E - D O K U M E N T I S J M Modern RISC-Societies Towards a New Paradigm for Societal Evolution Edited by Lučka Kajfež-Bogataj | Karl H. Müller Ivan Svetlik | Niko Toš Modern RISC-Societies Towards a New Paradigm for Societal Evolution Edited by Lučka Kajfež-Bogataj | Karl H. Müller Ivan Svetlik | Niko Toš E-DOKUMENTI SJM/Ljubljana, 2022 edition echoraum | Vienna Lučka Kajfež-Bogataj/Karl H. Muller/ Ivan Svetlik/Niko Toš (edited) Modern RISC-Societies Towards a New Paradigm for Societal Evolution First electronic edition Published by: Univerza v Ljubljani, Fakulteta za družbene vede, IDV, CJMMK Kardeljeva ploščad 5, Ljubljana Ljubljana, 2022 Book Series: E-Dokumenti SJM 10 Editor: Slavko Kurdija Layout: Werner Korn Text edition: Gertrud Hafner Book edition: Hannah Elmer Figures and Graphs: Michael Eigner This work is licensed under a Creative Commons AttributionShareAlike 4.0 International License. URL: https://knjigarna.fdv.si/ in www.cjm.si DOI: 10.51936/9789612950149 Kataložni zapis o publikaciji (CIP) pripravili v Narodni in univerzitetni knjižnici v Ljubljani COBISS.SI-ID 101935875 ISBN 978-961-295-014-9 (PDF) Printed edition was published by Edition Echoraum, Wien, 2010 ISBN 978-3-901941-23-8 Table of Content Foreword 9 Abstracts 11 1 The RISC-Program: An Experiment in Trans-Disciplinary Knowledge Production at the University of Ljubljana 25 Karl H. Müller | Ivan Svetlik | Niko Toš Part I — An Overview on RISC-Research 43 2 RISC-Processes and Societal Coevolution: Towards a Common Framework 63 Karl H. Müller Part II — RISC Modeling and RISC Theory 115 3 A Discussion on Zipf ’s Law 119 Heinz von Foerster et al. 4 Bubbles Everywhere in Human Affairs 137 Monika Gisler | Didier Sornette 5 New Models for Generating Power Law Distributions 155 Günter Haag 6 RISC-Processes and Their Weak Societal Protection Networks 169 Karl H. Müller 7 Zipf ’s Law in Labor Status Transitions: New Insights from Austrian Labor Market Data 185 Michael Schreiber 8 Self-Reflexive, Contagious, Attraction-Driven Networks (SCANs): Towards a New Transdisciplinary Framework for RISC-Modeling 203 Günter Haag | Karl H. Müller | Stuart A. Umpleby 9 The Organizing of Promises: Finance Capital as Tensegrity System 239 Adrian Lucas 10 The Poverty of Economic Explanations 247 Peter Štrukelj Part III — RISC-Applications 283 11 Weather- and Climate-Related Natural Hazards 287 Jože Rakovec 12 The RISC Potential of Converging Technologies 297 Toni Pustovrh 13 Risk, Crises and Control: Between Fear and Negligence 325 Marko Polič Part IV — RISC-Prevention and Damage Control 339 14 Natural and Other Disasters: A Social Work Perspective 343 Romana Zidar | Mojca Urek | Vili Lamovšek | Nino Rode | Jelka Škerjanc 15 Secondary Disaster and Social Work 369 Jelka Škerjanc 16 “Tsunami Project:” A Case of a Col aborative Project Between Two Universities 387 Mojca Urek | Bogdan Lešnik 17 Seismic Isolation for Asymmetric Building Structures 403 David Koren | Vojko Kilar Part V — Towards Inter- and Transdisciplinary Forms of Science 433 18 Socio-Economics and a New Scientific Paradigm 437 Rogers Hol ingsworth | Karl H. Müller | Ellen J. Hol ingsworth | David M. Gear 19 Turning Science Transdisciplinary: Is it Possible for the New Concept of Cross-Disciplinary Cooperations to Enter Slovenian Science and Policy? 461 Franc Mali 20 Approaches to Interdisciplinary Col aborative Research 477 Simona Tancig | Urban Kordeš Bibliography 503 Authors 551 Index 559 To Rogers J. Hol ingsworth and to Ellen Jane Hol ingsworth who enabled the formation of the RISC-Program and the publication of this book. To Yvonne Lucas (1922–2009) and to Hermann Müller (1925–2010) who would have liked to see this book come true. Foreword Working in an inter- and trans-disciplinary environment like in the RISC-program (Rare Incidents, Strong Consequences) at the University of Ljubljana requires a coordinated effort by a large number of persons across national boundaries and across different languages. In our case, this co-operation included several faculties at the University of Ljubljana, Ivan Svetlik as vice-rector and, subsequently, Lučka Kajfež-Bogataj from the University of Ljubljana as local coordinators and Karl H. Mül er form the Wiener Institute for Social Science Documentation and Methodology (WISDOM) as external adviser. Financial support for the RISC-program was provided by the faculties of the University of Ljubljana and two Austrian ministries, namely the Federal Ministry of Science and Research and the Federal Ministry of Labour, Social Affairs and Consumer Protection (BMASK). Thus, thanks go to – − Gertrud Hafner in Vienna who was confronted with the difficult tasks of transforming a very heterogeneous manuscript into a homogeneous format and into the new publishing program of the book series − Ivi Kecman, Anna Polajnar and Manca Poglajen in Ljubljana who served as a vital interface between Ljubljana and Vienna − Michael Eigner who was mainly responsible for the design and the redesign of the diagrams, figures and graphs in the book − Hannah Elmer who helped to transform the combined German/English or Slovenian/English manuscripts into homogeneous English texts − Andreja Kocijančič who as rector of the University of Ljubljana was an active supporter of the RISC-program from its earliest stages onwards − Andrea Schmölzer from the Federal Ministry of Science and Research who made it possible for WISDOM, to engage in cross-border research activities − Stefan Potmesil, Richard Fuchsbichler, Roland Hanak and Susanne Schlögl from the Federal Ministry of Labour, Social Affairs and Consumer Protection (BMASK) who continue to provide a stable support for cross-border cooperations with Slovenia − Werner Korn who acted as an unmoved prime mover behind this book-project and behind the entire series on “Complexity, Design, Society” which has reached already its fourteenth volume − a remarkably good spirit of stable cooperation and friendship between the editors which has overcome many obstacles and barriers and which will continue to last well-beyond the publication of this book. This book has been the outcome of trying something new and innovative in the field of inter- and transdisciplinary research on societal issues, past, present and 10 Foreword future. Like in most innovative attempts in this area, one should be reminded of three important features of innovation processes in general. First, Michael Vance stresses the peculiar fact that innovation processes should not be seen as radical departures, but in an intimately close relationship with their old environments. Innovation is the creation of the new or the re-arranging of the old in a new way. Consequently, the new RISC-framework takes large quantities of building blocks from traditional research across the natural or social sciences areas but provides a re-arrangement or, alternatively, a recombinative and integrative design for these many diverse and isolated components. Second, Sir Francis Bacon [1561–1626], in his “Essay on Innovations,” provides another important hint on innovation processes, namely the imperfect nature of the early stages of innovations which applies quite naturally to the emerging RISC-framework as well. As the births of living creatures, at first, are ill-shapen: so are all innovations, which are the births of time. With respect to the future path of RISC-research, Woody Allen gives a consolatory assessment even for the case of failed innovations. If you’re not failing every now and again, it’s a sign you’re not doing anything very innovative. In this sense, the general new outlooks and perspectives on the evolution of RISC-societies, even in the case of their unsuccessful diffusion, remain an honest attempt to address urgent societal problems in a genuinely new way. It should be emphasized that the present book in its final design fits very well into the overall context of the book series with its emphasis on complexity research or on new research designs, new methodologies or, as an essential element, new perspectives on the evolution of societies. We sincerely hope that the results of this book enable researchers in the field to widen their current perspectives on societal evolution significantly and to open up new ways for an inter- and trans-disciplinary research of contemporary societies with exciting and innovative results. Vienna, September 2010 Lučka Kajfež-Bogataj | Karl H. Müller | Ivan Svetlik | Niko Toš Abstracts The RISC-Program: An Experiment in Trans-Disciplinary Knowledge Production at the University of Ljubljana Karl H. Müller | Ivan Svetlik | Niko Toš Between 2007 and 2009, the University of Ljubljana initiated a trans-disciplinary research program on rare events with strong societal repercussions and effects, which has been labelled the RISC-program (Rare Incidents, Strong Consequences). In this process, the university established a small unit that sought to act as a catalyst for promoting trans-disciplinary research on rare events both inside and outside Ljubljana. Fol owing an international workshop in May 2007, the RISC-unit organized a series of talks, lectures, workshops and research activities, which highlighted the current knowledge frontiers on rare events as well as the available policy recommendations and best practices in reducing hazards and disasters related to rare events. From its overall goals, these RISC-activities were intended as a model for a new type of trans-disciplinary knowledge production that draws together expertise in the social, physical, biological and technical sciences to address urgent societal problems. The present article summarizes both the advances and the shortcomings of these RISC-activities. RISC-Processes and Societal Evolution: Towards a Common Framework Karl H. Müller This introductory article will present an overview of RISC-processes (Rare Incidents, Strong Consequences) and their changing relations and role in the evolution of societies, past and present. The article will make three central claims. First, RISC-processes can be considered as the missing link for an evolutionary theory of contemporary societies. Second, RISC-processes, in conjunction with additional building blocks within the wider evolutionary framework, become necessary and sufficient for a new and comprehensive theory of societal evolution. In this article, a broad outline of this new theoretical perspective on societal evolution will be provided. Third, the current stage of societal RISC-development makes it imperative to reconsider the problem of sustainability. In the light of the preceding RISC-discussion it will be argued that sustainability needs at least three main dimensions which are strictly independent from each other. The first one comprises the widely discussed sustainability issues with respect 12 Abstracts to globalization, namely the generalizability of today’s advanced development levels to the entire globe, the second one deals with the transferability of natural resources (environment, raw materials, water, air, etc) to future generations and the third main dimension of sustainability, however, must be related to RISC-processes and to the emergence of robust ensembles, resilient linkage structures and flexible support networks which, despite the impossibility to control RISC-processes locally or globally, are able to withstand most of the disastrous impacts of these rare events in the long run. A Discussion on Zipf ’s Law Heinz von Foerster et al. This article originated at the Macy-Conferences which took place between 1946 and 1953 and which, in retrospection, can be considered as the most important incubator for the subsequent developments of cybernetics, systems theory, artificial intelligence or the cognitive sciences. This particular presentation was made by Heinz von Foerster [1911–2002], an Austrian born scientist who emigrated to the United States in 1948 and who become widely known through his Biological Computer Laboratory [1958–1976] at the University of Il inois. Von Foerster’s discussion of Zipf ’s law took place during the 9th Macy Conference which was held in New York’s Beekman Hotel on March 20 and March 21, 1952. Since Zipf ’s law is just another word for power-law distributions which are in the center of the RISC-program, it is rather obvious why this hitherto unpublished manuscript has been included. The following group of persons was present at the meeting and participated briefly or extensively in the discussion on Zipf ’s Law: W. Ross Ashby (Psychiatry), Gregory Bateson (Anthropology), Julian Bigelow (Electrical Engineering), John Bowman (Sociology), Ralph W. Gerard (Neurophysiology), Heinrich Klüver (Psychology), Warren McCul och (Neuropsychiatry), Margaret Mead (Anthropology), Walter Pitts (Mathematics), Henry Quastler (Medicine and Computer Engineering), Gerhard von Bonin (Neurophysiology), Jerome Wiesner (Computer Engineering) and John Z. Young (Neuroanatomy). Bubbles Everywhere in Human Affairs Monika Gisler | Didier Sornette We review the “social bubble” hypothesis, which holds that strong social interactions between enthusiastic supporters of new ventures weave a network of re- Abstracts 13 inforcing feedbacks that lead to a widespread endorsement and extraordinary commitment by those involved in the projects, beyond what would be rationalized by a standard cost-benefit analysis in the presence of extraordinary uncertainties and risks. Starting with analyses of previous bubbles, in particular the famous “Tulip mania,” the social bubble hypothesis is illustrated by the example of the Apollo project. The social bubble hypothesis suggests novel mechanisms to catalyze long-term investments, innovations and risk-taking by the private sector, which otherwise would not be supported. New Models for Generating Power Law Distributions Günter Haag Power-law distribution, rank-size distribution, Zipf ’s law, hierarchy as a systemic organization into levels, self organized criticality and fractal phenomena are different aspects which may belong to the same coin. New models for generating power law distributions are discussed in order to demonstrate the typical aspects and issues of different modeling points of view. Moreover, some aspects of micro and macro based modeling approaches are discussed and shown. The interpretation of the models and the outputs of the different approaches are open for discussion and further research projects. RISC-Processes and Their Weak Societal Protection Networks Karl H. Müller This article provides an overview on the special relations between RISC-processes and their societal control potentials, be it at the national or at the global level. At the outset, the intricate relations between RISC-processes, controls or governance and forecasting will be discussed in greater detail and the classical equivalence of explanation control and prediction will be effectively abolished. The second major point places special emphasis on the inherent control mechanisms in self-organizing RISC-processes and their necessary failures in critical periods and stages. Finally, the third part of the article points to inherent vulnerabilities of globalized RISC-societies which lie clearly beyond any societal control. 14 Abstracts Zipf ’s Law in Labor Status Transitions: New Insights from Austrian Labor Market Data Michael Schreiber Motivated by discussions about competitive strategies of Europe that expect member states to implement flexicurity for employers and employees we present recent findings of research into new methods, tools and procedures for RISC-processes in labor markets. We studied the transitions in the employment status in Austria for a period of six months in 2009 by analyzing monthly data according to three distinctions among target groups: age, gender and education. It turned out that frequencies of changes in employment status followed a power law during these six months. Moreover, the complexity of the status change networks was shown to be reducible by cut-off values that enable schematic classifications of the different groupings. Self-Reflexive, Contagious, Attraction-Driven Networks (SCANs): Towards a New Transdisciplinary Framework for RISC-Modeling Günter Haag | Karl H. Müller | Stuart A. Umpleby This article extends the discussion of the modeling of RISC-processes to new clusters or families of network models which, so far, have not made their way to the core of socio-economic theory and model-constructions. These new network groups can be characterized as self-reflexive and contagiously attractive, i.e., as driven by internal learning and by external imitation processes where contagious attractions become an intrinsic property of the network relations themselves and not an exogenous factor that can influence or disturb network actors. Usually, these networks exhibit multi-level structures and are normally marked by high degrees of observer-dependencies. In the course of this article, a more general class of models will be introduced under the heading of self-reflexive, contagious, attraction-driven networks (SCANs) which can be used for a wide variety of complex self-organizing RISC-processes across nature or society. The Organizing of Promises: Finance Capital as Tensegrity System Adrian Lucas By pandering to promises of control ability, traditional analyses of financial crisis fail, and they fail to the extent that blame is attributed to subjective Abstracts 15 categories, irrespective whether the subjective scapegoats be the capitalists of Marxist analysis, or investment bankers self-rewarded for arbitraging regulatory frameworks, or over-enthusiastic credit borrowers, or whatever. This paper instead takes its cue from Le Corbusier’s desubjectifying, hence objective, reconfiguration of architecture, and applies Fuller’s tensegrity concept to configure a more objective, since desubjectivized, analysis of financial capital as an organization of promises, whose immanent topology is that of a self-dynamic tensegrity system. The Poverty of Economic Explanations Peter Štrukelj The aim of this article is to provide a thorough analysis of the current style of economic explanations for the severe global financial crisis from 2007 onwards. Until now, economists were believed to be capable of ex post explanations and, due to the complex nature of the economic system, unable to produce accurate forecasts, despite Karl R. Popper’s emphasis on the symmetry of explanation and forecasting schemes. This article tries to establish that economists, by and large, are incapable of generating reliable and robust ex post explanations. Phrased differently, although these accounts look prima facie explanatory and although their proponents believe that they have accomplished an economic explanation, these economic ex post explanation schemes should not even be considered as explanation sketches, let alone as explanations. Rather, these explanatory accounts are an expression of the nearly perfect blindness of an entire profession vis à vis the evolution of our global economic and financial system. Weather- and Climate-Related Natural Hazards Jože Rakovec Global statistics on weather-related natural disasters show that, of all natural hazards leading to disasters, 90% are linked with meteorological or hydro-meteorological extremes or conditions, and 75% of economic losses as well as 70% of lost lives are due to these events [Golnaraghi, 2008]. A system of forecasts and warnings became operational on the global scale last year, with one part including a system for Europe [Meteoalarm, 2008]. The global system was developed based on a pre-existing one for the Alpine region, which had resulted from a project within the 16 Abstracts scope of Interreg IIIb 2003–05. Some Slovenian early weather warnings [ARSO, 2008a] and earthquake warnings [ARSO, 2008b] contribute to it. An overview of the natural causes and development of these phenomena is given, accompanied by statistics for the whole world as well as for Slovenia. The concept of the early warning system is presented as it regards the global and regional levels. As an example, the case of the strong precipitation, flash floods and landslides on September 18, 2007 in Slovenia is discussed from the meteorological point of view, showing the main components that caused its severity: the long-lasting advection of warm, humid air ahead of a cold front and stationery convection. The components needed to help with the forecasting of such events and with real-time monitoring are briefly examined. The RISC Potential of Converging Technologies Toni Pustovrh Ever since their earliest inception, science and technology have played an increasingly important role as catalysts of cultural and social change, affecting and shaping human societies and the lives of individuals. In the past two decades, the rapid pace of scientific and technological development has opened many new fields and begun experimental work on numerous applications that may radically alter existing social relationships and structures, as well as challenge contemporary moral and ethical boundaries. The convergence of technologies, such as those arising from combinations and mutual stimulation among the rapidly growing domains of nanotechnology, biotechnology, information technology and cognitive science, is expected to yield insights and applications with the greatest transformative potential, while having a disruptive effect on existing technologies and on society as a whole [Roco and Bainbridge, 2003; Nordmann, 2004]. Of course it is also possible to claim that such a process in its essence does not represent anything new, since scientific and technological progress has always occurred by combining the findings, tools, methods and insights from a variety of different fields. But there are some aspects of converging technologies that could be seen as having deeply transformative features. Advances in converging nano-, bio-, info- and cognitive [NBIC] technologies potentially offer tools for the direct manipulation of the underlying biological mechanisms of the human mind and body, thus enabling the manipulation of the genome, the “blueprint” according to which each individual’s physiology develops. Manipulations of the brain, whether involving molecular interventions in the form of psychopharmacological agents or Abstracts 17 the implantation of cybernetic devices, could allow alterations of various cognitive functions of the human mind. Depending on whether we subscribe to a linear or an accelerating view of scientific and technological progress, we could claim that the number of innovations and their sophistication, power to manipulate and scope of influence is also increasing. The field of synthetic biology can serve as an illustrative example. Synthetic biology (drawing from nano-, bio- and information technology) is currently striving to redesign the constituent systems of naturally occurring microorganisms, so that these can be employed to perform other functions valuable to humans. The anticipated benefits and risks, though at this time necessarily still speculative, are great. Microorganisms that could be engineered to decompose currently non-biodegradable materials such as glass and plastics or break up dangerous chemicals would greatly contribute to recycling efforts as well as to the remediation of pol uted areas. Other uses could involve the production of useful materials and medicine, energy generation and the enhancement of human biological systems [Chopra and Kamma, 2006]. The risks could involve unanticipated side effects of interactions with the natural environment or the human body, the development of negative economic or social trends, or the risks emerging from intentional use of such technologies as weapons created for military or terrorist purposes [Boutin, 2006]. The release of engineered pathogens could thus present a catastrophic or even existential risk [Bostrom, 2002]. Another example is the potential ability to engineer offspring with specific physical or cognitive traits, popularly called “designer babies.” While truly sophisticated genetic engineering is not yet available, pre-implantation genetic diagnosis already offers the possibility of screening embryos for various genetically-based diseases and selecting those without such disorders. Sex selection was one of the first non-health-related traits offered to prospective parents, while recently The Fertility Institutes announced they would soon be offering selection of complexion, eye and hair color, as well as other customizations as they are made available by scientific progress [TFI, 2009]. These and similar technologies have been praised by some [Savulescu, 2001] as providing deeply transformative benefits, and criticized by others [Fukuyama, 2002] as harboring the potential to destroy human nature and society. The paper thus presents an overview of some fields and applications of converging NBIC technologies [Beckert et al. , 2008] that are expected to have the greatest transformative impacts on individuals and societies in the near future, and it explores some of their potential societal implications, encompassing both risks and benefits, as shown in the examples of synthetic biology and “designer babies.” A special emphasis will also be given to global catastrophic and existential risks [Bostrom et al. , 2008] potentially inherent in these technologies, as well as to the 18 Abstracts potential ability of such technologies to mitigate or control the aforementioned catastrophic risks. Risk, Crises and Control: Between Fear and Negligence Marko Polič Risk and crisis, although relatively rare for individuals, are common phenomena in human life. Ulrich Beck has even introduced the concept of a risk society. While the number of studies devoted to risk and crisis is constantly increasing, as is the understanding of these phenomena, this knowledge is still compartmentalized between different sciences, and the gap between the views of experts and those of the lay public is decreasing very slowly, if at al . Quite often, technocratic approaches prevail. Why are people sometimes afraid of matters that are not dangerous, while at the same time they ignore warnings about real dangers? Which risks are tolerable and which wil provoke human actions? And what wil these actions be? Awareness of risk and crisis, as phenomena or events that threaten important human values and cause pressure and uncertainty, is necessary for beginning any action. Although actions depend on risk perception, they depend even more on subjective control, culture and social factors like trust or stigma. The mutual dependencies of these factors are discussed, especially the psychological aspects of risk and crisis management—that is, those matters that influence the decision-making and behavior of individuals and groups and depend on their psychological nature. A multidisciplinary and integrated approach will serve as the necessary context for proposing a satisfactory societal response in different emergencies in the sense of Simon’s ‘bounded rationality’ model. Some psychological theories relevant for explaining people’s behavior during disasters, which are seen as examples of crisis, are presented. These range from decision-making, bounded rationality, to the changes in the organizations. Natural and Other Disasters: A Social Work Perspective Romana Zidar | Mojca Urek | Vili Lamovšek | Nino Rode | Jelka Škerjanc The majority of models for responding to natural and other disasters aim at harm reduction before, during and after the event through the four-phase model of readiness, risk reduction, response and recovery. Such interventions overlook the importance of community resilience and their ability to cope with such events. Since social work is able to function in unforeseen and unpredictable Abstracts 19 situations that demand innovative and original solutions [Flaker 2003], the role of the profession can contribute greatly to strengthening vulnerable individuals, groups and communities, when appropriately incorporated in the system of protection, rescue and relief. Presented research entitled The Analysis and Evaluation of Needs for Social Services in Cases of Natural and Other Disasters in the Municipality of Ljubljana, conducted from 2007 to 2009, focuses on the vulnerability of individuals and groups, the accessibility of institutions and services in times of disaster, the implementation of rights for those affected by disasters, the service coordination, the voluntary initiatives, the non-discriminatory practices, and the sensitivity to the needs of the affected population. In this research we used the concept of sense-making methodology [Dervin et al. , 2003]. This methodology enables the individuals who survived disaster, irrespective of their role in it, to reflect on their real experiences; this allows the research to focus on the event itself instead of on the formal social structures and/or societal roles in which the individual operates, as is typically done. The methodology was applied through thirty micro-moment time-line interviews, organized with the model of simultaneous sense-making. Results indicate that there are deficiencies in the existing formal, semi-formal and informal systems of protection, rescue and relief. Respondents identified as a gap the significant lack of community and social support. Through the research, the authors identified 10 categories of deficiencies: too great a response time, disconcerted performance of some organizations and inappropriate informing of people, lack of available staff or crucial person in some institutions, lack of clearly defined common protocols, lack of criteria for and opacity of relief distribution, imbalance of power and unequal distribution of resources and relief among the population, frequent discrimination and human rights violations, overlooking vulnerable (poor) groups of the population, inaccessibility of different forms of support in the field during and after the disaster, resorting to bureaucracy because of nonexistent protocols, lack of psychosocial support and relief for rescuers. Secondary Disaster and Social Work Jelka Škerjanc In July 2004, in the Upper Soča river valley, Slovenia, the third earthquake in the last 28 years affected the living conditions of the residents. Three main facts have previously defined the support after earthquakes in the Bovec area: the support for residents has not been sufficient; the support has been short-term and focused 20 Abstracts mainly on the material living conditions of residents, and there has been almost no support for people struggling with their every-day living conditions. Based on past experiences, the residents feared the reconstruction, claiming that for some of them, the reconstruction was itself another ordeal to be suffered. In a complex situation of extraordinary circumstances, the probabilities were high that the residents’ needs would be overlooked and their expertise in their lives ignored. A three-month-long, social work volunteer camp project was set up to support residents in organizing their lives. Volunteer social workers took the side of the residents. From this perspective, the residents’ situation became more visible, and some features that generate stress and concerns for the residents in reconstruction after a natural disaster were brought to light. In the article, we address the roles and characteristics of social work performance in organizing support for residents after a natural disaster. The support was organized according to the individual resident’s definition of his or her reality and the need for service provision. The tasks performed were recorded daily according to the methodology that allows further analysis about the following: the needs for citizens in organizing their lives after natural disaster; the roles of social workers in providing support; effects of stakeholders involved; the distribution of power between the citizen and structures involved in reconstruction. The statistical records of the services delivered by the project brought to light the experiences residents had with structures and institutions, political subjects, media, volunteers and charity organizations after being affected by a natural disaster. There emerged an accumulation of stress and trauma generators for residents who have little or no means of support for facing them, for reducing them or for actively coping with them. Every angle that we view the situation of the residents from shows us their loss of power in their lives and the little or no control over the solutions to improve it. At the moment when natural disaster hits citizen reality, along with the consequences that disaster creates, there emerge additional generators of trauma and distress. The findings also lead to the conclusion that in organizing support for residents after a natural disaster, social work has an important role to play. Its place is with the residents and at their side. From this standpoint, social work acts to support residents, making sure their voices are heard and that they maintain control over the support they receive. Thus, social work has a unique perspective on the residents’ situation and on their need for support. The information gathered from the resident’s perspective enables the design of original, genuine and creative answers to his or her situation. Since the answers respond directly to people’s needs, they are also efficient. Abstracts 21 “Tsunami Project:” A Case of a Collaborative Project Between Two Universities Mojca Urek | Bogdan Lešnik The paper is a report on the findings of a research camp held in a village on the southern coast of Sri Lanka eight months after the Indian Ocean tsunami. The camp was part of a wider project of col aboration between the University of Colombo and the University of Ljubljana, and its participants were students from both universities working together as a group. The report is mainly focused on the views and experiences of humanitarian aid as expressed by the people of this village. They keenly observed the distribution of aid and saw irregularities and abuses that only increased their distress. Among other issues, they questioned the methodology that caused less visible and socially excluded members of the community to be excluded once again from the distribution of aid, and they particularly resented being forced into submission. The project that started as a summer camp in this tsunami-affected village led to the signing of a Memorandum of Understanding between University of Colombo and University of Ljubljana for academic col aboration in the field of social work. This was followed by introducing social work as a stream within the special degree program in sociology at University of Colombo and by a fruitful exchange of knowledge, students and teachers between both universities. Seismic Isolation for Asymmetric Building Structures David Koren | Vojko Kilar The paper presents the summary of the main results of research work within the framework of the doctoral thesis of the first author performed in the past two years at the Faculty of Architecture, University of Ljubljana. The paper examines architectural-structural particularities of asymmetric buildings in earthquake-prone areas and the possibilities for the implementation of advanced technologies to increase earthquake resistance of such structures. In doing so, it examines new dimensions offered by the use of one such advanced technology ( i.e. , various devices of seismic isolation such as bearings, dampers, systems for displacement reduction) and their influence on architectural building design in earthquake-prone areas. Conventional design of structures, which is based on ensuring sufficient stiffness, strength and ductility, does not completely prevent the structure from damage. Contrarily, seismic isolation as a modern alternative in earthquake-resistant design offers a possibility of much higher damage protection, yet with much bigger 22 Abstracts financial input, which could be justified only for especially important buildings. In the first part of the paper, the architectural-urban reasons leading to the design of irregular buildings in architecture are determined, and explanations for its unfavorable seismic behavior from a structural point of view are given. Furthermore, the promising results of parametric study on the seismic behavior of structures with different levels of structural asymmetry isolated with lead rubber bearings are presented. The nonlinear dynamic analyses have been performed, and the results obtained have shown that the behavior of base-isolated structures is much affected by the distribution of isolators. It was observed that some distributions favored by common building codes are best only for accommodating the torsional effects in the base isolation system. A significantly different conclusion was found observing the nonlinear behavior of the superstructure, where such distributions might cause more damage in the flexible side of the structure. In the second part of the paper, a simplified nonlinear method is applied for analysis of base-isolated structures. For this purpose, a new bilinear idealization of the capacity curve for a base-isolated structure is proposed. In this way, the new method is capable of detecting the first damage (yielding point) of the superstructure as well as of estimating the behavior of the superstructure further in the nonlinear range. The results are presented in terms of top and base displacements as well as damage patterns of the superstructure. Comparisons of the results of the simplified method with the ‘exact’ results of nonlinear dynamic analyses have shown a very good agreement. It has been shown that the presented simplified approach might be a valuable tool for design, analysis and verification of the behavior of seismically isolated structures. It has been shown in the paper that the correct use of seismic isolation can contribute to freer design of architecture, which is in the interest of both structure and architecture designers. Moreover, the topic discussed is rather interdisciplinary and tries to improve the level of cooperation between architects and other experts. In doing so, it would also make it possible to shift certain interesting architectural concepts from earthquake-safe areas to earthquake-prone ones. In this way, more advanced, more daring and at the same time sufficiently earthquake-resistant architectural designs would be possible to build. Abstracts 23 Socio-Economics and a New Scientific Paradigm Rogers J. Hollingsworth | Karl H. Müller | Ellen J. Hollingsworth | David M. Gear This paper argues that a new scientific paradigm [Science II] is slowly emerging and is rivaling the Descartes-Newtonian paradigm [Science I], which has been dominant during the past several hundred years. The Science II paradigm places a great deal of emphasis on evolution, dynamism, randomness, chance, and/or pattern identification. As a cause and effect of the new paradigm, scholars in the physical, biological and social sciences are increasingly addressing common problems. Several of these are discussed. In their research, these scholars are using common models, methods, and metaphors. The paper focuses much of its attention on the field of socio-economics as an example of how the newly emerging paradigm [Science II] offers considerable potential for a hybrid field of social science to become more engaged with colleagues in the natural sciences. The convergence of interests across scientific fields has enormous implications for the appropriateness of reshaping the structure of existing universities that encourages high fragmentation, specialization and differentiation, but poor communication across academic disciplines. Turning Science Transdisciplinary: Is it Possible for the New Concept of Cross-Disciplinary Cooperations to Enter Slovenian Science and Policy? Franc Mali In the paper, some theoretical and empirical aspects of scientific trans-disciplinarity are presented. The development of recent post-academic science is characterized by a strong orientation to trans-disciplinarity in science. For example, the whole ‘philosophy’ underlying the European Research and Innovation Area places a strong emphasis on cross-, inter- and trans-disciplinarity in science. Discussed in more detail in the contribution, the concept of converging technologies represents a new phase in the development of trans-disciplinarity in science. In the paper, the main attention is given to the explanation of some barriers that hinder the realization of the new trans-disciplinarity in science and policy discourse in Slovenia. Here, the new role of the centers of excellence as new intermediary science organizations is highlighted. Namely, the centers of excellence could play an important role in shifting science and policy discourse from disciplinarity to trans-disciplinarity. Approaches to Interdisciplinary Collaborative Research Simona Tancig | Urban Kordeš The aim of this article is to identify some of the main issues related to inter- (trans-, cross-, multi) disciplinary work on complex problems. The focus is on concepts, theoretical frameworks, practical considerations and challenges, and possibilities of col aborative learning and problem solving. The RISC-Program: An Experiment in Trans-Disciplinary Knowledge Production at the University of Ljubljana 1 Karl H. Müller | Ivan Svetlik | Niko Toš 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 16 13 15 The last fifty years of the 20th century as well as the first decade of the new millennium have experienced a rapid and profound re-configuration of the science system as a whole. After a brief summary of these secular changes and phase transitions the article will present an overview of the organization and the outputs of a small research program at the University of Ljubljana which, from 2007 to 2009, ran under the name of RISC (Rare Incidents, Strong Consequences). The rationale for the focus on rare events was that, on the one hand, these rare events constitute a very important problem cluster for contemporary societies and that, on the other hand, rare events, due to their widespread occurrences in societal and natural systems, can serve as a paradigmatic example for a new type of trans-disciplinary knowledge production. Moreover, rare events belong to the class of complex scientific problems which share a number of characteristic features.1 For instance, rare events with big societal impacts are in most instances difficult or impossible to predict in terms of the time of their appearance as well as of their magnitude. They appear suddenly as a consequence of much longer term development processes. Additionally, these rare events question the conventional wisdom on the close linkages between explanation, prediction and control.2 Although nearly impossible to predict and very difficult to explain even ex post,3 these rare events can be controlled, nevertheless, in ways beyond traditional scientific explanations or predictions. 1.1 Secular Phase Transitions in Science Before entering into a detailed description of rare events a general background topic is to be discussed which became an important reference point in the organization of the RISC-program, namely a significant change in the overall science landscape. At the outset two major phase transition in the overall scientific landscapes will be laid out, albeit in a very brief manner. Currently, the cognitive and organizational landscapes of science are experiencing a secular shift which can be characterized in its epistemological, ontological, theoretical and organizational 1 On complex problems and their characteristic features, see, e.g. , Anderson, Arrow and Pines’, 1988, Stein, 1989, Casti, 1994, Coveney and Highfield, 1995, Cowan, Pines and Meltzer, 1999, Hol and, 1995, Kauffman, 1995 or Rescher, 1998. 2 On the conventional wisdom of the strong ties between prediction, control and explanation, see, for example, Casti, 1989, Popper, 1965 and 1972 or Suppe, 1977. 3 See, for example, the very revealing article on ex post explanations of financial crises by Peter Strukelj in this volume. 28 Karl H. Müller | Ivan Svetlik | Niko Toš dimensions as a transition from Science I to Science II4 and in its knowledge organization as a change from Mode I to Mode II.5 Science I corresponds to the organization of science from its initial modern phase in the 16th century to the 1940s and 1950s approximately. Science I is the long-term period of majestic clockworks, culminating at an early stage with the “Principia Mathematica” of Sir Isaac Newton in 1687. In contrast, Science II operates with blind watchmakers [Richard Dawkins] or, to use another metaphor from Karl R. Popper, works in a configuration of clouds.6 The transition from Science I to Science II should not be seen as replacements or substitutions of old homogeneous forms with new ones, but as a transition in terms of hegemony. The first phase from the long 16th century up to the period between 1900 and 1950 is characterized by the dominance of Science I and contained only a few elements of Science II, whereas the second phase from the 1950s onwards can be described as the peaceful co-existence between a dominant area of Science II and a substantial cluster, which remains structured and organized in Science I. Table 1.1 highlights some of the major shifts from Science I to Science II, which, once again, should be understood as an exchange between center and periphery. Following Table 1.1 more specifically, the following distinctions can be introduced between Science I and Science II.7 With respect to theory construction, general and universal laws, while at the core of Science I, move to the periphery of Science II, while pattern formation or pattern recognition, formerly marginally embedded in the domains of Science I, move to center stage of Science II. The leading discipline for Science I is theoretical physics whereas the core areas of Science II are the life sciences, broadly conceived. Science II addresses a large number of common problems, common metaphors, common methods as well as common models and mechanisms which can be identified across a wide variety of disciplines in the natural and in the social sciences. This common domain comprises issues like evolutionary dynamics, structural changes, multi-level organization or morphological issues. George Cowan identified a large set of problems which, contrary to the age of Science I, require the co-operative efforts of scientists across the Great Divides 4 See especially Hol ingsworth and Müller, 2008, Hol ingsworth, Müller and Hol ingsworth, 2008 or Hol ingsworth, Müller, Hol ingsworth and Gear, 2008 and in this volume. 5 On the transition from Mode I to Mode II, see especially Gibbons et al. 1994 or Nowotny, Scott and Gibbons, 2001, and Nowotny, 2005. 6 One could make the point that Science I and Science II correspond, at least partially, to the two epistemic cultures, identified by Knorr-Cetina, 1999, namely the closed culture of high-energy particle physics [Science I] and the open trial and error type of micro-biology [Science II]. 7 For more background materials, see also Price, 1963 or Rescher, 1982. The RISC-Program 29 of natural, technical and medical sciences on the one hand and of the social sciences and humanities on the other hand. TAble 1.1 Basic Differences between Science I and Science II Dimensions Science I Science II leading Fields of Science Classical Physics evolutionary biology and the Sciences of Complexity Theoretical Goals General, Universal laws Pattern Formation and Pattern Recognition Theory Structures Axiomatic, Reductionistic Nested, in Multiple levels Simultaneously Research-Programs Closed Deductive Open, Recombinative Heuristic Rules Induction Critical Phenomena, Qualitative Change Units of Method Causal Relations Generative Relations Forecasting Capacities High low levels of Complexity low High Ontology Dualism Monism epistemology Observer- Observer-Inclusion exclusion(“Objectivity”) Re-entries (Self-Reference) Re-entries excluded Re-entries Possible Dynamics linear Dynamics, Non-linear, far from equilibrium equilibrium Types of Distribution Normal Distributions, Mild Power-law Distributions, Distributions Wild Distributions Potential for Interdisciplinary low High Research Cognitive Distances High low between Natural and Social Sciences leading Metaphors Clocks Clouds Theoretical neurophysics; the modeling of evolution, including the evolution of behavior; strategies to troublesome states of minds and associated higher brain functions; nonlinear systems dynamics, pattern recognition and human thought; fundamental physics, astronomy, and mathematics; archaeology, archaeometry, and forces leading to extinction of flourishing cultures; an integrated approach to information science; (or) the heterogeneity of genetic inventories of individuals [Cowan, 1988:236]. 30 Karl H. Müller | Ivan Svetlik | Niko Toš It is interesting to note that the Austrian cybernetician and pioneer in the cognitive sciences, Heinz von Foerster, addressed the issue of the declining hegemony of Science I already in the 1970s and 1980 which, in his assessment, was about to crumble and fade away.8 For Heinz von Foerster, the classical scientific method was characterized by three general rules which can be found in Table 1.1, too. First, Science I adhered to the rule of objectivity— “the properties of the observer shall not enter the description of his observations” [Foerster, 2003:285]. This rule effectively eliminated selfish designs from the landscapes of Science I. The second rule was described as the rule of induction or of conservation of rules— “rules observed in the past shall apply to the future” [Foerster, 2003:203] and led to the general inability and insensitivity for structural or qualitative changes. Finally, the third general rule was classified as the rule of causality or, alternatively, of necessary and sufficient causes and was characterized by Heinz von Foerster with a sufficiently necessary amount of romantic irony: “Almost everything in the universe shall be irrelevant” [ Ibid.]. Resting on these three pillars, Heinz von Foerster concluded that the traditional scientific method is “counter-productive in contemplating any evolutionary process, be it the growing up of an individual, or a society in transition” [ Ibid:204]. Aside from the cognitive phase transition in the theoretical science landscapes, profound changes are occurring simultaneously in the domain of knowledge organization. Here, a phase transition is under way which leads from the traditional modes of knowledge organization along disciplinary boundaries, labeled as Mode I, to a new and trans-disciplinary form of organization under the name of Mode II.9 It must be stressed that the phases of Mode I and Mode II do not coincide with the two periods for Science I and Science II since Mode II emerges only in the course of the 1980s and 1990s and Mode I does not start prior to the massive “disciplinification” of the sciences during the second half of the 19th century. Nevertheless, both phase transitions should be viewed as complementary,10 the first one in the cognitive domains of theory structures and leading epistemologies, the second one in the arena of scientific organizations and forms of intra-scientific or science-society co-operations. Table 1.2 highlights the major differences for both forms of knowledge organization. According to Table 1.2, a substantial shift in knowledge organization is breaking its way. One of the changes, which is common to both phase transitions is the significant increase in inter- and transdisciplinary co-operations [Mode II] and 8 For more details, see Müller, 2007 and 2008 or Müller and Müller, 2007. 9 On these secular shifts, see Gibbons et al. , 1994, Nowotny et al. , 2000, Nowotny, 1999 and 2005. 10 On the important notion of complementarity, see especially French and Kennedy, 1985, Rozental, 1967 or Söderqvist, 2003. The RISC-Program 31 a significantly increased number of theory—as well as mechanism-, methods-and model-linkages across scientific fields and disciplines [Science II]. TAble 1.2 Two Types of Organizing Knowledge Production Dimensions Mode I Mode II Knowledge Organization Separation between context basic discoveries in the of discovery and application context of applications Separation of basic- science entanglement between and applied science applied and basic science Focus on disciplinary Focus on trans-disciplinary matrices problems, patterns, metaphors Organizational Units Homogeneous research- Heterogeneous Research- teams in stable spatial teams; Distributed across locations space long-term orientation limited periods, temporary configurations Quality Control Peer review, Internal Wider set of criteria, including extra-scientific elements like societal acceptance, etc. Extra-Scientific-Domains Irrelevant Societal responsibilities and the need for societal consensus formation The capacity to cooperate with experts from other fields, to come to see the world and its problems in a complementary way and to emphasize with different presuppositions, involves the capacity to assume multiple cognitive and societal identities ... Biologists working in environmental science, computer scientists in the analysis of gene sequences and mathematicians in ecological modeling can equally gain reputation on both their native and new grounds [Gibbons et al. 1994:149]. As can be seen in Table 1.1, one of the main distinctions between Science I and Science II lies in the two dimensions of potential for inter-disciplinary cooperation on the one hand and the distances between the natural and the social sciences on the other hand. Both dimensions change substantially between Science I and Science II. In short, Science II becomes a period with a high potential of inter-disciplinary co-operation and a phase with relatively small distances between the natural and the social sciences, due to a growing stock of common models, methods, mechanisms and metaphors which can be used across various disciplinary fields and due to the change in leading fields from theoretical physics to the life sciences. 1.2 The RISC-Program as an Experiment within Science II and Mode II The RISC-program at the University of Ljubljana was characterized by a common focus on processes which exhibit a characteristic distribution between a small number of very big events with major effects and repercussions and a very large number of events with very small or marginal effects only. Therefore, the acronym RISC—Rare Incidents, Strong Consequences—has been chosen which, additionally, has the advantage of sounding strikingly similar to the concept of risk. Furthermore, well-established compounds like risk-society, risk-assessments, risk-insurances, etc. can be complemented with the corresponding RISC-notions. RISC-processes can be found in many fields of research and clearly transcend the boundaries of academic disciplines or the big divide between the natural and the social sciences. But it was not only the ubiquity of RISC-processes in geology, finance, migration research or linguistic which was so fascinating to study. The long-term aims of the research program on RISC-processes were − to find common models, methods and mechanism across the various RISC-applications and, in doing this, establish a strong trans-disciplinary line of co-operations at the University of Ljubljana − to search for a common framework of societal evolution which would be able to integrate the various disciplinary and trans-disciplinary RISC-work − to provide the infrastructure for bringing together a number of international experts having experiences in dealing with RISC-processes and enable a reflection and exchange of experiences − to become an internationally recognized focus for policy making activities in the RISC-arena by identifying major policy issues to be dealt with, possible future research on RISC-related hazards and disasters and new policy approaches. As can be seen from the long-term goals and perspectives, the research program wanted to give, on the one hand, new impulses for trans-disciplinary studies in the social, physical, biological and technical sciences at the University of Ljubljana and, on the other hand, to work on a common theory framework for the evolution of modern societies, past, present and future.11 11 For more background materials on the comparative advantages of this type of transdisciplinary organization, see especially Hol ingsworth and Hol ingsworth, 2000 or Hol ingsworth, 2004. 1.3 The Profile of the RISC-Program Following an international workshop at the University of Ljubljana on May 25 and 26, 2007 with an overview of RISC-processes the research program for the period from 2007 to 2009 was concentrated on institutionalizing a RISC-based agenda that was capable of advancing the state of global RISC-research. The work areas of a new co-ordinating unit or hub have been laid out already in a programmatic paper, underlying the Ljubljana Workshop12 and are to be recapitulated briefly. All the activities of the new hub fell into three broad cognitive domains [see Figure 1.1]. FIGURe 1.1 The Overall RISC-Research Program ������������������ ������������� ��������� ��������� ������� ������������� ����������� ����������� ����� ���������� ��������� ���������� ��������� ������� ������������� ����������� ����� ��������� �������� ���� ��������� ������������������ With respect to the theoretical core, it was already noted in the Mission Statement13 of the new research program that models, methods and mechanisms which generate or account for RISC-processes and a general evolutionary 12 See Hol ingsworth/Müller/Svetlik and Toš, 2007. 13 See Hol ingsworth/Müller/Svetlik/Toš, 2007, especially page 5. 34 Karl H. Müller | Ivan Svetlik | Niko Toš framework which is able to integrate the various RISC-processes, lie at the core of the RISC-program.14 Different applications across different academic fields and disciplines constituted the second group of cognitive tasks. Since RISC-processes can be found in the natural as well as in the social sciences they can be studied in diverse academic fields like economics, sociology, historiography, information sciences, medical research, earth sciences, biology or the environmental sciences. Finally, the third area in Figure 1.1 covered the overall implications of RISC-processes for the sustainability of contemporary societies. Within the research program policies and plans were developed to cope with large scale disasters or negative effects before they occur, in an effort to prepare societies (ex ante policies). Moreover, attention was given to efforts to sustain a high level of quality of life even in the face of disasters and, consequently, plans and policy suggestions to safeguard societies from disasters (ex post policies). With respect to the organization of the hub it was envisioned that the overall number of personnel should be very small and limited to the following group of persons. − First, a high-ranking hub coordinator from the University of Ljubljana was chosen who was well embedded in the communication processes between the different faculties of the University of Ljubljana. This person was responsible to guarantee the annual budget through contributions from the faculties. Professor Ivan Svetlik as vice rector of the University of Ljubljana stepped in to take over this central position. After his move to become Minister of Labour, Family and Social Affairs on November 21, 2008, Professor Lučka Kajfež Bogataj became his successor. − Second, an external scientific advisor was needed as scientific hub coordinator, who should enable contacts to international networks dealing with RISC-issues. Working part-time, the advisor was expected to focus his activities on the development of the RISC-program. Here, Karl H. Müller from WISDOM in Vienna was invited to take this coordinating position. − A group of experts at the University of Ljubljana have been selected who acted as a Faculty Advisory Board and who had the important task of disseminating the RISC-program to their faculties. − Another group of local researchers were to be mobilized who should become the key actors for RISC-research. Ideally, these persons were assumed to be 14 For relevant literature of the RISC-core see, for example, Bak, 1996, Bal , 2004, Barabási, 2002, Beinhocker, 2006, Haken, 1982 and 1983, Jackson, 2006, Jantsch, 1972, Jensen, 1998, Mandelbrot, 2004, Newman, Barabási and Watts, 2006, Sornette, 2002 and 2006, Thom, 1989, Watts, 2003. For the general background, see especially Zipf, 1949. The RISC-Program 35 teachers and researchers from the University of Ljubljana, who already dealt with various aspects of RISC-issues or who might have interest to do so in the future. Their immediate interest should be focused on bringing their students into the emerging RISC-activities and on the development of new research and teaching programs. They were considered as essential for the mobilisation of internal financial resources at the University of Ljubljana. − An administrative and technical support was provided for the two coordinators. This included a part time employee, office, and infrastructure for meetings, workshops and conferences on RISC-processes, other services etc. From October 2007 onwards, Anja Polajnar helped to coordinate all the planned activities, to prepare materials, to manage the web site, to makes travelling and accommodation arrangements, to handle reimbursements etc. From October 2008 until summer 2009, Manca Poglajen fulfilled this role. − An International Scientific Advisory Committee was set up, composed of several highly regarded foreign and Slovenian experts. The committee should monitor and advice the RISC-program. It assisted in implementing the vision for the new research program, energized the local participants and advised about sources of funding. The members were encouraged also to contribute to the program through lectures or individual advice. The international members from abroad were Professor Didier Sornette from the ETH Zürich, Professor Robert Boyer (Ecole des Hautes Etudes en Sciences Sociales and Centre Pour la Recherche Economique et ses Applications, Paris), Professor Henry Abarbanel (Professor of Physics and Research Physicist (Scripps Institution of Oceanography), University of California San Diego, Director of the Institute for Nonlinear Science at the University of California) and Professor J. Rogers Hol ingsworth, from the University of Wisconsin, Madison as speaker of the board. From the University of Ljubljana, Professor Lučka Kajfež Bogataj and Professor Anuska Ferligoj participated in the Scientific Advisory Board. − The RISC-program was placed in the framework of integrated activities of the University of Ljubljana. Organizationally it was situated in IRI (Institute for Innovation and Development). This concludes the presentation of the organizational format of the RISC-program. 1.4 Main Groups of RISC-Activities As already indicated, the RISC-unit was operating in a large number of areas like organizing talks, lectures, small workshops, or preparing a summer school, applying for research projects or counselling political bodies both nationally and internationally on appropriate policies for RISC-processes. As an overview the RISC-program wanted to operate, aside from the coordination activities, in four broad areas which can also be seen in Figure 1.2. FIGURe 1.2 The Organization of the RISC-Program 2007–2009 ����� �������������� ���������� ������������� ������������ ������������� ��������������� ������������� In terms of outputs of the RISC-program, one expected to obtain substantial results in five groups of RISC-activities during the period from 2007 to 2009 which can also be found in Figure 1.3. RISC-Research and Publications One of the core tasks of the RISC-program was the building of research communities at the University of Ljubljana across the social, physical, biological and technical sciences. These RISC-research communities were to conduct RISC-research and enhance the local expertise in this field. Leaders of faculties, research groups or centres and existing post graduate study programs were invited to take part in the initiative. The expected effect was to increase the The RISC-Program 37 horizontal and trans- disciplinary integration of the University of Ljubljana. Applying for grants on RISC-research at the national and the international level as well as conducting research projects was to become the most prominent type of activity.15 Moreover, it was expected that a well structured research profile should emerge as a result of the two years activities. FIGURe 1.3 The Main Tasks of the RISC-Program ����������� ����� ������������ �� ������������� ������������ � ��������������� � ��������������� � ���������������� ����� ����� ������������ �������� �������� ��������������� �� ������������� � ������������������� ����������������� ����� � ��������������������� �������� �� ����������������� � ��������� � ����������� � ������������� �� ��������������� � ��������������� � ������������� ������������� �� ������������� ��������������� � ���� � �������� �������������� ������������ � ������������� ������������� Thus, by the end of the first period there should be a small number of research reports, resulting from national or internationally organized RISC-projects. Additionally, the RISC-program coordinators were to produce a number of articles and towards the end of 2009 also a book on RISC-processes. Aside from that, the RISC-program invited members of the University of Ljubljana 15 A total of seven research applications have been made to national and international funding bodies (ARSS, ESF, Norwegian Foundation) in the course of the two years. However, the outcome was only marginal and no important research project was approved. 38 Karl H. Müller | Ivan Svetlik | Niko Toš to publish relevant articles or overviews under the new RISC-umbrella. For this purpose, a new series, entitled RISC-Research Series, was founded in January 2008.16 Risk-Teaching The RISC-coordinators organized a series of lectures, seminars and courses by national and international experts as well as small workshops on different aspects of RISC-research on a regular basis. This main purpose of these activities was to explore the possibilities for further research and teaching activities and to mobilize national and international RISC-research experts. By far the most important activity during the two year period was a regular series of talks and workshops that were organized in intervals of approximately four weeks. Speakers in this series included Henry Abarbanel, Patrick Doreian, Günter Haag, J. Rogers Hol ingsworth, Lučka Kajfež Bogataj, Adrian Lucas and Didier Sornette. These mainly international experts on RISC-processes were expected to interact with different faculty members or students for a short period of three to four days. The intended format was a lecture to the interested students and staff, one or more informal discussion-rounds and workshops in different institutes as well as informal meetings with students and faculty members. A second type of event consisted of small workshops for one or two days. For these workshops, several additional RISC-experts outside the University of Ljubljana were invited as well. The main emphasis of these workshops was concentrated on one of the following topics: − on modeling and model-related problems of RISC-processes − on a specific RISC-application domain like financial crises, environmental hazards, etc. − on RISC-relevant research on hazards, disasters and policy-implications. Additionally, the hub was working on establishing a summer school on RISC-research. It was envisioned to create this summer school in a two stage process, the first stage as part of an already existing summer school and the second and final stage as an independent annual summer school on different aspects of RISC-research.17 Moreover, the RISC-program attempted to mobilize local resources at the University of Ljubljana to support RISC-research by theses and dissertations 16 Currently, nine research reports have been produced in the RISC-series. 17 As a role model one can refer to the Santa Fe Summer Schools in the late eighties and nineties. See, for example, Stein 1989. The RISC-Program 39 across various disciplines and faculties. Graduate and post-graduate students were invited to take part in a RISC-award and three students with relevant RISC-topics have been selected. Finally, the RISC-unit produced two general outlines towards a graduate and a postgraduate RISC-program as joint degree study programs in cooperation between the University of Ljubljana and other institutions. The postgraduate RISC-program was conceptualized for persons already engaged in research professionally who want to add a special qualification and who want to become RISC-specialists. Both programs were designed to provide special competencies in: − models and generative mechanisms for RISC-processes across a wide variety of areas − applying models, methods and data for RISC-analyses − hazards and disaster prevention − conducting transdisciplinary RISC-projects. Finally, one was also contemplating the idea of involving third countries’ universities in order to apply to the Erasmus Mundus scheme. Risk-Networking and Conferencing One of the important targets of the RISC-program was the organization of knowledge-transfers on RISC-processes to other universities in Europe on the one hand and to the media on the other hand. With respect to the networking within the university domain, the major target was the diffusion of the RISC-program as a role model for trans-disciplinary research within an university and for a new type of social science or, alternatively, for a new form of addressing serious societal problems. For this purpose, several meetings have been arranged with the European University Association and the design of the RISC-program was presented in several international conferences like the World University Forum. In the area of media mobilization, the aim was to use national and international media as a carrier to inform the wider public on RISC-issues and on strategies and policies to minimize the impact of rare events. Additionally, counselling local, national or international journalists on appropriate policy measures for taming RISC-related consequences became an integral task of the RISC-media activities. RISC-Research-Infrastructures Following Figure 1.3, research infrastructures were the fifth and final group of RISC-relevant outputs. Several of the RISC-research applications were focused 40 Karl H. Müller | Ivan Svetlik | Niko Toš on the issue of data and data management for RISC-research. Unfortunately, none of these applications turned out to be successful. As a consequence, the RISC-related research infrastructures were only marginally developed in the course of the two years. In terms of infrastructure, in January 2008 a web-page on the RISC-program became available which included also the TRASD-homepage, set up for the kick-off meeting in May 2007. The RISC-homepage provided an in-depth documentation of on-going activities and served at least two main important tasks, aside from being a vital communication tool both inside and outside the University of Ljubljana. − First, all the lectures and seminars were recorded and could be accessed through the RISC-homepage. − Second, the homepage offered the research papers of the RISC-series for downloads and provided essential information on relevant literature on RISC-processes. At this point, this short summary of five different types of activities under the RISC-umbrella has come to its end. 1.5 Shortcomings As it turned out, the RISC-program did not succeed to pass a critical threshold and, to use a terminology from Walt W. Rostow, to launch a “take off ” which would have moved it into a stage of “self-sustained growth.” A final project application as a nationally funded center of excellence did not succeed. In our view, three main reasons, the first one cognitive in nature, the second and the third one organizational, can be identified for the failure of the program. Probably the most important reason for the failure of the program was the inability to mobilize RISC-related research capacities at the University of Ljubljana. On various occasions a general interest was expressed with respect to risk-related issues, but no or marginal interests could be recorded for RISC-issues. One can offer several empirical instances for the claim that no manifest RISC-related research interests emerged in the course of the last two years. − All in al , there was a total of seven international lectures by specialists in RISC-modeling, most of them physicists, but the overall impact of these lectures in terms of contacts, networking or projects was close to zero. − The 1st University Workshop showed an interesting risk-related program with presentations on risk-perceptions, floods, droughts, etc. Nevertheless, with the exception of a single presentation from the Jožef Stefan-Institute, not a single RISC-related lecture was presented during the workshop. The RISC-Program 41 − The RISC-award produced a number of interesting risk-related submissions. However, the only RISC-centered submission on co-citation networks was given a low evaluation by the members of the Faculty Board whereas risk-related topics were very favorably assessed. − The current RISC-volume has very valuable contributions from the University of Ljubljana but all these contributions are situated either in the area of risk-prevention, damage-control or in the area of the methodology of inter- and transdisciplinary research. − Additionally, not a single long-term research program at the University of Ljubljana has added RISC-related topics into its menu of research goals. In essence, the long-term research programs remained essentially within their discipline-specific boundaries and RISC-issues were not part of any of these programs. − Finally, the RISC-series was mainly produced by external persons outside the University of Ljubljana. Out of a total of nine papers so far one finds only two Ljubljana papers in the series, one by Lučka Kajfež Bogataj on climate change and one by Peter Strukelj on the weaknesses of economic explanations of the current crisis. Both contributions are very interesting on risk-related or methodological grounds, but do not belong to the core field of RISC-modeling or empirical RISC-research. In a crucial sense, the RISC-program maintained itself on the semantic ambiguity between RISC as a study of rare events and risk as an inter-disciplinary research field associated with disasters, hazards or accidents. Due to the large research capacities of the University of Ljubljana in the fields of risk-research, the RISC-program was apparently conceived as a subset or as a pure variation of a risk-related program. A second decisive element of weakness came about when the RISC-program was initialized. In the beginning, the first programmatic RISC-papers were focused on establishing a RISC-research institute at the University of Ljubljana. Due to funding difficulties, Ivan Svetlik made a very interesting move and the RISC-program became an internally generated activity where several faculties of the University of Ljubljana were to provide a seed money for a period of two years. What remained relatively vague, at this point, were medium and long-term targets and success criteria. With Ivan Svetlik and his strong position as vice-rector these pending issues could be kept implicit or vaguely defined. There was a general understanding that the RISC-program should have, on the one hand, international lectures, workshops or publications, and should generate, on the other hand, new inter-disciplinary research projects for the University of Ljubljana. Aside from this general understanding, no explicit targets have been 42 Karl H. Müller | Ivan Svetlik | Niko Toš fixed and, especially important, no procedures have been discussed for the case that these general goals cannot be reached through the available instruments of lectures, workshops, publications. The third component for the failure of the RISC-program emerged after the transition from Ivan Svetlik to the Slovenian Ministry of Labour, Family, and Social Affairs. At this point, the targets of the RISC-program as well as the function and the task distribution of all relevant actors should have been made explicit to and for all relevant participants. This should have included the role of and the financial incentives for the new RISC-coordinator of the University of Ljubljana, the role and the requirements of the Faculty Board and, finally, the role of the external scientific advisor. As a consequence, the weak governance structure became even weaker and the overall goals of the program became even less clear and diverse. 1.6 Future Outlooks The present volume offers a first overview not only on the potential of RISC-research, but also on the potential of RISC-research as a new form for the study of contemporary societies. The subsequent article on the dimensions and the long-term evolution of global RISC-societies can serve as an evaluation instance whether the scope of the RISC-program was intrinsically too small and theoretically too restricted or whether the analysis of RISC-societies could and should become a new recombinant societal science to address important societal issues in a broader, more inter- and trans-disciplinary and, equally important, more interventionist manner. Part I — An Overview on RISC-Research 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 16 13 15 Introduction to Part I: RISC 101 or a Primer on the New RISC-Framework for Societal Evolution The introduction to Part I will deviate from the usual form of introductions in this volume which will be focused almost exclusively on the content of the articles included in the various parts. Within the present introduction however, a short summary will be given of the new framework for the study of societal evolution which runs under the name of RISC-framework. The initial article has already indicated that the RISC-program was a two year effort at the University of Ljubljana to increase the inter- and trans-disciplinary communication across the widely distributed faculties across Ljubljana. The common topic consisted in the study of rare events across a wide variety of fields and academic disciplines and RISC (rare incidents, strong consequences) was the intentionally ambiguous name for this endeavor. On the basis of the various workshops, lectures and publications during these two years, a new theoretical framework gradually emerged, which should shed fresh light on processes of societal evolution in the long run. Subsequently, a summary of important building blocks of the RISC-framework will be laid out which constitute this new research program or, more adequately, this novel research tradition. The RISC-Framework as a Comprehensive Research Tradition Before going into the different RISC-building blocks in more detail it is useful to describe the requirements or ingredients of empirical research programs on the one hand and of larger research traditions on the other hand. Figure I.1 shows the typical modules or building blocks for empirical research programs like a theoretical core (TC), a set of methods, models and mechanisms (MMM), linked to the theoretical core, the embeddedness of TC and MMM within a wider background-knowledge BK, the bridge-modules BM which link the theoretical domain with applications, a class of paradigmatic examples A to 1 A , i.e. , applications of TC, MMM and BM on observable or actually observed n processes as well as an underlying class of observations, data and measurements (DT).1 In Figure I.1, no arrows have been used in order to stress the duality of top-down and bottom up flows. Theoretical concepts, generative mechanisms 1 As a relevant selection from the philosophy of science literature, see Balzer/Moulines/Sneed, 1987, Curd/Cover, 1998, Bunge, 1998, 2002, 2003, Donovan/Laudan/Laudan, 1988, Ludwig, 1990, Schurz/Weingartner, 1998, Salmon, 1998, Sneed, 1991 or Stegmüller, 1981. 46 Karl H. Müller or transfer-modules are as much shaped by the DT-segment as observations, methods and data are determined by the theoretical core, the MMM-segment or the BM-domain. FIGURe I.1 Mapping Empirical Research Programs �� �� ��� �� � � � � � � � ��� � �� Research traditions can be described as a network of research programs and can be visualized in the way of Figure I.2. Here one can see networks of theoretical cores T, of methods, mechanisms and models, of bridge modules M on the one hand and a rich network of different classes of observations, methods and data (DT) and a network of wider application domains (D) on the other hand. The application area changes into larger application domains D to D where each 1 n domain captures a set of paradigmatic examples. The RISC-framework, due to its large set of application domains and to its heterogeneous composition of theories, generative mechanisms or models, can best be characterized as an emerging trans-disciplinary research tradition. The application domains D cover 1 unusually wide areas, ranging, as the introductory article already indicated, from earthquakes, forest fires or sun-flares to income distributions, financial markets, migrations and settlements or the evolution of languages. Introduction to Part I 47 FIGURe I.2 Mapping Empirical Research Traditions as Networks of Research Programs �� � � � � � � � � � � � � �� �� �� �� �� �� � � � � � � ��� � �� �� �� �� �� �� At this point the various components of the RISC-research tradition wil be summarized briefly, using the format of a RISC-introductory course or a RISC-primer with a smal number of basic propositions for each relevant RISC-segment. The TC-Networks of the RISC-Framework 1) The RISC-approach relies on four main theory groups, namely on a general theory of networks, a general theory of systems, a general theory of self-organization with a special emphasis on critical phenomena or structural changes and, finally, on a theory cluster from the cognitive sciences with a special focus on situated or embedded cognition.2 2 See, for example, Adams/Aizawa, 2010, Augoustinos/Walker, 2001, Clark, 1998, 2001 and 2008, Fiske/Taylor, 1991, Noe, 2006, Pfeifer/Bongard, 2007 or Robbins/Aydede, 2008. 48 Karl H. Müller 1.1) The emphasis on general theories of networks, systems, of self-organization and of situated or embedded cognition becomes necessary since these four theory groups comprise a set of different special theories which vary in their theoretical core or in their MMM-domains. For example, the general theory of networks includes different potential network formations like flow networks and relational networks or random networks and scale-free networks. 1.2) The relations between a general theory of networks and of systems should be understood in the following way. Both theory groups are considered as strictly equivalent and irreducible to each other. Both theory groups are powerful description devices for the societal and natural worlds and should be used according to criteria like adequacy, epistemic utility or cognitive viability. Systemic approaches are primarily required wherever boundaries, boundary conditions, closures and environmental interactions become of central importance. Likewise, network approaches are to be preferred when boundary issues are mostly irrelevant and when the main emphasis lies on the interactive micro-dynamics of a specific configuration.3 1.3) Each of the TC’s of special theories of networks, systems, self-organization and situated cognition constitutes a node in the TC-network. Figure I.2 indicates already that the theoretical core of the RISC-framework should be viewed as a network of TCs from network, systems, self-organization and cognitive science theories. 1.4) In general, the general theories of networks and of systems can be seen as the most general description devices within the RISC-framework, whereas the theories of self-organization become relevant for the specification of the structures of networks or systems.4 The theories of embedded cognition serve as the necessary micro-foundations for societal RISC-mechanisms and processes and for the subjective side of RISC-perceptions, expectations and actions. 1.5) The background-knowledge (BK) of the RISC-framework lies, therefore, in more general theories of space and time, of sub-atomic, atomic or molecular composition and dynamics or of general theories of evolution and of self-organization. 3 Of course, it would be possible to classify systems and networks in a different way, for example by characterizing networks as the more general configuration and systems as subsets of networks. This would imply, for example, that elementary systems like input-output systems would have to be classified as strongly reduced networks with basically one node only. Since systemic descriptions are well established across scientific disciplines, such a sub-set relation would violate the well-embedded usages of systemic concepts and apply network formations in domains where basically all ingredients for networks are missing. 4 For a very useful summary of the core concepts of a general theory of systems, see, for example, Bunge 1978, 1979 and for a systemic epistemological outline 1983a and 1983b. Introduction to Part I 49 1.6) In general, the RISC-framework places special emphasis on non-trivial theory compositions and recombinations. This means that only those components from the general theory of embedded cognition should be included which are based on (self )-reflective or state-determined practices and routines.5 1.7) The stock of available theories of societal differentiations and of the long-term historical dynamics becomes partially and selectively integrated into the network of bridge modules (BM) of the RISC-framework. The Network of Mechanisms, Methods and Models within the RISC-Framework 2) With respect to mechanisms,6 methods and models, the main focus in the RISC-framework lies in the specification of generative mechanisms and its corresponding models. Generative mechanisms must fulfill the following basic requirement. They must be able to specify how the interaction between micro-units produces or generates a specific macro-process which in the case of the RISC-framework must be a process with a power-law distribution. Generative mechanisms are to be understood as generic and qualitative explanation sketches. The RISC-framework provides a comprehensive set of generative mechanisms both within societies and in the environment of societies. The former are to be classified as internal RISC-mechanisms, the latter as external RISC-mechanisms. 2.1) Within the internal or the societal side of RISC-mechanisms, a special emphasis is placed on two generative mechanisms, one in the sphere of economic production and one in the field of knowledge production. 2.1.1) For the side of economic production, the RISC-framework provides the specification for a capitalist network formation with a large number of micro-units, with average profits as long-term reproduction requirement, and with innovations, very broadly understood, as central internal disturbance components. 2.1.2) For knowledge production, the science network can be characterized in its composition by a large number of micro-units, by a multiplicity of research programs and research traditions, by an average problem-solution capacity as long-term reproduction requirement for research programs and research 5 For more information on state-determined schemes of embedded cognition, see especially the article by Günter Haag, Karl H. Müller and Stuart A. Umpleby “Toward Self-Reflexive Forms of RISC-Modeling” in this volume. 6 On the seemingly contradictory notions of social or societal mechanisms, see, for example, Hedström/Swedberg, 1995, Schmid, 2006 or Schmidt/Florian/Hillebrandt, 2006. 50 Karl H. Müller traditions, and by scientific innovations, widely understood, as the central internal disturbance elements. 2.1.3) These two RISC-mechanisms, the one for economic production and the one for knowledge production, should be specified as coupled or linked. They operate, thus, in a coevolutionary fashion and are situated at the core of societal RISC-processes and their long-term coevolution. 2.2) Aside from the two generative RISC-mechanisms for economic and knowledge production, societies are characterized by a multiplicity of additional RISC-mechanisms in areas like migrations and settlements or in the cultural and artistic spheres. 2.3) For the external or the environmental side of RISC-processes, one finds a large number of independent RISC-mechanisms like tectonic plates and their dynamics or ecological systems and their long-term development patterns. 2.4) Furthermore, RISC-mechanisms can be specified across the five layers of the Earth’s atmosphere, namely in the troposphere, the stratosphere, the mesosphere, the thermosphere and the exosphere. 2.5) Relevant RISC-mechanisms are distributed outside the Earth’s atmosphere as well and processes like the permanent inflow of cosmic material, including large-sized meteorites, should be seen as the paradigmatic example for an external RISC-process outside the Earth’s atmosphere which has produced and still can cause a massive impact for regional societies and the global environment. 2.6) Each of these internal or external RISC-mechanisms can be characterized by a variety of RISC-models and each RISC-model, iin turn, can be used for a variety of RISC-mechanisms. For example, model groups like self-organized criticality (SOC), complex networks (CN) or master equation networks (ME)7 can be used for the modeling of a variety of internal and external RISC-mechanisms. 2.7) With respect to common methods, the statistical theory of extreme events constitutes the central component within the RISC-framework.8 Bridge Modules (BM) for the Study of Modern RISC-Societies 3) Due to wide diversity of application domains, bridge modules play an important triple role in the formation of the RISC-framework. These bridge modules serve three main purposes. First, they provide the guidelines for the 7 See, for example, the article by Günter Haag “New Models for Generating Power Law Distributions” in this volume. 8 As a relevant reference, see for example, Embrechts/Klüppelberg/Mikosch, 1996 or Resnick, 2007. Introduction to Part I 51 specification of a new architecture for contemporary societies. Here, these bridge modules must identify the constituent elements, the structures and, especially important, the inter-linkages and the coevolution between these constituent components. Second, along the vertical axis of Figure I.2, these bridge modules must establish the necessary theoretically mediated descriptive and historical accounts for a RISC-application domain D. Third, along the horizontal axis of Figure I.2, the bridge modules provide the necessary horizontal connections between various RISC-application domains as well. While in traditional approaches the specification of bridge modules for internal and for external RISC-processes is undertaken independently from each other, the RISC-framework puts heavy emphasis on the coevolution of internal and external RISC-processes which should be reflected in the construction of bridge modules as well. Thus, the horizontal components of bridge modules become of special importance since they specify the inter-linkages between the societal domains on the one hand and various groups of external RISC-networks or systems on the other hand. Within the RISC-framework, a relatively large number of modules is needed which, in conjunction, forms the network or RISC-relevant bridge modules. 3.1) The most important bridge modules for the RISC-framework lie in the specifications for a new societal architecture which can be classified, inter alia, as the infrastructural constitution of societies [see also Figure I.3]. Modern RISC-societies should not be specified, as in traditional perspectives, as a composition of functional systems or networks for economic production and distribution, for the political sphere, for the science field, for cultural fields, the life world domains, etc. Rather, the internal domains of modern RISC-societies should be conceptualized as a tripartite configuration, composed of (1) a variable number of functional systems or networks for economic production, for knowledge production, culture and arts, media, household (re)production, etc., (2) by infrastructural networks or systems in the three domains of energy, information and transport and (3) by RISC-protection networks or systems, understood in an unusually wide fashion. 3.2) A first bridge module is needed for the class of production or functional networks or systems with a large number of internal RISC-processes. These functional or production networks or systems lie in the sphere of economic production and distribution, educational production, knowledge production, household (re)production, news and entertainment production, arts and cultural production, health production and the like. Aside from the domains just mentioned, modern RISC-societies can be described by any re-combination of these ensembles, too. With respect to the bridge module for the economic sphere, a special emphasis should be laid on the level differentiation in the 52 Karl H. Müller economic sphere and on the phenomenon of vertical or horizontal production chains where economic production and distribution processes are organized at three levels, ranging from agriculture and raw material extraction (primary level) to industrial production (secondary level) and services (tertiary level).9 FIGURe I.3 The Infrastructural Constitution of RISC-Societies ������� ��������� ����������������� ��������� ������� ������ ����������� ��������� 3.3) A second bridge module must be focused on ensembles in the three infrastructural spheres of energy, information and transport. These infrastructural networks or systems are organized in an accumulative mode where a new infrastructural network or system operates in conjunction with already existing 9 See, for example, Fischer/Reiner/Staritz, 2010, Froebel/Heinrichs/Kreye, 1977 or 1986. Introduction to Part I 53 ones. In general, infrastructural networks or systems are not crowding out or substituting previous ones. As an important element in the specification of the infrastructural bridge modules, a special emphasis must be devoted on the coevolution of these three ensemble groups and on the bottlenecks and shortages in a specific infrastructural segment, due to large-scale innovations in another infrastructural segment. 3.4) The third bridge module is devoted to the area of RISC-protection networks or systems which cover a wide class of ensembles which, in the spirit of Karl Polanyi,10 can be seen as the societal protective belt which limit and contain the detrimental consequences of RISC-mechanisms in the production networks or systems. Following Karl Polanyi, the political system or network, NGOs or the civil society in general belong to the RISC-protection networks or systems. An important class of RISC-protection networks or systems consists in those ensembles that deal with the catastrophic consequences of rare events of external RISC-mechanisms and processes like droughts, floods or earthquakes. It seems very useful to treat the available societal support-capacities as an important sub-segment of the RISC-protection networks under the name of RISC-support networks or systems. 3.5) The fourth bridge module is concentrated on the overall structures, forms or designs of these three large societal ensembles. Here, the following guiding principles become relevant. 3.5.1) First, the infrastructural networks or systems in the three domains of energy, information and transport enable and constrain the unfolding of societal networks in their energy, information and transport potentials and capacities. 3.5.2) Second, functional systems and networks of production on the one hand and RISC-protection systems on the other hand should be specified according to a homeostat11 model where the RISC-protection networks assume the role of an under-critical regulator RPROT. 3.6 ) The fifth bridge module covers the subjective or the human side of networks and ensembles. Contrary to Jürgen Habermas’ separation between systems and life worlds, the RISC-approach stresses a variety of specification possibilities for networks and systems and points to the possibility for exclusive, trivial and highly inclusive non-trivial network or systems specifications. In short, the four bridge modules can be based on trivial mechanical models without any subjective or life world-components to them, on models with trivial action schemes like rational action or bounded rationally and, thus, of restricted or stylized life 10 See, for example, Polanyi, 1978. And for the Polanyi-configuration of networks or systems of production and for protection, see, for example, Müller, 1995. 11 On the model of the homeostat, see especially Ashby, 1957. 54 Karl H. Müller world-elements or, finally, on models with non-trivial action schemes which are based on the stock of theories on situated cognition. The last group, while highly complex, is able to account for life world components in their fullest form. 3.7) Additional bridge modules are required, obviously, for the empirical domains of external RISC-mechanisms and processes as well. For each of the larger application domains from Table 2.1 in the subsequent article on “RISC-Processes and Societal Coevolution” like earth quakes, epidemics, environmental systems or forest fires, but also for atmospheric RISC-processes in the form of tsunamis, floods or droughts specific bridge modules are needed which provide the general specification guidelines for RISC-applications. Observations, Measurements and Data (DT) for the RISC-Framework 4) The RISC-approach is focused on the totality of processes with a power law distribution and the network of observations, measurements and relevant RISC-data (DT) cover, thus, three very heterogeneous functional classes of information. 4.1) In general, the overall data network DT comprises both quantitative and qualitative data and includes different media like official macro-data, panels and surveys, documents, observation reports, diaries, pictures, films, etc. These different and heterogeneous sets can be qualified as DT-components of the RISC-DT-network. 4.2) The first functional data class comprises all DT-components which are needed for the statistical and for the model analysis of internal and external RISC-processes. 4.3) Aside from data sets for power law distributions as the constitutive empirical units of the RISC-framework, one must add a second functional class of relevant DT-components on the long-term coevolution of the three societal segments, namely functional production networks and systems, the three infrastructural networks or systems on energy, information and transport and the RISC-protection networks or systems, including the special segment of RISC-support networks or systems. 4.4) A third relevant functional class of DT-components is needed for the inner and the outer environments of modern RISC-societies and their long-term development or coevolution. 4.5) The RISC-framework, due to its heterogeneous DT components and due to its reliance on three functional sets of relevant DT-components, poses new challenges for data integration and data management which have to be solved along the unfolding of the RISC-framework as a trans-disciplinary research tradition. However, the problems of DT-integration are generic ones and apply to all trans-disciplinary research traditions with heterogeneous DT-networks. Introduction to Part I 55 Paradigmatic Applications for the RISC-Framework 5) The application domains of the RISC-framework can be summarized in two alternative ways, namely according to the different domains of RISC-mechanisms and processes or according to two very broad problem domains, namely with respect (1) to a very comprehensive notion of sustainability on the one hand and (2) to a RISC-based concept of viability on the other hand. 5.1) The separation according to the domains of RISC-mechanisms and processes can be conceptualized in the format of an additive list and Table 2.1 in the subsequent article provides such an overview of RISC-application domains. 5.2) Along the second perspective of RISC-applications, two very broad and general problem domains stand in the center of RISC-applications, namely sustainability and viability. Both domains must be separated into several principal dimensions each. Moreover, these sustainability and viability dimensions cover core problems with respect to the medium and long term maintenance and survival of contemporary RISC-societies. 5.3) Viability and sustainability of modern RISC-societies and their principal dimensions should be viewed as strictly independent from each other since RISC-societies may be RISC-sustainable and not RISC-viable—and vice versa. 5.4) The problem of sustainability of modern RISC-societies is dependent on three principal dimensions, namely on sustainability (1) with respect to the globalization of currently advanced production systems, (2) with respect to future generations and (3) with respect to RISC-robustness. Each of these dimensions is strictly independent from each other and the space of sustainability dimensions covers almost any recombination of specific sustainability states in each dimension. The central research problem with respect to the sustainability of modern RISC-societies addresses the feasibility of trajectories or of drifts towards high sustainability levels across all three sustainability dimensions. TAble I.1 A Typology of Modern RISC-Societies Viability low High low Type I Type II Sustainability High Type III Type IV 5.5) Turning to the problem of the viability of modern RISC-societies the main emphasis lies on two principal dimensions, namely (1) on the subjective 56 Karl H. Müller evaluations and assessments in core areas like overall life-satisfaction, happiness or quality of life and (2) on institutions of basic or fundamental rights and their actual performance levels. Thus, the central research problem with respect to the viability of modern RISC-societies is focused on the feasibility of trajectories or of drifts towards high viability levels across the two viability dimensions. 5.6) Aside from these two general core problems and applications, RISC-applications should provide at least partial answers to the two core problems of sustainability and viability. For example, studies of financial crashes or floods should be discussed alongside their sustainability and viability impacts. 5.7) The RISC-framework can be seen, thus, as an emerging research tradition which opens new perspectives and produces new solutions with respect to the sustainability and to the viability of contemporary societies which recombines large segments of scientific fields in the natural and in the social sciences which, so far, treat their domains of investigation in splendid isolation from each other. The Long-Term RISC- Perspectives on Societal Evolution 6) So far, the coevolution of RISC-societies can be described in three long-term stages where each stage is characterized by a unique distribution of RISC-mechanisms and processes. 6.1) The first stage of RISC-societies was characterized by the absence of RISC-mechanisms in the sphere of economic or knowledge production and by strong impacts of external RISC-processes. The dominant societal systems or networks were situated in the political sphere. Moreover, the world was separated into a configuration with a number of societal islands surrounded by a natural RISC-environment whose impact for economic production in agriculture was strong and direct. The first stage comprised RISC-societies of varying degrees of complexities, including world empires of limited regional scope and covers the entire period of human history prior to a critical transition phase in the 14th and 15th century. 6.2) The second stage in the evolution of RISC-societies starts with the long 16th century and brings the implantation of two societal mechanisms which, in combination, generate a variety of societal RISC-processes. Thus, the second stage sees the irreversible diffusion of the capitalist mode of network and system formations in the domain of economic production and of the science mode of network and system formations in the area of knowledge production. The second stage comes to an end approximately in the decades between 1900 to 1950 when the global diffusion of these two RISC-mechanisms has reached its regional limits. Introduction to Part I 57 6.3) The third stage of RISC-societies starts around 1945 and is characterized by new features like global economic micro-units in the form of transnational enterprises and of global actors in the RISC-protection systems or networks like the United Nations, the World Bank or the International Monetary Fund (IMF). The three most relevant new elements, however, come from the domain of RISC-mechanisms and processes. 6.3.1) First, a rare event in technological innovations is focused, for the first time, on the infrastructural networks and systems for information and brings about an enormous increase in network densities both at global and at local scales. 6.3.2) Second, internal and external RISC-mechanisms become linked and coupled in unprecedented ways and the internal RISC-engines exert their impacts on the external RISC-dynamics and vice versa. The most striking example of coupled RISC-processes runs under the heading of climate change where the outputs of the capitalist RISC-engine and its related societal production networks or systems change the composition of the atmosphere in a critical manner and alter, thus, the distributions for atmospheric RISC-processes like floods, droughts, tsunamis, hurricanes, tornados, etc. 6.3.3) Third, these new external RISC-distributions, in turn, have a direct impact on internal RISC-processes, most notably on the capitalist RISC-engines in areas like finance, insurance, but also in agricultural production and across the different levels of the production chains. 6.3.4) Thus, the third stage of RISC-societies can be described with attributes like high levels of complexity, unpredictability, emergent phenomena or a growing number of un-intentional effects. Central Theoretical Propositions of the RISC-Framework 7) The dynamics, the sustainability and the viability of modern RISC-societies results from the interplay of external and internal RISC-mechanisms and, correspondingly, from the distribution of internal and external RISC-processes and from the capacities of RISC-protection networks or systems. 7.1) Internal RISC-mechanisms are embedded in three main network groups or systems, namely, first, in the multi-level economic production and distribution networks or systems, in the knowledge production networks or systems and in other functional societal production ensembles, second, in the infrastructural networks or systems for energy, information and transport and, third, in the group of RISC-protection networks, including the RISC-support networks for prevention and damage control. During the second and the third stage of RISC-societies, both the inter-linkages between these three large-scale societal segments and between internal and external RISC-processes have increased significantly and continue to increase. 58 Karl H. Müller 7.1.1) With respect to internal RISC-mechanisms, very large-scale innovations in the field of economic or knowledge production should not be seen in the usual way as long swings or cycles, but as rare events and as the outcomes of a RISC-mechanisms in the sphere of economic or knowledge production. 7.1.2) Consequently, very large scale technological innovations happen, due to their very high diffusion potentials, in one of the three infrastructural networks or systems for energy, information and transport. Moreover, large scale innovations, medium innovations and smal -scale innovations should be specified in a generative mode where large scale innovations provide the basis for medium or smal -scale innovations and small or medium scale innovations give rise to large-scale innovations. Similarly, rare events or large-scale scientific innovations should be specified in a generative network of smal -, medium- and large-scale innovations. 7.1.3) The substitution potential of the global configuration of RISC-societies, due to higher network densities or systemic links, is growing substantially. In terms of network densities or systemic links, the global RISC-ensemble achieves, thus, higher level of S-robustness where S stands for substitution. 7.1.4) Conversely, the accumulation of scale-free networks in the infrastructural domains of transport and information leads to lower levels of RISC-NC-robustness where N stands for network nodes and C for systemic components. Phrased in network terms, the col apse of only a few central nodes, due to the high linkage densities across the infrastructural and societal domains, could lead to extremely serious effects for the maintenance of modern RISC-societies as a whole. 7.2) Due to the coupling of internal and external atmospheric RISC-processes both contemporary RISC-societies and their environments are entering a period of increasing stress.12 7.2.1) The increasing societal stress results from the fact that societal systems or networks operate on average events and smaller or larger deviations from relatively constant average events, and not on shifting averages. These shifting averages, however, are the necessary by-product of the new stage of coupled internal and external atmospheric RISC-processes. 7.2.2) Likewise, the growing environmental stress is due to the fact that the boundary conditions for environmental actors are changing more rapidly and, in many instances, cross critical survival and reproduction values for environmental actors. 12 For an interesting background on integrated societal-environmental couplings, see Hol ing, 1978, Gunderson/Hol ing, 2002, Gunderson/Allen/Hol ing, 2010 or Perrings/Mäler/Folke/ Hol ing/Jansson, 1995. In this context, stress has been introduced as a “general term … for effect of human use” Scheffer/Westley/Brock/Holmgren, 2002:197. Introduction to Part I 59 7.3) Due to the couplings of internal and external RISC-mechanisms and processes and due to the strongly interlinked connection patterns of contemporary RISC-societies, rare events in the environment of RISC-societies acquire an increasing potential for large-scale detrimental effects in two general directions, namely, first, along the functional direction across internal societal production networks and systems, the infrastructural networks or systems and the protection networks and systems and, second, along the spatial direction, following a path towards globalization.13 In other words, the E-robustness of RISC-societies with respect to external or environmental RISC-processes decreases. 7.4) Internal RISC-processes, due to their complex and self-organized dynamics, cannot be effectively controlled by the RISC-protection networks or systems. Even worse, the control capacities of RISC-processes in the third stage of RISC-societies diminish significantly due to a growing asymmetry between global actors in the sphere of economic production and distribution and the absence of strong global actors in RISC-protection networks and systems. For the decades ahead one will observe a governance gap between the global actors of the economic RISC-engine and the national and weakly supra-national or very weakly global actors of the RISC-protection networks. Moreover, RISC-protection networks, due to their under-critical control capacities, create a growing number of unintentional effects. 7.4.1) Rare events with highly negative societal impacts are usually accompanied by upward changes in the control- and steering capacities of the RISC-protection networks or systems. 7.4.2) The protection networks or systems are characterized by a time lag of one period since the increase in control or steering capacities is directed towards the control or the prevention of the last rare event and not towards the next one. 7.5) The RISC-support networks or systems for disasters operate usually in an under-critical stage, because they are based on the assumptions of small deviations from average events and not on large scale-deviations in the form of rare events. Rare events in the case of external RISC-processes are usually situated beyond the rescue and relief-capacities of the available support networks or systems. 13 Think, for example, of the eruption of the volcano underneath Iceland’s Eyjafjal ajokul around the year 1600, at the beginning of the second stage of RISC-societies. At this point of time, the eruption would have been unnoticed outside Iceland. By 1800, the eruption would have been recorded with long time-delays in areas outside Iceland, but no detrimental effects on economic production would have occurred. Around 1900, the same result would have been recorded. Even in 1940, the overall societal effects would have been smal . In April 2010, the eruption was linked primarily with the infrastructural transport networks (air traffic) and secondarily with economic production networks and other functional networks and systems as well. Moreover, these negative effects, while concentrated on Europe, were spread across the entire world. 60 Karl H. Müller 7.6) In instances of actual disasters, the logic of situated cognition and the subjective side of RISC-perceptions, of fear or of risk-expectations must be viewed as an essential element which should not be eliminated by trivial action models or by idealized schemes of local needs and a corresponding outside support. In general, RISC-support networks or systems do not operate in a blank slate and clashes between the local population, affected by a disaster, and support from outside are to be viewed as the necessary results of such a configuration. Usually, one can observe a striking incongruence between the perceptions and expectations on the part of the affected population and the perceptions and expectations of the outside support moving into a disaster region. 7.7) The growing couplings and interdependencies between societal and natural RISC-mechanisms and processes and the organization and structures of contemporary RISC-societies as hyper-complex adaptive networks or systems have multiple consequences for the sustainability and for the viability of the global network or system of contemporary RISC-societies. 7.7.1) Due to the architecture and the self-organizing nature of the dynamics of contemporary RISC-societies simultaneous increases across all five dimensions of sustainability and viability are clearly beyond the control-capacities of the RISC-protection networks or systems. 7.7.2) While the coevolution of RISC-societies during the second and the third stage exhibits an arrow of complexity in terms of composition and linkage densities, no arrow of sustainability and viability can be identified. Rather, the dynamics with respect to the sustainability and viability of contemporary RISC-societies depends on the self-organizing coevolution of the overall RISC-ensembles and on local or global drifts. With respect to sustainability and viability, the hyper-complex configuration of the overal RISC-ensembles operates far from equilibrium in a mutually adaptive or coevolutionary mode in both upwards and downwards directions of the five-dimensional sustainability and viability space. 7.7.3) While the coevolution of the economic RISC-mechanisms and processes operates on a global drift toward higher levels of globalization, differentiations and inequalities,14 no long-term natural drifts can be observed with respect to sustainability and viability. On the contrary, due to the couplings of external and internal RISC-mechanisms and processes one can assume, for the decades ahead, a negative correlation between the coevolution of the economic RISC-ensembles and the sustainability and viability dimensions. 14 According to an estimate by the economic historian Paul Bairoch, 1995, the levels of GDP p.c. between today’s First and Third World were roughly equivalent at roughly 180 US$ as late as 1750. Introduction to Part I 61 7.7.4) Within the five-dimensional sustainability and viability space, trade-offs between sustainability and viability should be expected as normal or reference cases, and correspondences or complementarities between sustainability and viability as rare events. Increasing the degree of robustness (sustainability ) for 3 example, does not lead to increases in the viability dimensions and vice versa. 7.7.5) Similarly, with respect to the sustainability dimensions alone, trade-offs between the three sustainability dimensions should be seen as the normal outcome and simultaneous increases across all three sustainability dimensions as improbable events. The same situation applies to the two viability dimensions where trade-offs are to be expected as the regular outcomes. 7.7.6) Moreover, the five-dimensional space of sustainability and viability is characterized by lock-ins and path dependencies where the coevolution of the overall RISC-ensembles has no effects on a particular sustainability or viability dimension and where a given sustainability or viability level can only be changed by rare internal or external RISC-events themselves. 7.7.7) From a global perspective the biggest challenge for contemporary RISC-societies lies in the construction of functional boundaries which are capable to contain the negative or disastrous effects of rare events. Such an emphasis on spatial separations should become of special relevance in the new designs for the emerging new cluster of converging technologies.15 15 On converging technologies, see the article by Toni Pustovrh in this volume. RISC-Processes and Societal Coevolution: Towards a Common Framework 2 Karl H. Müller 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 16 13 15 One of the persistently puzzling issues for contemporary theories of societal evolution or coevolution1 concerns the engines or, alternatively, the (f )actors for societal changes. To be sure, the literature is full of potential candidates for generative mechanisms2 of societal coevolution like, to mention a few, functional differentiations into specific societal sub-domains or systems,3 knowledge as a new wealth generating factor of production,4 configurational and structuration dynamics,5 great ideas,6 the Schumpeterian entrepreneurs pursuing their recombinations of factors of production or innovations for short,7 forces of production, revolutionizing both economic sub-structures and societal super-structures,8 a technological-ly driven transition from low and medium risk production to high risk production processes with substantial ramification for individual life-courses,9 urbanization,10 a small set of crucial variables relevant for long-term sustainability11 or a permanent interplay between systemic differentiations and life world developments.12 So 1 Societal evolution at the regional, national or global levels is almost by necessity coevolutionary in nature. Societies at the regional or national levels are horizontal y embedded in an environment of coevolving regional or national societies and the societies at the global levels interact with their global environments in a coevolutionary manner, too. Thus, in the context of this article, the terms societal evolution and societal coevolution are used in a strictly equivalent fashion. See also Durham, 1991, Lewontin/Oyama, 2000, Oyama, 2000 or Thompson, 1994. 2 Subsequently, the concepts of generative mechanisms or, alternatively, of generative engines will be used for the following minimal configuration, namely for a process P as a sequence of events and for an ensemble EN which can be attributed with the production of this sequence of events P. The ensemble EN could be a system, a network or any other complex configuration like a system of systems, a network of networks, a system of systems and networks, etc. Typically, generative mechanisms or engines involve a dual level configuration between a set of micro-actors AMI who are dynamically inter-linked with one another and who through their aggregated micro-dynamics produce or generate the macro-process PMA. 3 See, e.g. , Parsons, 1951, 1964 and 1994 or Luhmann, 1984 and 1997. 4 Along the knowledge line, see Bell, 1979a and 1979b, Drucker, 1993, Nelson, 1996 or Thurow, 1996, 1999 and 2005. 5 See especially Giddens, 1984, 1991, 2000 and 2009. 6 As a recent example, see Ogle, 2008. 7 On the Schumpeter system, see, for example, Arthur, 2009, Schumpeter, 1934, 1952, 1961, 1975, 1991, or Weidlich/Haag, 1983. 8 Under this category fall all Marxist approaches that use the distinction between an economic production ensemble and a collection of societal systems which are strongly influenced by the economic production ensemble and which, in turn, have a limited capacity to influence or control this production ensemble. 9 See, for example, Beck, 1986, 1997, 1998a, 1998b, 2000, 2002 or 2007. 10 On urbanization see, for example, Florida, 2002 or 2005. 11 See, especially, Diamond, 2005, with his five driving forces of environmental damages, climate changes, hostile neighbors, loss of trading partners, inappropriate reactions to change. 12 Of course, Jürgen Habermas is the most relevant source of reference in this area as can be seen from Habermas, 1968, 1981 or 1984. 66 Karl H. Müller far, the problem of rare events remained outside the mainstream discussions on the driving (f )actors for societal development, growth or coevolution.13 Thus, the next steps will introduce the notion of RISC-processes (Rare Incidents, Strong Consequences) within an evolutionary theory landscape and build up a set of crucial components for an evolutionary RISC-based approach to societal unfoldings, differentiations and complexifications. 2.1 RISC-Processes as the Missing Links in the Theories of Societal Coevolution Formally, a RISC-process is characterized by a specific distribution and an underlying distribution function where a very large number of minor or marginal events is accompanied by a very small number of very large-scale events. RISC-processes occur within societies as well as in their environment. Societal or internal RISC-processes comprise areas like the global finance system with rare occurrences of severe global crises in 1893, 1929, 1987 and 2008 or the current global information and communication networks with a very large number of marginal and local network defects and rare incidents of major failures with widespread and disastrous consequences. Natural RISC-processes in the environment of societies can be found, for example, in earthquakes with very rare instances of earthquakes with deep impact and catastrophic consequences and a very large number of very weak quakes. In the words of Didier Sornette, RISC-processes exhibit a wild distribution and can be qualified, thus, as wild processes. In contrast, the bell shaped normal distribution can be described as a mild distribution and the underlying processes, consequently, as mild ones. What is the probability that someone has twice your height? Essentially zero! The height, weight and many other variables are distributed with ‘mild’ probability distribution functions with a well-defined typical value and relatively small variations around it. What is the probability that someone has twice your wealth? The answer of course depends somewhat on your wealth but in general there is a non-vanishing fraction of the population twice, ten times, or even one hundred times wealthier as you are. [Sornette, 2006:104] 13 On the current scope of disaster research, see, e.g. , Felgentreff/Glade, 2008, Quarantelli, 1998 or Rodriguez/Quarantelli/Dynes, 2007. RISC-Processes and Societal Coevolution 67 FIGURe 2.1 Two Distributions of a RISC-Process 1a Linear Scale ��������� � ���������� 1b Dual Logarithmic Scales ��� ��������� � �������������� Table 2.1 offers a short overview on the ubiquity of RISC-processes with respect to natural or social domains and with respect to different scientific disciplines involved. TAble 2.1 RISC-Processes Across Different Scientific Disciplines RISC-Process/ Distribution-Characteristics Scientific Discipline Rare Incidents, Strong Very Frequent Incidents, Consequences Weak Consequences Natural Science Domains Sandpiles/ Physics very small number of very big very large number of very small avalanches avalanches earthquakes/ very small number of very large number of earth sciences earthquakes with very strong earthquakes with very small effects effects Solar flares/ very small number of very very large number of small Astronomy strong outbursts outbursts 68 Karl H. Müller CONTINUING TAble 2.1 Tornados, hurricanes, very small number of very very large number of taifuns, etc. Meteorology devastating tornados, atmospheric turbulences hurricanes, taifuns, etc. Forest fires/ very small number of fires with very large number of fires with environmental sciences very large scale consequences very local effects Viruses and epidemics/ very small number of new very large number of new Medical Research viruses with very large- scale viruses with no or marginal effects effects ecological systems/ very small number of very large number of vanished environmental sciences breakdowns with very large- species with no or marginal scale effects effects The brain/ Neuro- very small number of neurons very large number of neurons cognitive sciences with a very high number of links with a very low number of connections Social Science Fields language/ very small number of words very large number of words linguistics with a very large number of with a very small number of occurrences (in books, plays, occurrences (in books, plays, etc.) etc.) Scientific quotations/ very small number of articles, large number of articles with no Science studies quoted with very high quotations or zero-impact frequency or very high impact Scientific breakthroughs/ very small number of institutes very large number of institutes Science studies with a very large number of with no scientific breakthroughs scientific breakthroughs Innovations/ Science- very small number of very large number of technology-society innovations with far- reaching innovations with near-zero effects and very strong effects or re-percussions repercussions Financial markets/ very small number of crashes very large number of Finance with very strong effects fluctuations with very small consequences Wealth and income/ very small number of very high very large number of persons economics income or wealth or households with small or medium income Power grid/ energy very small number of very large number of minor sciences accidents/failures with very accidents/failures with no or widespread consequences marginal effects Migration and settlement/ very small number of very large very large number of small Sociology, demography cities within a nation settlements RISC-Processes and Societal Coevolution 69 Another interesting aspect of the overall RISC-framework lies in the micro-operators or actors and in their relevant operations which constitute or produce a RISC-process. As can be seen from Table 2.2, operators and operations comprise very heterogeneous sets, ranging from simple operators like a grain of sand to complex ones like firms, organizations or scientific institutes and from simple operations like dropping/fal ing down to very complex ones like doing research work, publishing or quoting articles. It is important to note that the focus on operators and operations does not assume a reference model of behaviour/action or a set of maximization or minimization rules. Operators and operations can be specified freely and can include trivial and non-trivial operation schemes.14 TAble 2.2 Selected Examples for Micro-Operators, Micro-Operations and Mechanisms in the Production of RISC-Processes RISC-Building Blocks RISC-Micro- RISC-Micro-Operations RISC-Mechanism Operators Natural Science Domains Sandpiles a single grain of dropping down (on a a non-random growth sand specific spot) mechanism earthquakes a tectonic plate movements and inter- dynamic plate mechanism Actions Forest Fires a tree, bush, etc. inflaming, interactions local diffusion mechanism with neighbouring trees or bushes, etc. Viruses and a single virus replication, movement, epidemiological diffusion epidemics interactions, mutation mechanism ecological a species reproduction, inter- eco-network-mechanism systems actions, mutations Cognitive a neuron signaling neural propagation systems mechanism Social Science Fields language a competent bindings of language societal network language user elements mechanism Scientific a single scientist quoting articles science network quotations mechanism Scientific a scientific solving new research science network breakthroughs institute problems mechanism 14 On situated or embedded cognition, see, for example, Adams/Aizawa, 2010, Augoustinos/ Walker, 2001, Fiske/Taylor, 1991, Noe, 2006 or Pfeifer/Bongard, 2007. 70 Karl H. Müller CONTINUING TAble 2.2 Innovations a firm or an changing the production economic network organization processes mechanism Financial a single trader selling and buying, economic network markets innovating products or mechanism processes Income and a single individual income from economic network wealth or household employment or self- mechanism employment Migration and a single individual moving to a new place, societal network settlement or household spatial attractivity mechanism differences From Tables 2.1 and 2.2 it becomes clear that RISC-processes can be characterized as self-organizing since all the examples in Tables 2.1 or 2.2 have no units for effective steering and control. No malevolent or benevolent controller or demon is in sight for arranging, fol owing Table 2.2, the succession of sand avalanches, the order of magnitude of earthquakes, the severity of forest fires, the power of hurricanes or tornados, the diffusion degree of epidemics, the severity in the breakdown of ecological systems, the ordering of neural waves, the frequency order of words, the frequency distribution of scientific quotations, the orchestration of scientific breakthroughs across institutes, the rank-size distribution of firms or, finally, the severity of financial crises. In other words, RISC-processes are mostly generated through the characteristic micro-operations of their micro-operators although one can identify control units in many of the examples in Table 2.2 where these control units remain well embedded and couched in an overall self-organization ensemble. Furthermore, while all RISC-processes can be classified as self-organizing, the relations between RISC-processes, self-organization and evolution are complex and can be captured with the help of Figure 2.2. While all evolutionary processes can be characterized as self-organizing, the converse relation does not hold since not all self-organization processes should be qualified, at the same time, as evolutionary ones. As an example, take the case of the dynamics of tectonic plates and the issue of earthquakes which, for obvious reasons, is to be qualified as a non-evolutionary self-organizing process. Similarly, all RISC-processes turn out to be self-organizing ones, but the reverse side does not hold either since not all self-organization processes are distributed in a RISC-like fashion. As an empirical example, take the size distribution in a species which is normally and, thus, mildly distributed. Finally, not all RISC-processes are to be qualified as evolutionary ones and not all evolutionary processes exhibit a RISC-distribution. Thus, Figure 2.2 summarizes the intricate relations between RISC-processes, RISC-Processes and Societal Coevolution 71 evolution and self-organization. Additionally, RISC-processes are produced by generative mechanisms which can be divided into two broad clusters, namely into non-evolutionary and evolutionary mechanisms. In terms of demarcation criteria, an evolutionary mechanism requires an endogenous proliferation of novelty as well as a dualism in the micro-constitution of evolutionary actors. In biology, this dualism has a well-defined meaning,15 since the observable properties, structures and processes of an organism as micro-actor belong to its phenotype and the sequence of nucleotides, forming the DNA of an organism are qualified as its genotype.16 FIGURe 2.2 The Relations between RISC-Processes, Evolutionary Processes and Self-Organization Processes ��������������� ��������� ������������ ������������ ����� ��������� ��������� Moreover, the evolution of evolution in general exhibits a very illuminating RISC-characteristic as well because innovations or mutations in the history of life exhibit only a very small number of very profound changes and transitions and a very large number of marginal or minimal changes. The following table, compiled by John Maynard Smith and Eörs Szathmáry [1996:5], offers an overview on the very small number of big evolutionary jumps. Moreover, it must be added that these innovative jumps and transitions were accompanied by a widening of the evolutionary landscape and not by a complete substitution, something which can be captured in Figure 2.3. 15 On this point, see, for example, Feldman 1988:43 and Maynard-Smith, 1974, 1982b or 1989. 16 Due to this separation of domains, an interesting point could be made that any evolutionary theory from its very outset is coevolutionary in nature. On this point, see especially Margulis, 1981, 1993 or 1998. 72 Karl H. Müller Focusing on the generative mechanisms and models of RISC-processes, the relations between RISC-processes and generative mechanisms are intricate and can best be described as a dual one:many-relationship. For each RISC-process one usually finds a non-empty set of generative mechanisms and models which can be used for empirical analysis. Likewise, each generative mechanism or model can be utilized in several different contexts and, thus, for a variety of different RISC-processes. TAble 2.3 Very Large Scale Transitions in Evolutionary Time Previous State Transition Phase New State Replicating molecules → populations of molecules in compartments Unlinked replicators → chromosomes RNA as gene and enzyme → DNA and protein (genetic code) Prokaryotes → eukaryotes Asexual clones → sexual populations Protists → animals, plants and fungi (cell differentiation) Solitary individuals → colonies Primate societies → Human societies (language) FIGURe 2.3 The Pattern of Evolutionary Unfoldings ������������ ������ ����������������� At the current point it must be posed as an open question whether the different generative mechanisms and models will converge to one or a small number of second-order mechanisms and models generating mechanisms and models. Thus, Figure 2.5 leaves it open whether the model cluster at the deep-structure will consist of one, two or several independent clusters of generating mechanisms of generating mechanisms. RISC-Processes and Societal Coevolution 73 FIGURe 2.4 The One-Many Relationships between RISC-Processes and Generative Mechanisms ����� �������� ���������� ���������� ���������� ������ ������ ������ ����� ����� ����� �������� �������� �������� ���������� ������ At this point it could be interesting to continue with a short discussion on the differences and similarities between the concept of RISC-processes and the focus on rare events with large-scale societal consequences on the one hand with the meanwhile well-recognized notion of risk-societies17 on the other hand. At the outset, RISC-societies, i.e. , societies with an ensemble of internal and external RISC-mechanisms and processes, differ from post-modern risk-societies in a fundamental manner, since risk-societies, following Ulrich Beck and others,18 emerged as the latest phase of capitalist development only whereas 17 As locus classicus, see Beck, 1986. 18 On risks and risk-research, aside from Ulrich Beck’s risk-society, see also Adams, 1995, Banse, 1996, Bernstein 1996, Bonß, 1995, Caplan, 2000, Dembo and Freeman, 1998, Douglas, 1992, Dowd, 2005, Fischhoff et al., 1981, Gardner, 2008, Graham and Wiener, 1995, Pidgeron, 74 Karl H. Müller RISC-societies and their evolution can be traced throughout the entire history of human societies. Figure 2.6 and Table 2.4 offer some guidelines on the special relations between contemporary risk-societies and RISC-societies. FIGURe 2.5 The Convergence Towards Small Clusters of RISC- Mechanisms ����� ����� ����� �������� �������� �������� ���������� ���������� ���������� ����� ����� � � ������ ����� ����� �������� �������� Due to the profound dissimilarities between RISC-societies and their risky counterparts it seems worthwhile to continue this article with a long-term historical perspective which should shed new light on the intricate relations between rare events and societal evolution. Kasperson and Paul Slovic, 2003, Reason, 1994, Slovic, 2000 or Wisner et al. 1994. RISC-Processes and Societal Coevolution 75 FIGURe 2.6 The Relations between Risk-Societies and RISC-Processes ���� �������� ��� ��� � � � �� � � �� � � � � � � � � � TAble 2.4 Shared Domains and Differences between Risk-Societies and RISC-Processes Shared Domains RISC-Processes Aspects of Risk-Societies Independent of Risk- without RISC-Processes Societies Production processes and solar flares processes of the indivi- large-scale Innovations; sandpiles dualization of life Courses; Relations between pro- tectonic formations drifts towards duction processes and brain mechanisms scientific self- the environment; word-frequency reflexivity, etc. Size distribution of firms; distributions; indisvidualization of Income and wealth scientific quotations risks from Systemic distribution; complex networks (pre- domains to private Financial markets; ferential attachments), etc. households, individuals, etc. Migration and settlements, etc. 2.2 The Three Very Long-Term Stages in the Evolution of RISC-Societies As stated already, rare events with far-reaching societal consequences have been part and parcel of the evolution of modern societies from its earliest stages onwards. So far, the history of rare events has been considered as exogenous and as residing outside the driving (f )actors or the generative mechanisms for societal complexifications and differentiations. Earthquakes, like forest fires, 76 Karl H. Müller floods, droughts, famines or epidemics were seen as short and catastrophic episodes beyond the scope of endogenous socio-economic development and growth patterns. In sharp contrast, it is suggested here that rare events play a genuinely endogenous role. Moreover, the history of RISC-societies can be separated into three distinct stages where Stage I comprises, by and large, all human societies prior to the irreversible capitalist expansion in the 16th century, Stage II consists of the evolution of a package of societal RISC-mechanisms and processes and their global diffusion in the half millennium between the 1450s to the 1950s and Stage III is breaking its way from the 1950s onward as new forms of couplings and interplays between societal and natural RISC-processes emerge. Table 2.5 presents an overview of the three stages of RISC-societies. TAble 2.5 Three Historical Stages in the Diffusion of RISC-Processes Time Distribution of RISC-Processes Stage I [prior to 1450] Societies under the A large class of RISC-processes in the dominance of natural RISC-processes environment of societies A relatively small class of RISC-processes in the sphere of economic production and knowledge generation Stage II [1450–1950] Aside from the RISC-ensembles already in Societies under the dominance operation during the periods prior to 1450: of societal RISC-processes 1) A collection of RISC-processes in the spheres of economic production (2) A set of RISC-processes in the domain of knowledge generation (3) Secondary, tertiary, n-ary RISC-processes as a consequence of (1) and (2) (4) The self-organization of the interlinked RISC- ensembles from and across (1) to (3) Stage III [after 1950] The growing interdependencies and inter- Societies under the dominance connections between natural and societal of couplings between natural (external) RISC-processes and societal (internal) RISC-processes RISC-Processes and Societal Coevolution 77 The RISC-World before 1450 The RISC-worlds prior to 1450 were relatively smal , with a world population of approximately 50 million persons around 1000 B.C. to roughly 400 million people around 1450. For centuries, the world population remained stable, with the exception of the period from 600 to 300 B.C. and around the decades of the year 1000. One of the most striking features prior to the long 16th century lies in the near absence of RISC-mechanisms inside the economic or knowledge production spheres of societies. FIGURe 2.7 Globally Distributed Societal Ensembles Prior to 1450 ������������ ����������� According to the stylized Figure 2.7, the socio-economic worlds prior to 1450 were composed of a large number of societal islands, separated by no man’s land or the seas between them and embedded in natural environments which were self-organized in a large class of external RISC-processes. In essence, during the very long first RISC-stage, the regionally separated states, empires or other societal ensembles of varying degrees of organizational complexity and embedded in different natural settings were affected and threatened by natural RISC-processes in the form of earthquakes, fires, famines, droughts, floods, epidemics, etc. Societal RISC-processes played a dominant role in the field of languages, and were possibly involved in migration patterns within empires, in selected cases of societal wealth distribution, etc. Inventions, innovations and 78 Karl H. Müller new technologies, although deeply engrained into the societal fabric, were not organized in a permanent and continuous way and remained largely outside the scope of societal RISC-processes. Following Karl Polanyi [1978, 1979], these different isolated and globally distributed societal ensembles can be classified as reciprocal or as re-distributive. Re-distributive formations exhibit a higher degree of organizational complexity, are clearly dominated by the political system and are, therefore, regionally confined up to the level of empires. These more complex re-distributive societal formations are characterized by the dominance of agricultural production and by a small and limited number of trades, by a rigid stratification system where the economic production processes mainly in agriculture reflect this stratification system. Capitalist engines, as they will be introduced in the next section, were only present to a sub-critical degree. This state of independent reciprocal and re-distributive worlds has been captured, aside from Figure 2.7, also by Table 2.6 which exhibit a small number of empires and societal ensembles, all embedded in a natural environment and subject to the influence of natural RISC-processes. The second noteworthy feature in the first RISC-stage lies in the absence of RISC-processes in the domain of knowledge production. Analogous to Polanyi’s scheme, societal knowledge-bases prior to 1450 can be characterized either as distributed or as centralized. Centralized knowledge bases were dominated by special knowledge preserving groups usually also associated with a dominant religion and empirical knowledge remained, using a term from Michael Polanyi, largely implicit. [See also Table 2.7] TAble 2.6 Societal Ensembles Prior to 1450 Societal Ensembles Reciprocal Redistributive Formations Formations Societies under Societies under the the dominance of dominance of the personal political System exchanges Quite obviously, rare events like strong earthquakes, catastrophic eruptions of volcanoes, long-enduring famines, wide-spread epidemics left a deep imprint in the collective memory of societies and it would be a fascinating research task, to link these rare events prior to 1450 to traditional knowledge systems, especially to religious and magic practices which should safeguard these societal ensembles from the occurrence of these rare events. RISC-Processes and Societal Coevolution 79 TAble 2.7 Knowledge Production Systems prior to 1450 Varieties of Knowledge Productions Distributed Knowledge Centralized Knowledge Bases Bases Knowledge bases Knowledge bases shared among under the dominance the members of a of a knowledge societal ensemble preserving group Two Great RISC-Transformations and Their Global Diffusion 1450–1950 At a surface-level it is widely recognized, following Immanuel Wallerstein,19 Fernand Braudel,20 and others,21 that the five centuries between 1450 and 1950 experienced a gradual integration of reciprocal as well as re-distributive societal formations into a relatively small global production system which was composed, initially, of a core region in the north-western parts of Europe (France, England,. The Netherlands), of semiperipheral areas (Spain, Portugal, Eastern Europe, the Italian peninsula, etc.) and of small bands of peripheral regions along the coasts of Africa, of Latin America or the East Indies. This global production system was fueled by a capitalist engine and was bound or linked together during its first centuries by a more and more differentiated system of trade relations. This global production system had placed itself outside the realm of effective political controls through regional political systems in Europe like the one in Spain, in France, in the Netherlands or in England. Subsequently specific development patterns emerged in each of the three global regions, reaching from differences in world trade-relations to significantly different roles and control capacities of national governments or to different compositions with respect to socioeconomic groups or classes. 19 On Immanuel Wallerstein’s world systems perspective, see especially Wallerstein, 1974, 1979, 1980, 1984 or 1989. 20 With respect to Fernand Braudel, see especially his three volume compilation on Civilization and Capitalism with a time span from the 15th to the 18th century in Braudel, 1979, 1982 and 1986. 21 From Karl Marx, 1964 to Max Weber, 1949, 1951 or 1978 one finds a wide-spread consensus on the uniqueness of the capitalist transformation. Relatively seldom, however, one finds an emphasis of a dual movement in economy and science. [Müller, 1999] To view both production processes within the economic sphere and within science under a homogeneous RISC-perspective constitutes a novel feature of the present approach. 80 Karl H. Müller − Core regions were characterized by an advanced economic production system and by a powerful state apparatus which shielded, supported and stimulated the national system of production both internally and externally. − Semiperipheral areas occupied an intermediary position both in the sphere of production and, especially important, in trade relations. − Peripheral domains were specializing in the role of agricultural producers and providers of raw materials. The state apparatus was comparatively weak and acted in a core-oriented manner. Likewise, trade relations with the center were highly asymmetrical. Over the centuries, one can observe the gradual emergence of global instruments like the gold-standard for coordinating and balancing the evolving world economic system. By 1930, towards the end of Stage II, weak global institutions and organizations like the League of Nations emerged which attempted to fulfill essential co-ordinating functions, too. Thus, the First Great Transformation can be characterized as the world-historic turn towards an economic RISC-mechanism as network of networks of self-organizing and globally distributed markets. Following Karl Polanyi, the societal consequences were twofold. On the one hand, “the economy is no longer embedded in social relations, but the social relations are embedded into the economic system” [Polanyi 1978:88].22 On the other hand, during the half millennium between 1450 and 1950 all hitherto external societal formations were gradually integrated into this expanding network of networks which had conquered every continent and every country within the five continents. As a consequence, reciprocal and redistributive societal formations became market or, alternatively, capitalist formations. [See also Table 2.8] TAble 2.8 The Great Transformation I: Local Reciprocal or Redistributive Formations Become Integrated in a Single Global RISC-Mechanism The First Great Transformation: Economic RISC-Production Reciprocal Redistributive → Capitalist Formations Formations → Formations Societies under Societies under → Societies under dominance of dominance of the → dominance of personal political system → markets exchanges → 22 Translation by Karl H. Müller from Polanyi, 1978. RISC-Processes and Societal Coevolution 81 FIGURe 2.8 The Irreversible Expansion of the Capitalist Engine ����������� The second Great Transformation is much less recognized and occurred in the domain of knowledge production. The half millennium between 1450 and 1950 saw a gradual integration of traditional forms of knowledge production and of societal knowledge bases which could be characterized either as distributed or as centralized knowledge bases. Thus, between 1450 and 1950 a globally distributed science-driven knowledge base emerged [See also Table 2.9]. Between 1450 and 1950, a global differentiation can be observed which brought about a separation into three distinct regional types of knowledge bases with respect to the (re) production and to the accessibilities of knowledge: central, semi-peripheral and peripheral knowledge-bases, all of them, of course, science-driven. − Core knowledge bases turn out to be highly distribution-oriented, setting the standards of the state of the art within specific fields of inquiry elsewhere, too. Judged from an intellectual balance of international exchanges, the core knowledge base is diffusion driven, exhibiting a global diffusion potential but being highly selective, in turn, with respect to knowledge bases and contributions from other regions. In terms of operationalizations and measurements for the present time, SCI-groups (science citation indicators) must exhibit a clearly asymmetrical pattern in which articles from core knowledge producing regions quote mainly other core region publications while, in turn, they are being quoted throughout the semiperipheral or peripheral knowledge bases. 82 Karl H. Müller − For semiperipheral knowledge bases, a genuine mixture between core features and peripheral features can be recorded, since semiperipheral knowledge bases show areas of high global competence with a correspondingly high diffusion potential as well as research fields with predominantly reception-centered features only. − Peripheral knowledge bases are mainly reception driven, exemplifying a high reception potential but being only marginally reproduced and recombined in other regions. Once again seen from an intellectual balance of international exchanges, the peripheral knowledge base is characterized by a local diffusion potential only, although it is able, albeit with a certain time lag, to reproduce the state of the art-standards set in core or semiperipheral knowledge bases. Again, peripheral knowledge production is highly asymmetrical in terms of SCI-values, exhibiting comparatively low impact values for other regions of the world. TAble 2.9 The Great Transformation II: Distributed or Centralized Knowledge Bases Become Integrated in a Single Global Knowledge Base The Second Great Transformation: Knowledge Production Distributed Centralized → Science-Centered Knowledge Knowledge → Knowledge Bases Bases → Bases Knowledge bases Knowledge bases → Knowledge bases under shared among the under the dominance → the dominance of an open members of a of a knowledge → mode of knowledge societal ensemble preserving group → production (modern science) The most surprising feature of Table 2.9 lies in the fact that the essential spatio-temporal differentiations for the evolving economic RISC-mechanism can be applied to the evolving RISC-mechanism for knowledge production as well. Although some important differences prevail, the deep similarity in the evolutionary development patterns of knowledge bases and economic production formations remains unaffected. Thus, it is not only possible and heuristically fruitful, to differentiate between core, semiperipheral and peripheral knowledge bases, but it is also rewarding from a cognitive point of view, to study the coevolution of economic and knowledge production throughout the five centuries of Stage II. The scientific production, like the evolving capitalist mechanism, have always carried with them a strong tendency toward globalization, although globalization RISC-Processes and Societal Coevolution 83 is to be understood in the spatio-temporal contexts of the evolving world-economy only.23 Thus, despite the seemingly global discourses between scientific centers throughout the 18th century in Paris, London, Edinburgh, Berlin, the American East Coast or St. Petersburg, many external territories and their knowledge traditions, especially in Africa, have not only been excluded, but also de-qualified and mis-understood in a very profound manner.24 The Third Stage of RISC-Societies after 1950 The decades between 1900 and 1950 have seen the emergence of three major shifts which justify to categorize the period after 1950 as the third stage in the very long-term evolution of RISC-societies. These three new components will be outlined in greater detail in the fifth section of this article. At this point, the three basic stages of RISC-societies as well as the two Great Transformations in economic and in knowledge production have been introduced. In the next section both transformations will be discussed in greater detail from a RISC-point of view. 2.3 The Two Great Transformations as Complex Network Mechanisms The present section deals with the incorporation of RISC-processes into the metabolisms of economic production and into the forms of global knowledge production of the 17th, 18th, 19th or 20th century. Within this section, four strong claims will be developed and supported with historical evidence. First, it will be argued that both for the sphere of economic production as well as for the domain of knowledge proliferation two self-organizing mechanisms or, alternatively, two self-similar network structures and dynamics can be specified which account for the spectacular growth in the area of economic and knowledge production. Second, the embedding of a capitalist and a science RISC-mechanism and their self-organized evolution has led to a swarm of interlinked RISC-processes in the economic and scientific spheres as well as in societal arenas. As a third thesis, the capitalist RISC-mechanism in its long-term unfoldings has generated and continues to generate a series of infra-structural networks which, 23 On this point, see especially Merton, 1985. 24 See, for example, Raynal and Diderot, 1988, Hegel, 1956. 84 Karl H. Müller over the last decades, became organized more and more as complex networks. Finally, from the 18th century onward, the capitalist and the science mechanism started to operate in a synchronous and, thus, coevolutionary manner. At the outset, Table 2.10 summarizes basic operations for the capitalist RISC-mechanism and the swarm of RISC-processes, associated with this capitalist engine. Of course, the term RISC-mechanism does not stand here for a hidden mechanical construct in the centre of production processes, but for a basic set of network structures and a for a growth dynamics of this network.25 TAble 2.10 The Great Capitalist Transformation and the Emergence of Socio-Economic RISC-Processes Operations of the RISC-Processes Capitalist Engine Average profits as basic Firm size distribution requirement for long-term reproduction Distribution of income and wealth Growth processes for micro-units Distribution of the frequencies of horizontally and vertically (production chains) changes in jobs Search for extra-profits Distribution of innovations New technology as a constant Distribution of technological accidents source of extra profits Rank-size distribution of birth-processes for new spatial ensembles for economic micro-units (firms) in a non-random production and distribution (cities, manner villages, etc.) Death-processes for established micro-units in a non-random manner Turning to the nodes of the capitalist RISC-mechanism, the micro-units of the capitalist engine are composed of firms, farms, associations and the like, capable of economic production or service provision and embedded in a monetary system of purchases and sales. The basic reproduction requirement for these micro-units is to reach an average profit-rate in the medium and in the long-run. Furthermore, these micro-units are capable of expanding horizontally, i.e. , regionally, or vertically, i.e. , by integrating pre- or a after production processes. The capitalist engine is energized by the search for extra-profits which can be accomplished in a variety of ways. Following Schumpeter, extra profits can be earned via technological change in the production of commodities already in use, the opening up of new markets or of new sources of supply, Taylorization of work, improved handling of material, the setting up of new business organizations such as department 25 On this point, see especially White, 2002. RISC-Processes and Societal Coevolution 85 stores—in short, any ‘doing things differently’ in the realm of economic life—these are instances of what we shall refer to as by the term innovation.26 Thus, innovations in their broadest possible variety constitute the incentives and attractors inherent in a global capitalist network. Again from a dynamic network perspective, the second basic feature of a capitalist network, aside from the search for extra profits or gainful innovations for short, lies in the creation of new actors with strong preferential attachments. New actors come into play preferably in those network domains or niches which are characterized by extra profits like in the case of the integration of new territories or in very large-scale technological innovations like in the diffusion of railroads or automobiles. Finally, network actors, failing the survival condition in the medium or in the long run, vanish from the network, although, in the long run, the birth rate of new micro-actors surpasses the death rate of outgoing micro-units to a significant degree. From the right hand side of Table 2.10 it becomes clear that the economic RISC-mechanism for the first Great Transformation is accompanied by a large number of socio-economic RISC-processes which became part and parcel of societal reproduction processes. Out of the total number of new societal RISC-processes a special emphasis will be devoted to Schumpeterian innovation processes since a permanent flow of innovations, due to their creative destruction and building capacities, turn out to be the most important single feature of the capitalist RISC-mechanism. As one of the new elements of the RISC-framework, Schumpeterian innovations are assumed to follow a power-law distribution and, thus, to fall under the class of RISC-processes. At first sight, this point may be surprising to economists, historians or sociologists of innovation processes. So far, socio-economic historians of innovations were rather convinced that the Industrial Revolution marked the first of a series of long-term cycles or long waves.27 Table 2.11 provides a customary overview of the succession of long innovation cycles from the 18th century onward.28 These long innovation waves usual y exerted significant effects on the overall economy. Table 2.12, for example, enables one to understand that before and after the railroad revolution the levels of societal transport, production and maintenance levels differed significantly. Around 1873, the world had become clearly different with regard to the transport dimensions than it had been thirty years earlier. Even 26 Formally, Schumpeter defines innovations as any change that does not alter the quantities of production factors of a production function, but that leads to a variation in the production function itself. 27 On long waves, see only Devezas, 2006, Kondratieff, 1928 or Schumpeter, 1961. 28 Table 2.11 is strikingly similar to the one found in Ayres, 2006. 86 Karl H. Müller though the question as to the proper dimensions and the adequacy of the historical data generally meets with great difficulties, the example of Germany between 1850 and 1873, around the time of the founding of the German Empire, provides substantial evidence that a big technological innovation wave must have taken place both in the overall performance dimensions of the transportation network and in its various other and more specific dimensions as wel . TAble 2.11 Long Innovation Waves as Rare Events of a RISC-Process Rare Innovations Long-Term Diffusion With Strong Consequences Peak Steam engines/ Textile industry 1788 1814 1848 Railway 1848 1873 1896 electrical industry 1896 1914 1945 Automobile 1945 1973 1996 ICT/Internet 1996 ? ? TAble 2.12 The Very Strong Effects of the New Railroad Transportation System in Germany, 1850–1873 Beginning of the 2nd End of the Industrial 2nd Industrial Revolution Revolution*) Network Dimensions Persons employed in transport 132.000 349.000 Value added railroads (mil. Mark) 17 bill. M. 274 mill. M. Value added transport (total) (mil. Mark) 53 bill. M. 387 bill. M. Freight traffic (in bill. Ton kilometers) 0,23 mill. TKM 9,9 bill. TKM Capital stock of the railroads 1,15 bill. M. 6,74 bill. M. (in bill. Mark) Dimension for the economy as a whole Capital stock in trade (in bill. Mark) 7,16 bill. M. 13,70 bill. M. *) Following Joseph A. Schumpter, each of the long cycles can be qualified as an industrial revolution Source: Hoffmann 1965 The repercussions of the railway construction can be quickly described in quantitative terms, using data for Germany. For iron and steel production, the RISC-Processes and Societal Coevolution 87 strength of the linkages was assumed to be 40% to 50% of the output, which, in turn, induced additional effects in iron ore and coal mining. Moreover, iron ore and coal were also linked directly to the railroad cluster between 5% and 10%. “The significance of railway construction as driving force in the process of industrialization … with a strong inducive effect cannot be called into question.” [Spree, 1977:288] It can also be shown from the perspective of capital mobilization that the railroads changed the financial world, figuring both as an obscure object of speculation and desired object of financing. If one takes into account the typical way capital was raised for railroads, mainly by way of the stock emissions as well as by way of the issuance bonds—as in the case of government financing—… the dominant role of railroads in the capital market becomes particularly evident in the 1840s to the 1860s. [Spree, 1977:266]29 This brief description shows that the construction and expansion of the railway system resulted in a fundament shift in the traditional means of transportation and actually had revolutionary effects for both the economy and society. These radical changes to the previous production and distribution modes as a result of rapidly growing transport capacities were achieved by a cluster of enterprises in the domains of private railroad companies proper, of machine construction, iron industry, coal production but also in the realm of banks and insurance companies. This core segment in the second industrial revolution succeeded within only a few decades—almost from nothing—in increasing the distribution capacities of the productive sectors within and, above al , between regional or national economies to significantly higher levels. In broad terms, one can agree with Walt W. Rostow, who emphasized that the construction of railroads lowered transport costs; brought new areas and supplies into national and international markets: helped in some areas to generate new export earnings which permitted the whole process of development to move ahead at a higher rate; stimulated expansion in output and the accelerated adoption of new technologies in the coal, iron and engineering industries; set up pressures (via the need for more durable rails) which helped give birth to the modern steel industry; altered and modernized the institutions of capital formation; and accelerated the pace of urbanization, with all its dynamic feedback effects on economic as well as social and political development.30 [Rostow, 1978:153] By substituting a RISC-perspective for the usual cyclical view, one is led, invariably, to assume that these long innovation waves constitute the peak in a distribution of innovations where a very large number of innovations has small and marginal impacts only. Table 2.13 provides a typology for four basic innovation types which 29 Translation by K.H. Müller. 30 Rostow 1978:153. 88 Karl H. Müller separates innovations into different diffusion potentials (low/high), different outputs (product/process) and, thus, into different innovation classes. TAble 2.13 Types of Innovations Type of Innovation Product Process High Type I Type II Diffusion (long waves) Potential low Type III Type IV Thus, the emphasis of long waves research should shift from the few rare events to a spectrum of innovation types and their overall distributions. Here, the expanded typology for innovations in general places the Kondratieff technology waves or the Schumpeterian industrial revolutions in the broader context of four possible innovation types. If one differentiates—as in Table 2.13—between innovations according to their output in product and process innovations, as well as differentiating by high and low diffusion potential in innovations, then one reaches a configuration with a total of four different types of innovation, of which only a single group, i.e. the one at the interface of product innovation/ high diffusion potential, corresponds to the Kondratieff or Schumpeter waves. Thus, large-scale innovations are not primarily arranged as long waves, reserved for a very few industrial revolutions, but innovations are arranged in a power law distribution, composed of a very large number of marginal or incremental innovations, of a limited number of medium-sized innovations and of a very small number of innovations with a very high diffusion level. These very few very large-scale innovations constitute the rare events or incidents with strong economic and societal consequences. Figure 2.9A reproduces fluctuations in a typical sequence of smal , medium-sized and large innovations, which, at first glance, appear as three large innovation waves. Figure 2.9B, by contrast, shows that there is a highly structured power distribution behind or underlying the three long cycles from 2.9A. Power-law distributions, as shown in Figure 2.9B, can be either very steep or rather flat, depending on the exponential coefficient in equation (1). For instance, for g = 1, the slope consists of a 45o gradient. P (k) ~ k -γ (1) In the context of Figure 2.9B, the long waves or the seemingly cyclical patterns of big surges in technological development and in large-scale innovations are transformed into a highly structured configuration with a very small number of RISC-Processes and Societal Coevolution 89 innovations with very large-scale effects—the classical long waves—and a very large number of small changes with marginal or small effects. According to the distribution in Figure 2.9B, the cycles from 2.9A only depict those rare peaks that are embedded in an immense number of small changes. FIGURe 2.9A Long Waves at First Sight ������ ���� FIGURe 2.9b A Power Law Distribution at Second Glance ��� ������ ������������� Yet Figure 2.9B does not only indicate a special distribution of innovations. These innovations, of different sizes and with different numbers, turn out to be generative with respect to their internal linkages. A very large-scale innovation enables, as is indicated by Figure 2.10, a number of medium-sized as well as 90 Karl H. Müller smaller innovations across economic domains. Many small innovations, in turn, lead to bottlenecks and shortages in other areas of the economy and in the infrastructural domains which increases the propensity for a new large-scale innovation and so on. Figure 2.10 illustrates this generative and productive pattern between the many smal , the less frequent medium-sized and the rare large-scale innovations. In this way, an important substitute and a considerably richer empirical basis can be provided for the traditional view of long waves, which does not focus solely on the large-scale innovations alias long waves, but on the overall context of many small innovations, a smaller number of medium-sized innovations and a few large-scale innovations. It is in this sense that the substitution of the traditional cyclical paradigm in favor of fluctuations or, more generally, a RISC-framework is to be understood. FIGURe 2.10 The Generative Relation between Small, Medium-Sized and Large-Scale Innovations �� � � �� � �� GI. large innovations MI: medium-sized innovations KI: small innovations An important aspect of substitution has to do with the question why and how such power law distributions occur, why innovations are distributed in such an asymmetric way and what general mechanism not only produces, but also reproduces this special configuration. Here, it is important to point out that networks with power law distributions are not only found in the economy but are distributed ubiquitously across society and nature, and range from economic domains like firm sizes or income and wealth, to other societal spheres such as migration and agglomeration processes, to the evolution of languages and their word distributions or to ecological systems with their forest fires or ecological disasters to geological formations— RISC-Processes and Societal Coevolution 91 earthquakes—and to many other phenomena of the natural world. In general, these processes are characterized by a high level of compositional complexity, by both global and local interactions, a relatively slow global dynamic and by critical thresholds and phase transitions.31 Turning to the engine for the second Great Transformation, Table 2.14 presents an overview of the basic operations for the science-based engine and the cluster of RISC-processes, linked with this knowledge-producing engine. Once again, the concept of an engine refers to a basic network structure and to a growth dynamics of this science-based network.32 TAble 2.14 The Great Knowledge Transformation and the Emergence of Science-Based RISC-Processes Operations of the Science-Based Engine RISC-Processes Average problem solving capacities Innovations in science as requirement for long-term reproduction locations of radical breakthroughs Growth processes for micro-units in science horizontally and vertically (program chains) Quotations of scientific articles Search for extra-cognitive gains Co-operation networks between New research programs as a constant scientists in a given field (Erdös source of extra cognitive gains number, etc.) birth-processes for new Reputation distribution across micro-units (institutes) in a non-random scientific domains manner Death-processes for established micro-units in a non-random manner With respect to the nodes of the science-based network, the micro-units of the science engine consist of research units in the form of labs, institutes, research centers, scientific schools or even individual scientists, capable of scientific production or service provision and embedded in a global system of scientific cooperations and publications. The basic reproduction requirement for these micro-units is to reach an average co-operation and publication rate in the medium and in the long-run. Furthermore, these micro-units are capable of expanding 31 Regarding individual types of models see also Bak, 1996, Barenblatt, 2003, Jensen, 1996, Laughlin, 2005, McComb, 2004 or Ong and Bhatt, 2001. 32 On this point, see especially White, 2002. 92 Karl H. Müller horizontally, i.e. , regionally, or across disciplinary boundaries, i.e. , by stretching out to other scientific disciplines. The science-based engine is energized by the search for radical breakthroughs in particular and for scientific innovations in general which can be accomplished in a variety of ways. These radical breakthroughs or scientific innovations can be accomplished, in a paraphrase to Joseph A. Schumpeter, via significant improvements in the available scientific research programs already in use, the opening up of new domains in the scientific landscape or of new sources of research infrastructures, advances in general or special methodologies, the setting up of new research organizations such as small trans-disciplinary research institutes—in short, any ‘doing things differently’ in the realm of scientific life—these are instances of what we shall refer to as by the term scientific innovation.33 Thus, scientific innovations in their broadest possible variety constitute the incentives and attractions inherent in the global science network. Again from a dynamic network perspective, the second basic dynamic feature of a science-based network, aside from the search for radical breakthroughs or, more generally, for gainful scientific innovations, lies in the creation of new actors with strong preferential attachments. New actors come into play preferably in those network domains or scientific niches which are characterized by a high innovation potential like in the case of the integration of new scientific domains of inquiry or in very large-scale scientific breakthroughs like the rapid diffusion of molecular biology in the 1950s or 1960s. Finally, scientific network actors, failing the survival condition in the medium or in the long run tend to vanish from the science network, although, in the long run, the birth rate of new micro-actors surpasses the death rate of outgoing micro-units significantly. From the right hand side of Table 2.14 one can see that the science-based engine for the second Great Transformation is accompanied by a large number of socio-cognitive RISC-processes, too, which, by now, belong to the normal scientific reproduction processes. As an empirical corollary, the extensive studies of Rogers and Ellen Jane Hol ingsworth on the locations of radical breakthroughs point to a clear power law distribution where a very small number of institutes is responsible for a very large number of these radical breakthroughs whereas most institutes are characterized by no radical breakthroughs at al .34 33 Formally, Schumpeter defines innovations as any change that does not alter the quantities of production factors of a production function, but that leads to a variation in the production function itself. 34 See, for example, Hol ingsworth and Hol ingsworth, 2000a and 2000b. 2.4 The Infrastructural Constitution of Modern Societies The discussion on the long-term evolution of societal RISC-processes offers a new perspective on the constitution of modern societies which places a special focus on their infrastructural domains and dimensions. Upon closer inspection, the rare events in technological innovations with a very large impact for societies occurred in special segments of the economic sphere which can be qualified as the infrastructural domain. These infrastructures, while part of the evolving economic production networks, occupy a special place since they offer and limit vital inputs and outputs for economic micro-units in three major areas, namely in the domain of energy, information and transport.35 Table 2.15 offers an interesting overview which connects the discussion on long waves with the three infrastructural segments of energy, information and transport. TAble 2.15 Rare Innovation Events and Their Infrastructural Domains and Their Generating Capacities Rare Innovations Long-Term Diffusion Infrastructural Infrastructural with Strong Peak Segment Networks Consequences Steam engines/ Textile industry 1788 1814 1848 energy No Networks Railway 1848 1873 1896 Transport Railroad network electrical industry 1896 1914 1945 energy Power grid Automobiles 1945 1973 1996 Transport Road network ICT/Internet 1996 ? ? Information Information net- Works (Internet, etc.) A rare innovation event with a very large impact across several dimensions of infrastructural capacities can also be visualized diagrammatically. Sticking to the second industrial railroad revolution, Figure 2.1136 creates a multi-dimensional space in the form of five performance dimensions for the three infrastructural domains of energy, information and transport as well as for the development of 35 One should add that infrastructures can be extended to other domains as well like water supplies or sewage systems. The focus on energy, information and transport has been made primarily because these three infrastructural areas are directly linked with the succession of industrial revolutions, discussed so far. 36 Figure 2.11 is based on Parallel Coordinates where each dimension is visualized by a single vertical line and the different time-trajectories from t0 to tn move horizontally through the different dimensions. On Parallel Coordinates, see especially Inselberg, 2009. 94 Karl H. Müller economy and society as a whole. The general dimensions can be measured by overal societal performance indicators such as gross national product per capita or by the capital stock in trade for the economic system. For the infrastructural transport segment the dimensions could include, for example, the transport of goods as a whole, the number of employees in the transportation network, the value added in railroads, the total freight traffic and the capital stock in railroads. Figure 2.11, which is vaguely based on the German data from Table 2.12, illustrates in general terms how between 1850 and 1873 in large parts of Europe as well as in the United States a very rare technological innovation with very large effects must have taken place within the infrastructural transportation domain. As can be seen from Figure 2.11, changes in the infrastructural transport domain lead to significant shifts both in the overall economic system and in societal spheres as well. FIGURe 2.11 The Scheme of a Rare Innovation with Very Large-Scale Consequences in the Infrastructural Transport Domain �������������������������� ������ ��������� ����������� ����� �� ���� �� ��� Over a period of two decades only, these changes led to a transformation of the overall societal production regime.37 With the descriptions so far and with the help of Figure 2.11 it becomes possible to summarize the different effects, generated by a rare innovation event. Rare innovation events in a single infrastructural network create a large number of primary and secondary effects and, equally importantly, also second-order 37 For instance, the dimension of network density must have also increased on a priori considerations, namely, through the increased number of actors (new companies), through a rising network density in railroads, through the linkages of the financial sector with the railroad system, the linkages of domestic markets and finally through the new connections in the export sector. RISC-Processes and Societal Coevolution 95 effects. More specifically, an industrial revolution within a single infrastructural network like the rapid diffusion of railroads induces effects − first, by revolutionizing an existing infrastructural domain itself (primary effects) − second, by inducing a bundle of changes in other infrastructural areas (secondary effects for the infrastructure domain) − third, by creating new nodes, linkages and linkage patterns for various economic arenas (secondary economic effects) − fourth, by the changes in domains outside the economic sphere, like the state, households, the science system, etc. (secondary societal effects) − fifth, by the changes within the induced changes of the infrastructural arena (second-order infrastructural effects) − sixth, by the changes of these induced infrastructural changes in the economic sphere both for the immediately effected areas and for other segments of the economic system (second-order economic effects) − seventh, by the influence of these induced impacts on societal arenas, etc. (second-order societal effects). To sum up, very rare technological innovation events from the 18th century onward can be described as a number of simultaneous, substantial and mutually reinforcing changes38 within performance dimensions and specific dimensions in one of the three infrastructural domains of energy, information and transport. More specifically, such a rare innovation event occurs − within the diffusion period39 of a new infrastructural regime either in energy, information or transport, which brings about − significant and strong shifts in performance dimensions of the particular domain − with strong repercussions or resonances [Niklas Luhmann] in other infrastructural arenas and in the economy and society as a whole. This concludes the infrastructural dimensions of rare innovation events with very large-scale economic and societal effects. Another striking feature becomes apparent in Table 2.15, however, as well. With the exception of the first industrial revolution which produced a significant rise in the energy sector for important parts of the economy, most notably for the cotton and textile industry, these rare innovation events with very large-scale consequences were accompanied by the emergence of new economic networks. Table 2.16 introduces—following the dimensions of Table 2.13 on innovation 38 See on this also point the account of Solomou, 1990. 39 Traditionally, one would write about the upswing phase of a so-called long innovation wave. 96 Karl H. Müller types—various network types, which are separated, on the one hand, according to the number of network actors into smal -scale and large-scale networks and, on the other hand, according to their form of connections, into flow networks and into relational networks. − Flow networks have, as their differentia specifica, physical connections between their network actors or nodes and manifest themselves, inter alia, as railroad networks, power grids, water networks, road networks, the Internet, high-frequency networks for mobile telephony, etc. − Relational networks comprise attributes or operations of network actors or nodes and do not involve physical connection lines. Relational networks can be based on kinship, on friendship, on acquaintance, on cooperation, etc. What is absent in all these instances is a direct physical link between actors like in the case of a railroad-network, the power grid or a telephone network. Furthermore, it becomes interesting to focus on the topology of these infrastructural flow networks.40 As for the topology, networks—flow networks as well as relational networks—can be differentiated into two different architectures, namely, into so-called random or egalitarian networks and into complex networks, scale-free or aristocratic networks. − Random networks, as depicted in Figure 2.12, evolve in a configuration in which the emergence of new nodes or links is not dependent on the pre-history of the overall topology. To put it differently, one does not find traces of memory effects. A typical random network, like the street connections of a region, links all relevant nodes. Here, the central goal is to reach all other nodes from one specific node with as little loss of distance as possible. − Scale-free networks, as shown in Figure 2.13, emerge where the creation of new connections does not take place randomly, but are formed based on a non-random rule, giving higher probability to nodes with a high number of links already. Global air traffic—but also the Internet—exhibit such a complex architecture in which new connections tend to orient themselves to the strongest, i.e. , most connected nodes. From a constitutional perspective, the new focus on infrastructure networks in the domain of energy, information and transport is so fascinating because these three areas—transport (or more generally: transport of material), energy and information—are, at the same time, the three basic ingredients in the constitution of the natural and the social world. What cannot be linked 40 For an overview see Barabási, 2000, Buchanan, 2002, Newman, 2005, Newman/Barabási/ Watts, 2006, Sornette, 2003, 2006, Watts, 1999, 2003 and 2004. RISC-Processes and Societal Coevolution 97 to matter, energy or information41 can hardly be imagined as being part of a natural or social ensemble. More importantly, energy, transport and information constitute a substructure that enables and maintains other societal domains as superstructures.42 Because of their central importance, these three infrastructural domains define and restrict the societal developmental capacities not only with respect to transport, energy and information, but also with respect to other capacities, like production, distribution or communication. FIGURe 2.12 A Scheme for a Random Network FIGURe 2.13 A Scale-Free Network 41 See, for instance, Norbert Wiener’s dictum that information is information, not matter and not energy, or Horst Völz’s definition of information as everything that is not matter or energy. [Völz, 1994]. 42 One might add water and air as necessary ingredients for societal infrastructures and metabolisms. But while air has been and still remains a free and public good, in the long run, water infrastructures, while being modernized continuously, have not had the potential for large-scale societal innovations during the last three centuries. 98 Karl H. Müller Several characteristics can be associated with each of these three infrastructural networks and their long-term developments. First, these three infrastructural networks, in the course of their evolution, have become organized as globally distributed networks across the territories and regions of contemporary societies. In this sense, one can speak today of a global societal transportation system, a global information system or a global energy system, all of which extend to every relevant place in a particular nation, region or settlement. Second, these three infrastructural networks have become intrinsic components of the economic world system and of practically all major societal configurations outside the economic arena, like the national systems of education, of the national systems of health or of the political-administrative systems. Thus, an economic system becomes embedded in infrastructural networks and other societal systems turn out to be embedded in precisely these three infrastructural networks as well. Third, each of the three infrastructural networks can be observed by means of one or several performance indicators, which measure the overall capacity levels in the domains of information, energy or transport and by means of a large number of ensemble-specific indicators, which record particular levels or flows of an infrastructural ensemble. Performance indicators, such as the per capita-consumption of energy, capacities for information flows or transport capacities, serve as overall parameters for these three infrastructural ensembles. Generally, one can assume that each of the three infrastructural networks can be measured with at least one, but usually several performance indicators and a comparatively large number of network-specific indicators. In sum, each of these indicators constitutes one dimension of a given infrastructural network, be it transport, energy or information. Fourth, the observations and measurements at the level of performance dimensions for these three infrastructural networks can be assumed to be evolutionarily stable. No matter how societies in the distant past, in recent history, in the present or in the future evolve, they can continue to be observed and compared on the basis of these performance dimensions. Evolutionarily stable dimensions have the advantage that they can be applied to almost any interval of societal time scale or time horizon. Over very long stretches of time, the distribution of values may change within these performance dimensions, but the dimensions themselves do not change. Fifth, these three infrastructural networks also constrain the societal potential for self-organization and become, thus, critical parameters for dimensions such as degrees of societal complexity, of order or other structural configurations. For example, societal self-organization depends on the amount of transferable RISC-Processes and Societal Coevolution 99 information and transport speed. A small message that needs several weeks to be transported restricts the domain of feasible forms of self-organization in the same way as a globally distributed instant access to very large quantities of information generates entire clusters of new forms of societal self-organization. Sixth, these different dimensions constitute, naturally and formally, specific spaces. Since the observations or measurements of infrastructural networks can be carried out in these multi-dimensional spaces across time, the three infrastructural networks can be characterized by specific dynamics as well. Thus, regions, cities or nations change within these multi-dimensional spaces at different speeds, depending on the underlying capacities and dynamics of their infrastructural networks. Seventh, in the long run, these infrastructural networks turn out to be accumulative. Figure 2.14 shows that from the mid-19th century to the present day several such large-scale flow-networks have evolved cumulatively, and at present can be found in an accumulation of co-existing and interactively linked large-scale infrastructural networks.43 FIGURe 2.14 The Accumulative Evolution of Five Infrastructural Networks—Railroads, Electrification, Roads, Airports and the Internet between 1845 and 2045 ���� ���� ���� ���� ���� What is interesting here is not just the cumulative character—at present all large-scale flow-networks are operating simultaneously. The reciprocal linkages, too, between these infrastructural networks in the three domains of energy, information and transport deserve special notice. In a certain sense, 43 For clarification I would like to point out again that in the last two centuries several medium-sized and smaller flow-networks have also emerged, e.g. , networks for water, telegraphy or urban trolleys, which have played an important role in the third infrastructural revolution, which Joseph A. Schumpeter described as the “electric revolution.” 100 Karl H. Müller these networks mutually generate and stabilize each other. For their operation, railroads require a functioning power network, which in turn is based on a novel information network, which, in turn, requires a power network and so on. The accumulation of infrastructural networks in the course of modernity puts a new flavor to Manuel Castell’s notion of the “rise of the network society”44 although these infrastructural networks, aside from the current ICT-revolution, play a very minor role Castell’s oeuvre. Eighth, these infrastructural networks—irrespective of whether they are random networks or scale-free networks—exhibit an interesting life cycle dynamic because, at an early stage, these networks are to be found practically nowhere. Railroads were a typical non-issue around 1760 as much as electric power in 1830 or automobiles and airplanes in 1875. These networks start as marginal forms in a new energy, information or transport niche, increase considerably in compositional or functional complexity45 and reach their full diffusion potential in the context of one of the three infrastructural domains at a later period, only to lose their dominant position within a few decades. Ninth, another interesting point lies in the sequence of these infrastructural revolutions. A specific rare innovation event in one of the three infrastructural networks of energy, information or transport leads to an asymmetry with respect to the two other networks. Historically, it becomes relevant that the sequence of such infrastructural revolutions changes the probability for subsequent rare innovation events. Relations (2) and (3) demonstrate that a rare innovation event RIE does not take place sequentially within the same infrastructural domain (IN ) & (IN ) at two consecutive points in time. Rather, a rare innovation event i,t i,t’ in one infrastructural network results, almost by necessity, in a subsequent, new rare innovation event in another infrastructural ensemble (IN ). j, t’ RIE (IN ) : < P(IN ) (2) i,t i, t’ RIE (IN ) : > P (IN ) (3) i,t j,t’ After this intensive discussion of infrastructure domains, of flow, random or scale-free networks and of rare innovation events which are concentrated exclusively in these infrastructure domains, it becomes possible to develop a new architecture for modern societies. Figure 2.15 exhibits a macro-pattern of vertical and horizontal linkages that run counter to the conventional wisdom of the constitution of modern societies, especially in the modernization or 44 See, for example, Castells, 1996, 1997, 1998 or 2000. 45 Compositional complexity refers—according to Rescher 1998—either to the number of modules or elements in a configuration or to the number of different types of such modules or elements. Functional complexity refers to the variety of modes of operation. RISC-Processes and Societal Coevolution 101 neo-modernization tradition. In Figure 2.15, the economic ensembles as well as various societal systems all are horizontally linked with one another and vertically connected to a societal sub-structure which comprises infrastructural networks for energy, information and transport. ��������� ������������� FIGURe 2.15 The Infrastructural Macro-Constitution of Modern Societies �������� ��������� �������� ������������������ �������� ����������� ������������� ������� ����������������� ������ ����������� ��������� FIGURe 2.16 The Infrastructural Network Micro-Constitution of Modern Societies ��������� ������������� ���������� ����������������� �������� ��������� ��������������� 102 Karl H. Müller In this sense, Figure 2.15 shifts the emphasis from the economic sphere to infrastructural network as the core of societal sub-structures. Figure 2.16 exhibits the micro-infrastructural constitution of modern societies which in its overall structures reproduces the macro-ensemble of Figure 2.15 and which contains infrastructural networks for energy, information and transport as its basic sub-structure. Figures 2.17 and 2.18 summarize the different effects of the rare innovation events in infrastructural networks. Figure 2.17 presents these primary and secondary effects and Figure 2.18 exhibits the second-order effects and adaptations, induced by these primary and secondary changes. In sum, rare innovation events in infrastructural networks imply four large groups or clusters of restructurings or reconfigurations. FIGURe 2.17 Primary and Secondary Effects of a Rare Innovation Event in a Single Infrastructural Network (Transport) �������� ����������������� ��������� ��������������� First, the primary effects usually manifest themselves as a big jump both in relevant performance dimensions as well as in specific network dimensions. RISC-Processes and Societal Coevolution 103 Second, a rare innovation within a specific infrastructural network opens up new horizons for innovations and can stimulate innumerable search processes for inventions, which can result in medium-sized or smaller innovations in other infrastructural networks, too. For instance, the expanding railroad network led to an end of the previous means of transport means for mails within the infrastructural information segment. In 1878 the last stagecoach closed its service in Germany and a series of innovations was needed to adapt the postal delivery system to railroads. By the same token, the packaging and delivery forms changed for many companies in response to the operations and requirements of railroad transport. FIGURe 2.18 Second-Order Effect of a Rare Innovation Event in a Single Infrastructural Network (Transport) ����������������� �������� ��������� ��������������� 104 Karl H. Müller Third, a big fluctuation in a specific infrastructure domain like transport leads to second-order changes that result from the ongoing diffusion of such an infrastructural revolution and its continued proliferation of primary and secondary changes. For instance, a European railway network also means that the transportation potential for commodities within and between countries can be significantly expanded in the long run. More export and import diversifications, in turn, open up new directions of specialization, comparative advantages and economies of scale, which are then reflected as second-order effects on small business or on the local production.46 Fourth, a significant element of diversity comes into play because a rare innovation event does not reproduce itself identically and does not occur at a uniform speed. From a global perspective, different types or families of development trajectories become possible, depending on the time-frame of a rare innovation event. Despite regional variations, however, a rare innovation event—for instance, the one that took place during the rapid expansion of the European or North American railroad networks—proves sufficiently homogenous throughout various regions, cities or nations that the same general directions prevail within the various dimensions of the infrastructural transport network. With reference to the revolution in railroads one will look in vain for regions in which the transport capacities have declined during the course of the building up of the railroad network. 2.5 The Third Stage in the Coevolution of RISC-Societies: The Recombinations between Societal and Natural RISC-Processes After the presentation of a new infrastructural perspective on the constitution of modern RISC-societies the present chapter will be concentrated on the current phase of RISC-societies and on the new configuration of natural and societal RISC-processes across regional, national or global levels. The Recombinations between Societal and Natural RISC-Processes Figure 2.19 presents, once again, the state of RISC-societies around 1950 which marks the end of the second stage and the beginning of the third stage. Due 46 On both the positive and the negative effects of railroad construction on processes such as the development of cities, of national companies and the destruction of local production cf. the classical studies by Al an R. Pred 1966, 1973 and 1980. RISC-Processes and Societal Coevolution 105 to the two Great Transformations and the two RISC-network mechanisms for economic and knowledge production, the RISC-societies worldwide have become a densely interlinked ensemble with a large quantity of societal RISC-processes and, initially in strict independence, a relatively stable amount of natural RISC-processes. Prior to the third stage in the evolution of RISC-societies rare events with a high damaging societal impact shared the following three characteristics. − First, these rare events, despite their usually strong effects, were local and affected relatively small parts of the world and left other regions virtually unaffected. − Second, rare events were strictly independent from each other. Earthquakes and technological accidents, for example, were in no way related to each other and occurred for strictly independent reasons and within different generative mechanisms. − Third, rare events, due to their local and strictly independent form of production, had rather weak effects for the continued development of entire nations or for the global survival in general. FIGURe 2.19 The Global Configuration at the End of Stage II ����������� Within a few decades, however, the third stage in the evolution of RISC-societies has led to an entirely new configuration and to a recombination of natural and societal RISC-processes. Currently, all three characteristic features, associated 106 Karl H. Müller with rare events during the second stage, have changed dramatically. In recent decades, these rare events have been undergoing a triple transformation process which, as will be shown below, has spectacular consequences for current and future problems of regional, national or global stability or sustainability, for that matter. The first novel feature lies in a new phase of globalization which differs significantly from the global diffusion processes of the economic and of the knowledge engine of the second stage. The decades after 1950 produced a broad range of new globalization processes through the emergence of transnational enterprises and global institutions like the World Bank, the IMF and the United Nations, and, more importantly, through new global infrastructural networks for transport, i.e. , the global airline network, and, especially important, through new ICT-based networks. In terms of RISC-processes, globalization has lifted societal RISC-processes to global dimensions as well. − The global air-traffic-system has been organized as a global scale-free network with a small number of central nodes, i.e. , airports with large a number of global destinations, and a very large number of small nodes, i.e. airports with connecting flights to the large nodes. Complex networks possess a special vulnerability pattern since the elimination of two or three central nodes can lead to the col apse of the entire air traffic-network. − The world-wide web is organized as a global scale-free network in a number of ways. Most notably, the global information network itself is characterized by a distribution of a small number of central nodes, a large number of small connectivity nodes and the same vulnerability pattern as for the global air traffic-network applies. − Computer viruses, due to the architecture of the world wide web, have the potential of affecting a huge number of businesses, state-administrations and private households practically simultaneously. − The combination of global information and communication technologies and financial markets has produced a global financial system which, so far, has been growing exponentially and outside the control of global, supra-national or national RISC-protection networks or systems. Moreover, due to the ICT-support financial markets have advanced to a stage of automatic trading where the number of actors are considerably enlarged through automatic algorithmic traders with a vastly increased operation speed. Currently, this new financial system experiences the rare event of a global crash which spreads throughout the global economic sphere and which affects national conomies, regions or communities around the world in an unprecedented scale within the time-frame of the third stage of RISC-societies. RISC-Processes and Societal Coevolution 107 Aside from the globalization of economic, scientific or societal RISC-processes, one is confronted with a second fairly recent development which, however, becomes characteristic for the third stage in the evolution of RISC-societies as well. Economic development processes in goods and services, housing, transportation, energy, etc. in Asia, Latin America and Africa have passed, in conjunction with on-going processes of post-industrial growth in the highly developed regions, a critical level in greenhouse gas emissions, most notably in CO . Figure 2.20 points to the fact that global societal development processes 2 have become critically related to the world climate and, as a consequence, to global climate change in the form of global warming.47 FIGURe 2.20 The Re-Combination of Societal RISC-Processes ����������� The third stage in the evolution of RISC-societies experiences, thus, a double movement of RISC-processes. − On the one hand, societal RISC-processes become rapidly global in scale. ICT-technologies, air-traffic systems, financial markets, but also wealth, income distributions or urban agglomerations have left their national or local boundaries and enter a new phase of global distributions. − On the other hand, the capitalist RISC-mechanism with its associated RISC-processes becomes critically linked with a bundle of natural RISC-mechanisms for the global climate. 47 For summaries, see, for example, Dessler and Parson, 2010 or Giddens, 2009. 108 Karl H. Müller While a steadily increasing amount of CO emissions is not a RISC-process by 2 itself, climate change has direct effects on the distribution of a large number of atmospheric and natural RISC-processes like floods, droughts, hurricanes, tornados, forest fires, and the like. This, in turn, leads to the third new critical feature of the current stage in the evolution of RISC-societies. The third critical feature stresses the fact that societal and natural RISC-processes begin to interact and become entwined and entangled in unprecedented and unintended ways. Climate change with its new distributions for RISC-processes can affect infrastructural networks like the power grid which, as a direct consequence, can lead to energy shortages and, as an obvious secondary effect, to the inability to sustain the internet-economy in an undisrupted and continuous manner. Figure 2.21 summarizes the new challenges for the infrastructural architecture of contemporary societies through the three new features of the third stage of societal RISC-evolution. FIGURe 2.21 The Constitution of Contemporary RISC-Societies �������������������� ������������� ���������� �������� �������� �������� ��������� ��������������� The Growing Risk-Potentials of Contemporary RISC-Societies The three new features of the third stage in the evolution of RISC-societies, i.e. , a new phase in globalization, critical links between the capitalist RISC-engine RISC-Processes and Societal Coevolution 109 and societal RISC-mechanisms on the one hand and global atmospheric RISC-mechanism on the other hand and, finally, new couplings and recombinations between natural and societal RISC-processes, have far-reaching implications for the future of RISC-societies at regional, national or global levels. Due to the three-fold transformation of natural and societal RISC-processes contemporary societies across the world are faced with an entirely new pattern of threats to societal maintenance, sustainability or viability.48 As a catchphrase, contemporary RISC-societies are at risk in an unprecedented manner. At the global level, the following mismatches and maladaptations can be observed, namely − a diffusion mismatch because more and more societal or natural RISC-processes can have non-local or even global effects − a governance mismatch because economic globalization, despite a fast growing segment of global NGOs, has not been accompanied by the buildup of global institutions49 − a mismatch in the protection networks or systems because of changing distribution patterns of atmospheric RISC-processes due to climate changes and global warming and unchanging RISC-protection and RISC-support networks or systems, operating on constant average conditions with smaller or larger deviations. For the national levels, new risk-potentials in the field of national security emerge which clearly transcend the usual profiles of military or terrorist attacks. Here, national security is understood in a comprehensive manner which includes military or terrorist issues alongside with economic, social, political, cultural or environmental dimensions. In this comprehensive context, national security is not simply synonymous with an absence of military or terrorist conflicts and violence, but depends on the well-functioning of the entire spectrum of societal and environmental networks and systems and their interactions in the face of the three new features of the third stage In the evolution of RISC-societies. Thus, countries like Slovenia are confronted with a challenging new configuration which, inter alia, includes the following characteristics: − an increased domestic vulnerability, due to a changing pattern of natural RISC-processes in Slovenia which are coupled with the economic RISC-mechanism in the form of global warming and climate change − an increased vulnerability from outside, due to internal or external rare events in areas outside Slovenia, but with a high impact for Slovenia as well − new requirements for regulation and prevention-policies not only in 48 On the differences between sustainability and viability, see the introductory article to Part I. 49 See the paper by Stark, 2009 which has been published as RISC-Research Paper Nr. 5. 110 Karl H. Müller traditional areas like building and construction, but also in financial markets, in the three infrastructural networks, in web-based technologies and in those environmental domains affected significantly from climate changes and from a different distribution of natural RISC-processes − new capacities for damage control and for support networks which in the case of rare events are capable to operate with strongly improved soft skills and with locally adequate and accepted interventions. In a separate article50 different types of control challenges for contemporary RISC-societies will be outlined in greater detail. 2.6 Re-Dimensioning Sustainability: RISC-Robustness as a Third Dimension of Sustainability The discussion so far on the evolution of RISC-societies opens up new perspectives for the concept of sustainability as well. The overall challenges ahead can be captured by Figure 2.22 which exhibits a new and third sustainability dimension alongside the two previously known ones. So far, sustainability has been discussed broadly in the following two dimensions: − first, sustainability along a spatial dimension, i.e. the globalization of today’s high socio-economic development levels to the entire globe and − second, sustainability along a temporal dimension, i.e. the inclusion of future generations at today’s high socio-economic development levels. But sustainability has to be enlarged with a third RISC-dimension, namely − third, sustainability along the RISC-dimension, i.e. , robust, or, alternatively, resilient51 designs for the entire set of RISC-ensembles and for their environments across regional, national or global levels. Figure 2.22 points to the new three-dimensional space for sustainability where the arrow indicates a drift towards higher levels in each of the three dimensions simultaneously. Currently, very little is known whether an arrow as depicted in Figure 2.22 is feasible at all or whether substantial increases of sustainability along all three main sustainability dimensions simultaneously cannot be accomplished under the given constraints, barriers and empirically observable distributions and reconfigurations of societal and natural RISC-processes at present. 50 See Karl H. Müller “RISC-Processes and Their Control Potentials” in this volume. 51 On the two equivalent notions of robustness or resilience, see Gunderson/Hol ing, 2002 or Gunderson/Allen/Hol ing, 2010. RISC-Processes and Societal Coevolution 111 In this way, a new and richer perspective on sustainability can be gained which moves sustainability problems into a vastly unexplored three-dimensional configuration of time, space and RISC-robustness. FIGURe 2.22 The Three Principal Components of Sustainability ������������� ���������� ������������� 2.7 Three Big Challenges for Future RISC-Research Due to the globalization and the partial recombinations of natural and societal RISC-processes it should become obvious why RISC-research should provide the missing links for a theory of societal evolution. The building blocks and the theories for a dynamic and evolutionary perspective of societal evolution, so far, lacked three crucial elements. − The first missing component was a comprehensive and structured arrangement of both internal and external RISC-mechanisms and their production of RISC-processes. Past or contemporary societies are not experiencing rare events as a random sequence, but as a highly complex and coupled arrangement of RISC-mechanisms inside and outside of society. − The second missing link consisted of appropriate accounts of actions or operations at the micro-levels which so far were dominated by schemes like Rational Choice or weaker variants like bounded rationality. Here, the RISC-framework operates with non-trivial models of embedded cognition which rely and require a rich observation basis of actions observed, of the internal stances of actors of themselves as well as assessments by others. 112 Karl H. Müller − The third missing building block is a result from the consequences of the previous two missing components. Societal learning processes as a consequence of rare events can be described and analyzed in a far more comprehensive and more trans-disciplinary manner since these societal learning processes take place simultaneously across a wide variety of societal domains. The present overview of RISC-processes and their generative mechanisms can be concluded with a short summary of necessary steps and research tasks for the immediate or medium-term future. Quite obviously, the discussion so far on the long-term evolution of RISC-societies has led to a long list of extremely demanding research challenges and to a wide variety of potential RISC-applications. Towards the end of this article three of these major survival-critical research challenges and their appropriate RISC-applications will be outlined briefly. − First, there is a highly demanding challenge for coupled RISC-modeling. The new stage in the evolution of RISC-societies around the globe requires intensive explorations of model groups with positive mutual feedbacks and the general format of: M = F (M , M ) 1 2 1 M = F (M , M ) 2 1 2 These coupled RISC-models should be capable to specify the potential impacts for societal and environmental domains in a more comprehensive way than isolated and independent models. − The second research challenge has to do with governance issues and new forms of institution building. The two most relevant governance challenges are directly linked with the ongoing globalization processes. The first very big problem has been mentioned already and lies in the asymmetry between the increasing set of global actors in the economic RISC-mechanism, the lack of actors of similar strength and scope at the global level within the RTISC-protection networks or systems and the need for a rapid catching up-process of global actors in the RISC-protection networks or systems. The second very large-scale problem lies in the opposite direction of globalizing actors, namely in a de-globalization of the impacts of rare events. In this respect the construction of institutional firewalls and institutional barriers becomes of utmost importance in order to contain rare events in relatively small areas only and to prevent a rapid diffusion to global levels. Such an institutional containment of rare events is urgently needed in areas like computer viruses, the power grids or financial bubbles, to name only a few prominent instances. − Finally, the third research challenge lies in the utilization of transactional or electronic data in the study, in the prevention or in the rescue operations RISC-Processes and Societal Coevolution 113 of RISC-processes. The diffusion patterns of epidemics can be reproduced by studying Google-search operations, the contagious attractors inherent in financial markets become more transparent by the micro-data for selling and buying, or informal support networks can be created or activated in the case of disasters through recent Web 2-tools like twitter, facebook and the like. In this area, new observation devices and tools become available which must be integrated and utilized within the RISC-research tradition. This small list of very big research challenges completes, in combination with the introductory RISC-primer, the overview on the potentials and the application domains for RISC-research, present and future. Part II — RISC Modeling and RISC Theory 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 13 16 15 Introduction to Part II Part II summarizes contributions which are aimed at a better understanding of the theoretical and modeling backgrounds of RISC-processes across contemporary societies. Part II can be subdivided into three rather homogeneous segments. The first section comprises the two articles by Heinz von Foerster and by Monika Gisler and Didier Sornette. They provide a general qualitative discussion for an in-depth understanding of RISC-processes and their generative mechanisms. Heinz von Foerster discusses Zipf ’s law in the context of the 9th Macy Conference and he stresses the importance of self-similar or scale-invariant distributions. Gisler and Sornette advance the social bubble hypothesis as a general generative mechanism for societal RISC-processes. The social bubble hypothesis stresses network configurations with positive feedbacks unhampered by countervailing factors as the decisive element for the emergence of rare events across science, technology, financial markets, and one could add, politics, fashions, etc. Turning to the second section four articles have been compiled on the modeling aspects of RISC-processes. Günter Haag proposes new models for generating power law distributions and suggests new micro-macro relations in this respect. Karl H. Müller discusses the wider issue of control ing RISC-processes by pointing, on the one hand, to a typology of RISC-control configurations and to a special theorem on the vulnerability of the emerging RISC-societies of the 21st century. Michael Schreiber, after giving an overview of labor markets as self-organizing systems, finds a new RISC-process in the field of labor market transitions. Finally, Günter Haag, Karl H. Müller and Stuart A. Umpleby offer a new type of self-reflexive RISC-model which draws considerably on the work by George Soros. Finally, the third section offers two interesting perspectives which should be seen as complementary to the previous two sections. Adrian Lucas points to a potentially relevant framework for RISC-mechanisms in financial markets which is heavily based on the work of Richard Buckminster Fuller. And Peter Štrukelj reminds us that our state of knowledge on RISC-processes in financial markets is not only very restricted with respect to forecasting and quantitative modeling capacities, but also with respect to ex-post qualitative explanations. Investigating current studies on the financial crisis by renowned economists in closer details, he comes to the conclusion that these accounts do not even qualify as explanation sketches and simply qualify as an expression of relative blindness even ex post. A Discussion on Zipf ’s Law 3 Heinz von Foerster et al. 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 13 16 15 Heinz von Foerster et al. : A Discussion on Zipf ’s Law1 [Titlepage 1] Zipf ’s Law. Macy Meeting 1952 [Titlepage 2] Ninth Conference on Cybernetics Josiah Macy, Jr. Foundation March 20–21, 1952 Hotel Beekman New York, N.Y. [Transcript] VON FOERSTER: I will try to make Zipf ’s law as short as possible. About two years ago, a professor in Harvard, Professor Zipf, published a book which was called Human Learning and the Principle of Least Effort. This book was an extensive survey of statistics which dealt with the frequency of occurrence of elements of various kinds which belonged to a particular class. I will give you immediately a concrete example. For instance, classes of things which were dealt with were words in a book or beetles in a backyard or occupations in the country, and the question was, how often did a particular occupation, for instance, occur in the country; for instance, how many barber shops can be found in the United States, or how many beauty parlors can be found in the United States, and so on and so forth; so that occupation would be a class, and then the kind of barbers or beauty shops, and so on and so forth. In the case of the words, I would say it would be the different words. He found the following situation: In counting, for instance, all the different words in a book, he found that there is one word which occurs most frequently. Let’s say in English it would be “the.” As a particular example, in James Joyce’s Ulysses, you have about twenty thousand “the’s.” This word is a word with a rank, No. 1, and we give it the first prize in the occurrence of words. The word with rank No. 2 will be another word—and I will put down here [on blackboard] the rank and here I wil put down the frequency. The second word—I have forgotten it, unfortunately, but it might be “a,” the general article. He found it had a 1 Slightly edited text following an unpublished typescript kept in the Heinz von Foerster archive at the Department of Contemporary History, University of Vienna, Sign. DO 968, 4–8–1. Words and signs in brackets indicate corrections or additions to the transcript. 122 Heinz von Foerster et al. frequency of 10,000, which is precisely half of the most frequent word. The word which ranked No. 3 as the third most frequent word would have a frequency of about 6,666, which would be a third of the maximum frequency. If you call the maximum frequency F, he found that the frequency of a word with the particular rank could be written down in a formula which is F/r. This does not hold only for words, but it holds also, for instance, in counting the number of different kinds of beetles in the backyard. It is also valid for counting the different occupations in the country. One can find out, for instance, that the most frequent occupation in the United States is actually the barber shop, the second frequent one is the beauty parlor, and the third frequent one is the laundry, and, also, they fol ow precisely this kind of statistics. This is certainly very peculiar, and lots of people have thought about how such statistics actually occur. Now, the usual way of plotting the results of this kind, of making statistics, is plotting rank logarithmically and the frequency in units, so that in this direction, the logarithm of frequency is plotted, and in this direction, the logarithm of rank is plotted, and as a result of the original equation I put on the blackboard, F/r, as a constant in this time, the maximum frequency F, you have a straight line in that logarithmic representation. This straight line cuts the frequency, so that 1 is about here, and it goes through the axis of the rank just at a point which indicates the total number of different kinds. With respect to words, it would be the total number of different words used in a book, or the total number of different species found in the backyard, or the total number of different occupations existing in the country. It can easily be found, for instance, if you put the frequency, F, equal to 1, then certainly r becomes equal to F, and since the highest rank must automatically, by the way in which we count the rank, be identical with the number of different kinds, that means that n, the number of different kinds, equals then the number of or the frequency of the most frequent word. That means that if you actually want to know how many different words are used in a book, the only thing you have to do is to count the number of times the word, “the,” occurs in that particular book and then you know immediately quite well how many different words occur in that book. That is certainly very startling and very peculiar. There is a possibility of making an indication why that counts, and it is as follows: If we make statistics at all of such a sample of the different kinds that can occur, where every kind is represented by lots of individuals, then what you have to look for is certainly a function which I would cal , rather, generally, P, which should be a function of the different kinds, which I would call n, and the possible rank. I would say this is just the thing that I select out of the whole thing. Our particular example, for instance, for actual type distribution is n/r, because n is identical with the maximum frequency and r gives the number of ranks. A Discussion on Zipf’s law 123 Now, let us forget for the moment what actually is found out, and let us consider a very general picture. Let us assume that I have here a universe in which lots of different kinds are floating, by representing a lot of different individuals within that kind. Let us make several triangles that would represent, for instance, “the’s” or barber shops or beauty parlors in the country and all those things, another kind, and so forth. Let us assume that such a universe exists. To know what actually occurs in that universe, we have to make a sample universe. Making a sample means collecting a particular region out of those things. Immediately, we can make statistics. For instance, in this particular sample, with three plus signs, one circle and one triangle, our rank frequency diagram would certainly rank 1 for the crosses, it would be frequency 3; rank 2 for the circles would also be frequency 1; and rank 3 for the triangles would also be frequency 1. Now, in this particular example, you can see that the definition of what we call a rank becomes a little bit arbitrary in the moment where several different kinds have the same frequency. But I will show you a little bit later that it is possible to avoid this arbitrariness and it is also possible to establish for any particular kind a real rank, a value. Now, I asked for this kind of universe the following things—and this is a particular assumption which is made, and I would like to call this particular assumption for lack of a better word, the generalization of the cosmological postulate, and what I mean by a cosmological postulate—which was the famous idea pointed out by Eddington, Millikan, Jennings, and so forth—is that if we look at the world, the world should, from all points in the universe, look alike; that is, wherever you put an observer in the universe he should see the same kind of a universe, the same distribution of stars or nebulae. Let us ask precisely the same thing for this kind of a universe; that is, that every observer should see precisely the same kind of distribution. Then we must ask for such a distribution free play, which actually does that job. In doing that, it means the following thing: Whether I go in my backyard or whether you go in your backyard or whether somebody else goes in his backyard, he should find a similar distribution. On the other hand, it should also be the case that if you increase your sample, if you make it a little bit larger, you should also see a similar distribution. Now, I have to state precisely what I mean by a similar distribution. By a similar distribution, I mean that the rank-frequency diagram should actually give the same kind of a function. For instance, we have a rank-frequency diagram of any particular form; we should expect, if we increase the sample, to find a similar form. That would be a generalization of the rank-frequency function. Now, increasing the sample could also refer to two different ways of doing it; namely, increasing the sample could mean going forward until I find the next element, and these elements could be two different things: either it could be 124 Heinz von Foerster et al. an element which I already have in my sample, or it could be a new element with a new species. Now, I define a proper increase of my sample until I reach a new species, and this for the very simple reason that if I increase my sample and I have a particular rank-frequency diagram already established, [of ] any kind, let’s say—let’s make it general—then it would mean only that I am going to disturb this kind of a rank-frequency diagram, which I would call a fluctuation, a particular fluctuation of that diagram. A proper increase means only if I go further and catch a new sample. In establishing this as a proper increase of a particular sample—and it should give exactly the same function of a rank because of the distribution—I have to say the following thing: The change of Φ in respect to n, with the number of different samples, should be a mere function of r, the rank alone. That means it should be independent of the number of different kinds I actually observe in that sample. Making this very specific definition, it can be integrated immediately and I get an integral which gives [the result]: Φ must be n times the functional r, plus, I would say, an integration constant which would be a function of r. These can be shown to be zero, because if I have no sample at al , I certainly must have a frequency function of zero. Φ must be zero; therefore, [c] turns out to be zero. The next thing we can establish immediately is the following— BIGELOW: Those two functions of r aren’t the same thing, are they? PITTS: Those two Fs are the same, are they? Yes, they obviously are, as a matter of fact. VON BONIN: You mean this F? PITTS: He is just integrating directly. VON FOERSTER: This is a partial derivative of Φ in respect to n. BIGELOW: F is a constant, then, with respect to n? VON FOERSTER: Yes, with respect to n. Now, I can immediately ask what will happen if I put r to n; that is if I make my rank the highest rank which exists. Then, certainly, by the definition that I would go and increase my sample so long until I catch a new sample or a new species, then this new kind can occur only once. That means a function of n, n, whether r is equal to n, must equal 1, because when this sample occurs just one time in the whole sample, by putting r to n, I get an equation of n to n must equal 1. That means that our function of F to n must equal 1 over n, for all n’s, and this is precisely the thing which has been proven; namely, that there is a general function which shows always 1 over r as a distribution function which I actually get in the sampling of those things. This is always the case, then, only if I am operating in a universe where the things are so distributed that wherever I start in that universe, and how large I make the sample, I get the same distribution. A Discussion on Zipf’s law 125 You see the restriction of this law immediately, because it is restricted at that point by already touching the limits of the universe, because then I really know I am on the edge. That is information, and it means that the whole rank and distribution would be completely changed. It would not be possible, therefore, by knowing the total number of beetles, kind of beetles, which we have on the earth, to infer about these numbers directly to the number of actual beetles living in the world, because then we have already touched the limits of the universe and we can’t make any such statement. In the ease of our mental patients, let’s say, they are living somewhat at the edge of the universe and therefore, in the universe language, they do not have those words more available and therefore you have distortions of the straight line occurring. Now, I must confess that I do not recall precisely. BIGELOW: Hold it a minute, will you? Within that first sample box, you contain a certain sample; right? A certain sample is contained there? VON FOERSTER: I collect a sample there. BIGELOW: The sample is finite? VON FOERSTER: Absolutely finite, yes. BIGELOW: Then there are some, by the law, which are not in the sample, because if the sample contains a finite number, if you keep finding those species which are rarer, you will finally find a species which is so rare that not one of them will be in this sample. However, if you expand the size of the sample, you will get this new species. VON FOERSTER: That is right. I get this new species and I say, by proper increase of the sample, it is only that increase which gives me that new very rare species. McCULLOCH: Let’s put it this way. BIGELOW: This thing implies that there are an infinite variety of species. VON FOERSTER: Right, and that is the ease with your sample in comparison to all possible species. McCULLOCH: Zipf ’ s law would not hold for the chemical atoms. BIGELOW: How can it hold for language? WIESNER: Does this mean, then, a man uses only a small portion of his total vocabulary? VON FOERSTER: That is right. McCULLOCH: Unless you deal with a lobotomized individual; then the peculiarity is that the lobotomized patient runs out of words for rare cases, Zipf ’s law curve comes down and is chopped off at the bottom. He has run out of kinds. VON FOERSTER: Precisely. It assumes that the universe is infinitely larger than compared to that which you are actually able to observe. That means you are sampling only a small section of the potentially available universe. 126 Heinz von Foerster et al. BIGELOW: How do we get this third out of this? VON FOERSTER: This third? BIGELOW: Yes. You were bringing them down, and you indicated that n was a certain constant. The rate at which they fall off, the most common number, is a definite constant in the first case you put on—your 20,000, 10,000, and 6,666. VON FOERSTER: Oh, yes. I mean the function, F/r, should be a function of the form 1/r. That means, if I have n different kinds which I actually observe, then my function must look equal to n/r; that is, if I am collecting a thousand samples or a thousand kinds, rather, of things, then my rank-frequency diagram must look like the following thing: rank 1, 1000. This is the maximum frequency. Second rank, 500. Third rank, 1000 over 3, 333, and so on. PITTS: I must say that since you have increased the size of the sample to the degree necessary to find one more species, which may be a very large degree, replacing those finite differences by derivatives leaves me with some doubts. VON FOERSTER: That is true, that is perfectly [loss of text]. ... start from a point, you take a square and take everything in that square, and make a list of all the dots belonging to each class. Naturally, if you allow that square to end at a limit, there will be the mean density of spots or dots of each particular type in the universe, and that forms an arbitrary set of numbers that can vary, I think, somewhere between zero and 1, all independently. PITTS: Now, what you are saying is that no matter what that set of population density is, for smaller samples at any rate, you will always get Zipf ’s distribution, with respect to the numbers inside any given square. Is that right? VON FOERSTER: No, I am not saying that. PITTS: Well, this is certainly a case where it is homogeneous from the point of view. VON FOERSTER: In that case, you don’t get Zipf ’s distribution. There is a particular complication in assuming the following thing: In my work, I had to assume an infinite number of kinds, that each kind should represent an infinite number of elements. Now, in the case of, let’s say, a [Poisson] distribution off a plane. PITTS: Well, a collection of objects, each distributed with a [Poisson] distribution, and there will be an infinite sequence of densities which we suppose converge to an infinite number of objects. VON FOERSTER: You get a peculiar result, a result which is much more probable; that with the next increase of your sample, you get always a new kind into the whole thing. Let’s say we can treat the whole thing the following way: Assume you have a tremendously large urn and in this urn are a lot of different marbles with an infinite number of colors. You are drawing marbles and you always make a note of what kind of marble you get. The highest probability you ever get is a new color with every draw. That is the highest probability. Actually, A Discussion on Zipf’s law 127 to get—and this would be a distinguishable case because if you are going on and drawing marbles, you get a certain series of red, blue and green, for instance. I collect those marbles also and I get violet, pink and orange, or something like that, out of the urn. That means we would both draw different samples. Therefore, our urn must have such a kind of statistics that we cannot distinguish whether you are going to draw samples or I am going to draw samples, or anybody is going to draw any kind of large samples; that we get precisely the same kind of samples out, it must have an internal distribution at this point of 1/r, so we get a like result. PITTS: But the crux of the process that distinguishes it from the ordinary sample is that you take the size of your sequence of samples, determined by the number of kinds in the content. VON FOERSTER: Yes. This I must say, to make a generalization of the function, F/r, because if I am increasing r my sample and I get something which I have already counted, then it would just perturb a little bit my rank-frequency curve, because then I have a particular unit to a certain frequency level for a particular rank. That means it would just be counted as a fluctuation, so I have to increase the sample until I get a new kind. BATESON: I am in some doubt about the definition of the classes that you are doing this with. Suppose you were doing it with beetles, and you extend the area at which you count beetles and find that you have included barber shops. Could this statistical method be made to discriminate between barber shops and beetles or is that left to you? VON FOERSTER: That is a very peculiar thing, and I confess I don’t know what are classes of elements, say, whether elements are in a certain structural relationship to each other. Talking about beetles, we know that we mean beetles, and beetles are not barber shops. But I must confess that I don’t know where to make the distinction, because you can invent some rather close questions, where it is rather arbitrary which is which. I must purposely exclude the barber shops. BIGELOW: What happens when you carry your sampling out to three dimensions? McCULLOCH: Oh, that doesn’t matter. VON FOERSTER: That is exactly the same. WIESNER: I didn’t understand the answer to Pitts’ question, which I think is rather important. PITTS: What apparently is the case is this: that apparently he so increases his size of samples, according to a curious sequence, depending upon the actual number of kinds visible in the content, in such a way that it converges to a different limit. BIGELOW: I don’t get that. 128 Heinz von Foerster et al. PITTS: I don’t see any reason for supposing that this curious sampling process actually acts in the accumulation of data to which Zipf ’s law applies. GERARD: Let’s go back to a single book. You say that the most common word is going to be twice as common as the next common and three times as common as the third most common one. Then you say, in English, the most common word is going to be “the.” If we go to German and that one word breaks up–. McCULLOCH: It holds. GERARD: –into three others, why does that still hold? McCULLOCH: Well, it wouldn’t be “the.” GERARD: Then it won’t be “the.” McCULLOCH: It would be “ein” or something else like that. WIESNER: There is a fundamental of the sampling process which I don’t quite understand. McCULLOCH: Let’s take a few of the things for which Zipf ’s law won’t hold. PITTS: What reason has he got for supposing the sampling process is actually carried out in that way in cases that give Zipf ’s law? McCULLOCH: Let’s wait one moment and let’s see the things for which Zipf ’s law doesn’t hold. It will not hold for chemical elements, or atoms. It will not hold for simple inorganic molecules. It may hold for protein molecules. We are not in any reasonable sample going to exhaust the richness of our universe. Wherever you exhaust the richness of your universe, you are going to come down. WIESNER: I want to bore a little further into the answer to Walter’s question. Are you saying that if I have a universe in which the populations are equally likely, I can get this kind of sample? PITTS; He is saying that it doesn’t depend on what the potentials of the population are. WIESNER: But if he says he doesn’t care, it implies that you can use the one you want to use. PITTS: Then you wouldn’t get Zipf ’s law. BIGELOW: This means that in each case, the increment he takes is a function of what he gets in the increment. He doesn’t really know that, however. PITTS: Well, he could. He could keep adding elements. BIGELOW: This is not the sampling process. PITTS: Not in the ordinary sense—well, yes, it is, really, because he doesn’t determine which element he picks. He keeps picking elements at random, but he stops as soon as he finds a new kind that isn’t already in the sample. He just determines where he stops adding to his sample and then looks at it. BIGELOW: But a random sample is one in which you don’t do this, in the ordinary sense. A Discussion on Zipf’s law 129 PITTS: Well, this is random in the sense that any particular sample is composed of elements, each one of which has been randomly selected. BIGELOW: But if you play a statistical game in which you are allowed to terminate the game when the variables have certain properties, you can always affect the answer. PITTS: Well, of course, but I mean, the individual choices are independent. WIESNER: May I ask what is the probability of getting this result? It must be very smal . If you allow me to start, as you said you would, with an equally likely distribution of all the elements in the population. PITTS: Incidentally, if that is so, of course, an argument of this kind will not proof it. BOWMAN: I would be inclined to put the statement more or less the other way round, and regard this as an observation rather than a proof of anything. VON FOERSTER: Right. BOWMAN: That there are a large number of classes of natural things for which the cosmological postulate holds, and let us just accept as a strange empirical fact. KLÜVER: It is not so difficult to think of this in connection with lobotomy. What I find surprising is that in a population of any city in the United States, five mil ion, ten million, or so, you have the same distribution. I am surprised at this strange observation, the facts here, if this is so. Al the other laws of history, then, do not seem to count for very much. What actually is underlying such a strange equation? MEAD: How many languages were used? KLÜVER: I would first like to know what is the fact. McCULLOCH: Three—Latin, English and Old German. BIGELOW: How close does it fit in these cases? McCULLOCH: Oh, surprisingly well. PITTS: He has samples of 10.000 or 20.000 at least and taken from all sorts of books, picked at random, at least so I would suppose. It impressed me. McCULLOCH: And it doesn’t matter whether you take English, modern English, or anything else, you get Zipf ’s law. There is one language that is deviant. Let me tell you, there is another thing that is rather interesting here. There is one language that is deviant, and that is Old Gothic. MEAD: Well, are there any non-European Languages in this? McCULLOCH: I don’t know of any that have been studied. I heard that somebody was studying Chinese. Now, here is one more point I would like to draw in here, if I may. If you study musical intervals, you find a corresponding law for Zipf ’s law holding for musical intervals. The Zipf ’s law breaks immediately. That has been studied from Haydn up to modern times, and it breaks immediately. BIGELOW: By whom, sir? Who carried out this study? McCULLOCH: One of Zipf ’s students. 130 Heinz von Foerster et al. PITTS: He has a whole project of people counting things. He did it for many years. McCULLOCH: It holds beautifully for musical intervals until you hit a man who set himself to write music in the following way: that he would use no interval twice until he had used all others once [laughter], and, of course, Zipf ’s law doesn’t hold. It is a perfectly flat horizontal line. VON FOERSTER: That was Schoenberg because he introduces very artificial means of creating music. McCULLOCH: Schoenberg was the first deviant from it. PITTS: The cosmological postulate is quite irrelevant. That is generally the assumption that is made in ordinary sampling theory, that if you take one additional element, the probability of getting a given value for it will not depend upon the ones you have selected, and they will be essentially the same and independent of the ones you have already taken, which is what you usually assume. This is the assumption of ordinary sampling theory. The way in which you get it, of course, is this very special method of sampling, and that is the assumption that really underlies it, not the assumption of the uniformity of probability, which certainly by itself would not do it. Now, what strikes me as strange is the supposition that the data are so gathered, according to the special sampling rule which alone would lead to Zipf ’s law. Can one suppose that all the people who have collected data on words have constructed a sequence of samples according to these rules? VON FOERSTER: I don’t know. PITTS: Or collected moths in a moth trap according to this rule? BIGELOW: One might, in fact, find such a trick in their data-gathering system. McCULLOCH: No, the whole of the book was counted. All of Joyce’s Ulysses and the whole of Homer’s Ulysses were counted, and Zipf ’s law holds for both. BIGELOW: There is one comment, that it is extremely difficult to find general empirical laws which hold like this in the history of science. There have been a number, all of which have been proven to be false alarms. This one may be true, but take the question of the Gaussian distribution, which is one of the things statisticians and scientists really believe in. You can produce situations where the Gaussian distribution does not hold in nature, where it should be thought to hold. A typical example is if you go out and measure the distribution of trees in a forest. If you measure the distribution as the function of diameter, the function of area, the law of the distribution of size of trees can’t be Gaussian and be this, too. VON FOERSTER: Yes, but I think the difference in the two kinds of statistics is that here we are dealing with a number of different kinds, whereas in the other case in the Gaussian situation, for instance,—we are dealing with one kind only. BIGELOW: But there can be many weasels where you identify a kind, you see. I don’t think that is the ease in this situation, but I merely say one should be A Discussion on Zipf’s law 131 cynical of rules which purport to extend over a very wide class of phenomena, and also; the statistics have to be examined extremely carefully, as well as the method by which they were gathered. PITTS: If this were true, of course, it would imply in a case subject to Zipf ’s law that the sampling data were of absolutely no value, and Zipf ’s law was completely an artifact and told you absolutely nothing about the actual population frequencies, say, of words, in actual English speech. But, really, it can have a very large random sample, and it is purely an accidental sampling. McCULLOCH: In essence, that is right; that is, if you had a large enough sampling of English so that you ran out of all the words in English, your Zipf ’s law from that time on cannot possibly hold. All you can do is shove this segment up. You can get more of those than you already have. PITTS: I think you have to be extremely careful about your sampling procedures to yield a result like this. ASHBY: Is it true to say that the old method of sampling is merely a device to get to the form of [P]. It is just a device for constructing the method for the argument. It doesn’t necessarily assume that that method is usually in practice. PITTS: It was that method that is used in practice. I would say that what you do in forming the statistics is actually the following thing: You go through the book, count the number of different words, and afterwards you order them in the rank-frequency diagram. That means you do certain things to that kind of a diagram. Therefore, you can already expect a certain kind of restriction of possible results because you are already doing something to the sampling. I would say certainly it doesn’t give any kind of an explanation to the whole thing. It just suggests the following thing, that if we have such a kind of universe, we would get a Zipf distribution. That is the only thing. I would say there is no explanation for the whole process. The next question which can immediately be thrown up is, why do we have this kind of universe? I can’t answer that. YOUNG: If you deal with letters, what would be the relation? PITTS: I don’t know. With letters, we are, unfortunately, very much restricted because we have only 24 letters. Therefore, the sample should always be smaller than 24, and then I would say that the whole thing is certainly far away from any possibility of getting Zipf results, because you have to assume an infinite number of kinds here, which is certainly not the ease in the language. But if you get a smaller sample, where not all words in that language are actually occurring, then you can fairly say you have not yet touched the limits of the universe but you have made a large enough sample that you can’t say it is too smal . MEAD: But is your assumption of occupations in a country that you have an infinite number of kinds? 132 Heinz von Foerster et al. VON FOERSTER: Well, at least the number of kinds must be related to the large number of kinds you actually collect. BATESON: Is this the same law that was running around in the twenties~ that came out from a botanist in Cambridge? 1t was called “aged area.” He argued, and had some very pretty curves, that if you got the number of seeds in a genus, you would find a very large number of genera, with only one species. McCULLOCH: The botanists have run into it, the biologists have run into it, and many others. BOWMAN: Will somebody define a genus for me? VON FOERSTER: No, nobody will. BATESON: The genus has to be set up by somebody who knows this law. YOUNG: Who knows something about the nature of genera of this sort, that all these rather arbitrarily defined genera– MEAD: Isn’t it a function of the method of definition, conceivably? Genera, after al , is a man-made definition of the universe. This may be merely a statement about a properly functioning Western mind. The two cases where it doesn’t work nicely are the schizophrenic and the lobotomized. All of these classifications are man-made; they are all made by Western science. KLÜVER: The population of Chicago and New York are not by definition. MEAD: Well, the conception of population is. BOWMAN: Where do we put the city limits? MEAD: And are dogs part of the population or not? Are unborn babies part of the population, or are potentially conceived babies part of the population, or are mice part of the population? You see, you have to work that all out. VON FOERSTER: But there is another conclusion which you can get from the etymological consideration. For instance, it shows that the barber shop is the leading profession in the United States. It shows that this would be the ideal, approximating a straight line. On the other hand, funeral homes are a very high frequency occupation. But it turned out to be that the barbers are a little bit off the curve, too low. This immediately suggests to me to become a barber [laughter], because then I would have a chance in that distribution and I would make an efficient barber. PITTS: Incidentally, I might say something about what Zipf did. Since he is now dead, it can’t redound to his harm. He got himself into some trouble about this. He took the size distribution of cities and countries such as the United States and Germany and so forth, and found that on the whole, they obeyed Zipf ’s law. Then he discovered that before the Nazis came into Germany, the distribution of population sizes by ranks was quite different from Zipf ’s law. It exhibited systematic deviations. Sometime after the Nazis came in, it was much closer to Zipf ’s law. He concluded from this that the Nazi government A Discussion on Zipf’s law 133 had been beneficial to Germany because it had corrected the distribution of the population in frequency. [Laughter] QUASTLER: That was because of the Anschluss, Germany plus Austria. PITTS: No, he took account of that, but the changes in population were in that direction. I think this is a very useful example of what a scientist can do if he is not careful. QUASTLER: I don’t want to darken the memory of the late Dr. Zipf, but why call this law by his name? He did not discover it. He wrote a big book about it in which he gave a very bad explanation. Incidentally, he stated that frequency and rank distribution in most of the objects you look at fall into one of three classifications, which is not generally the case. The distribution in words was discovered a long time before, by some Frenchman. McCULLOCH: Oh, it has been discovered in lots of fields, separately. The first man who gathered it all together, so to speak, was Zipf. That is the only reason, I think, for attaching his name to it. PITTS: He had a whole institute of graduate students who did nothing but count curious objects of one kind or another; that is, count members of the classes in classes of classes. WIESNER: Did he publish a book of those things? PITTS: I don’t know. McCULLOCH: In collections of coins, large collections of coins, you get Zipf ’s law again. It turned out that way. BIGELOW: It is fantastic. BATESON: If I have a multifaceted penny, say, of a hundred facets, with equal chances of throwing any one of those facets, and I experimentally start to throw and I chop off any sample according to von Foerster’s resume– VON FOERSTER: No, no! McCULLOCH: Will you tell him the comminutorial story? VON FOERSTER: I would say this is precisely the situation. The following thing does not work. Assume for instance, the urn, with an infinite number of kinds, where each kind is represented by an infinite number of samples. For each, there is the probability, P, an equal P. Then the most probable distribution curve you can get is the distribution curve of each sample occurring for each kind occurs once in your sampling. You can do it also in a different kind of way; for instance, if you have n different words, you ask what would be the book which would give the maximum amount of information. With [these] n different words, it would be a book in which each word would occur just once, or if you are writing a book which is much larger than the number of different words, then each word should occur with the same frequency. The result would be a curve like this [indicating on board]. This is the most 134 Heinz von Foerster et al. probable curve. This, on the other hand, is the curve with the highest amount of information. There is another thing which Warren and I were thinking about for a long time. It was the following situation: Assume you have a box with lots of equal white marbles in it. Now, you take one of those marbles out and you take a second marble out and you glue them together so that you get an element of a higher complexity. Afterwards, you throw that into the box. You again take out two things. It might be, by some chance, the already glued-together marble, or a marble which is not yet glued together. Then you glue those two elements together again. And so you proceed, setting up entities from higher complexity. Now assuming that the marbles would stick together, that the glue would be excellent, then after a certain number of processes, you would have one big ball with the complexity of n, where n is the initial number of marbles. But to get different kinds of the whole thing, where the kind now means complexity, namely, consisting of small and different marbles, you have to assume a particular half-life of any of these complex structures. You come immediately into the tremendous mess in the making of a particular theory, because you can assume the following thing: This complex element can break up in a situation where only one marble drops off or where two marbles drop off at the same time or where three marbles drop off at the same time or something like that. That means the problem of assuming a particular breakdown situation for this complex entity leads you to so many different possibilities that you can with ease establish a particular theory which would very nicely fit almost any kind of statistics. That means, finally, that it doesn’t mean a thing. That means we would just have to assume that the marbles are breaking up in that fashion, to furnish the kind of distribution in the box. PITTS: Have you seen Kolmogoroff ’s paper about the distribution of sizes of stones in a river bed, where he supposes that smaller ones have been derived by breaking up the larger ones? In a rather general condition, he shows that the logarithms of the sizes are Gaussian. VON FOERSTER: No, I have not seen that. I would say, as a probability, this does not work. I would say it does not work, to assume this principle. The information content idea, that it maximizes the information, is also important, because it is absolutely the same story as the probability. I tried another thing which was extremely complicated. I was asking myself, how many different kinds of those rank-frequency diagrams is it ever possible to draw? That is, if I plot here n different kinds, and I distribute them with a proper rank-frequency diagram, I can make, with a certain amount of things, a lot of different distributions which would look so and so. This problem, for instance, breaks up into what is known in mathematics as a partition problem, namely breaking up a particular number of things into components; that is, for instance, if I have a total number of words, A Discussion on Zipf’s law 135 let us assume it is 10, then I can assume the following: 10 can be established in having just one word occurring 10 times, or I can have two words, one occurring nine times and the other occurring one time, or eight times the one word and two times the other word, or seven times the one word and three times the other word, and so on. That means the maximum rank is equal to 1 or the maximum rank is equal to 2 in the first rank. I can go further and further, increasing the rank. I can ask myself, finally, what is the highest probability for having such a distribution? There is establishable a certain law, because this is breaking up the particular number, M, into a smaller number, m, and this is a particular number and you can maximize it with respect to n. Unfortunately, it doesn’t turn out that this is connected directly with that, which establishes what can cause an equal probability situation in this case. I would say there are lots of particular possibilities in which you can look into the whole thing, but for most of these things, it does not work so far as I can find out. PITTS: Speaking for myself at least, and probably also for Bigelow, I think it would be very good if you would insert even at somewhat greater length, the convincing mathematical derivation of your result in the transactions, because we don’t consider that this argument even makes it probable at al . It has so many holes in it. VON FOERSTER: Yes. Perhaps in a different discussion, we can talk about the whole thing. PITTS: You can easily insert it in two pages in the transactions. McCULLOCH: There are two more things. I know that Ralph Gerard has to catch a train in the not too distant future. It is about twenty minutes after four. I would like very much if we might deviate from the program I outlined and ask Walter Pitts to say a word or two, but to keep it pretty brief, about the investigations under way on synaptic transmission. That would tie back into the thing I know we want. Will you do it? Bubbles Everywhere in Human Affairs 4 Monika Gisler | Didier Sornette 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 13 16 15 4.1 Introduction Casual observations of the development of large scale entrepreneurial projects, reinforced by two detailed case studies, the Apollo program [Gisler and Sornette, 2009] and the Human Genome Project [Gisler et al. , 2010], has shaped our hypothesis that bubbles constitute an essential element in societal processes and in the dynamics of society at large. We believe that bubbles are ubiquitous in human affairs, providing fundamental driving forces for the societal courses of action. In that, they provide an important positive contribution to society. This view is a somewhat paradoxical statement because bubbles are usually understood as being bad. Our intention here is to outline our concept of ‘social bubbles.’ The roots of this concept grew out of Sornette’s research on the triggering mechanisms for market failures. In contrast to the view that the triggering mechanisms occur only hours, days, or weeks before the col apse, Sornette and col aborators have proposed that the underlying cause of market failure can be sought months and even years before the abrupt, catastrophic event [Sornette, 2003a; 2003b; 2005; 2008; Kaizoji and Sornette, 2009; Sornette et al. , 2004; 2009]. Furthermore, Sornette [2003b] has presented a general framework to characterize, quantify, model and predict financial bubbles and their aftermath. The main concepts explained and illustrated in details are imitation, herding, self-organized cooperation and positive feedback, leading to the development of endogenous instabilities. By probing major historical precedents, from the Dutch tulip mania that wilted suddenly in 1637, to the South Sea Bubble that ended with the first huge market crash in England in 1720, to the Great Crash of October 1929, and Black Monday in 1987, the conclusion was that most explanations other than cooperative self-organization failed to account for the subtle bubbles by which the markets lay the groundwork for a subsequent crash. According to this theory, the fundamental causes of the burst of bubbles lie in the maturation towards instability, due to positive feedbacks. As a speculative bubble develops, it becomes more and more unstable and very susceptible to any disturbance. If it has not been the raise of interest rate, it would have been something else that was the triggering factor for the crash. With the concept of social bubbles, we intend to open up this initial focus on financial bubbles and dig into the spheres of politics and social thinking, by investigating the investment in major innovations. Innovations are framed by interactions and novel relationships among science and technology, business and industry, economy, and politics [Freeman, 1974; Rosenberg, 1974; Dosi, 1982; Mansfield, 1995; Nelson, 2005]. Because change is so vital for long-run economic growth, it is of fundamental importance to understand how creative individuals and firms obtain the resources needed to undertake their investments in innovation and in- 140 Monika Gisler | Didier Sornette vention. It is also important to understand how the availability of such resources, including the manner in which they are accessed as well as the amounts that can be raised, influences the rate, direction, and organization of maturity. It is well known that investing in an innovation is not a clear cut choice. On the private side, it is imperative that the entrepreneurs decide, if, where, and how much to invest. In the case of public innovations, the situation is much more complicated. Increased government spending is often seen as critical to ongoing technological progress. On the other hand, the decision to invest in a specific project means to back out from other potential y social y worthy projects. In addition, commitment of public funding comes at the cost of taxation and/or increased debt for future generations, which may crowd out private investments and their associated initiatives. Several actors, decision makers, politicians, the public, the future generation, are involved. Personal involvements, the interest of the outcome, the incentives in general, are completely different among these different participants. A somewhat complex social network is at the heart of innovation processes. We believe that the social bubble concept provides a good tool for analyzing such processes. To do so, however, we first have to make sure how to understand what a social bubble essentially is. For this essay, we thus constrain ourselves to the discussion of the concept of social bubbles. Its concrete implementation will be a topic for further studies. Since our concept of social bubbles stems originally from the financial field, in a first chapter [2], we discuss the topic of financial bubbles in the literature. We do so mainly to show that the topic is not as clear cut even in finance as one would wish it to be. In chapter [3] we provide an analysis of the modern literature of an episode that is believed to be the first known bubble in history, the so-called tulip mania of 1634–37 in the Netherlands. It is quite interesting that the same historical episode evoked very different interpretations from different scholars (economists, historian of economics, social historians): whereas some bluntly refer to it as a bubble, others reject the idea entirely. This is a good example after all to enlighten the difficulties when it comes to define an economic episode in history as a bubble. The following chapter [4] gives an example of one of our own understanding of a social bubble, the Apollo Space Program. The factors combined in this endeavor wove a network of positive feedback that led to widespread enthusiasm and extraordinary commitment by those involved. Social bubbles are thus on the horizon when several arrangements, such as technological, economics, and political, become intertwined into a self-reinforcing spiral. The generic set-up will be the subject matter of our last chapter [5]. 4.2 Bubbles in the Literature Economic Bubbles (sometimes referred to as speculative bubbles, market bubbles, price bubbles, financial bubbles or speculative mania) form in economies, securities, stock markets and business sectors because of a change in the way players conduct business. This can be a paradigm shift, with the invention of new technologies [Perez, 2002], or a change of regulation, e.g. a (partial) deregulation of finance institutions. During the boom, people buy assets (stocks, real estates, etc.) at high prices, believing they can sell them at even higher prices, until confidence is lost and a large market correction, or crash, occurs [Kindleberger, 2005]. At the end, resources built during a bubble seem to be lost, causing prices to deflate. From this vantage point, most research on bubbles has proceeded by assuming it to be a ‘truism’. However, not all behavior of asset prices is necessarily a bubble, it is in fact difficult to determine whether and when an asset price is a bubble [Bhattacharya and Yu, 2008a], and more so, why the bubble crashes. The concept of the ‘new economy,’ a term used during the 1920s (the utility bubble), the 1960s (the ‘tronic’ boom) and the 1990s (the internet and communication technology bubble), captures vividly this mindset held by a majority of investors and firm managers at times of bubbles: there is widespread belief that times have changed irreversibly and for the better, and that a new epoch, a new economy, a new prosperity without business cycles and recessions is the novel emergent rule with the expectation of endless profits. Then, the ‘new economy’ bubble crashes, which is taken by the majority as a diagnostic that the bubble has been a waste of resources. There is a rich literature in economics and finance on the topic of bubbles, concerned with defining what is a bubble and developing a suitable theoretical framework [White and Rappoport, 1995; White, 1996; Galbraith, 1997; Shefrin, 2000; Shiller, 2000; Shleifer, 2000; Sornette, 2003b; 2005; Kindleberger, 2005, Bhattacharya and Yu, 2008b]. Not all of the writers were equally attentive to the fact that the definition of bubbles is far from being unambiguous. O’Hara [2008] has recently provided her own review and offers a somewhat disenchanting conclusion: “Are there bubbles? Are markets real y irrational? I am not sure. I do know that markets are very hard to predict and thus can seem ‘irrational.’ But I prefer a more neutral view.” [p. 16] Charles Kindleberger, in his book Manias, Crashes, and Panics [2005, 5th edition], was quite determined in identifying the financial processes at the origin of bubbles. He proposes that a bubble is an increase in asset prices in the mania of the cycle, in that asset prices today are not consistent with asset prices at distant future dates. Unsustainable patterns of financial behavior are thus at the origin of a bubble [p. 13]. Brunnermeier’s [2007] description that ‘bubbles are typi- 142 Monika Gisler | Didier Sornette cally associated with dramatic asset price increases, followed by a col apse’ goes in a similar direction. The concept of a bubble implies the interplay between an underlying mechanism, excess elements and necessary burst. The upswing usually starts with an opportunity (‘displacement’)—new markets, new technologies or some dramatic political change, and investors looking for good returns. Robert Shiller in Irrational Exuberance [2000]—published at the height of the dot-com bubble— proposed twelve factors that ‘propelled the market bubble,’ among them cultural and political changes favoring business success, challenging the role of specific judgment biases in finance. Similar characteristic scenarios have been described by Galbraith [1997], Kindleberger [2005], Sornette [2003b] and Sornette and Woodard [2009], corresponding to five steps: i) displacement ii) credit creation iii) euphoria iv) critical stage/financial distress v) revulsion. The scenario proceeds through the euphoria of rising prices, particularly of assets, while an expansion of credit inflates the bubble. In the manic euphoric phase, investors scramble to get out of money and into illiquid things such as stocks, commodities, real estate or tulip bulbs: a larger and larger group of people seeks to become rich without a real understanding of the processes involved. Ultimately, the markets stop rising and people who have borrowed heavily find themselves overstretched. This is distress, which generates unexpected failures, followed by revulsion or discredit. The final phase is a self-feeding panic, where the bubble bursts. People of wealth and credit scramble to unload whatever they have bought at greater and greater losses. The sudden fal , first in the price of the primary object of speculation, then in most or all assets, is associated with a reverse rush for liquidity. Bankruptcies increase. Liquidation speeds up, sometimes degenerating into panic. The value of col ateral (credit and money) sharply contracts. Then, debt deflation ends as productive assets move from financially weak owners (often speculators or the original entrepreneurs) to financially strong owners (well capitalized financiers). This provides the foundation for another cycle, assuming that all the required factors (displacement, monetary expansion, appetite for speculation) are present [Sornette and Woodard, 2009]. Since bubbles occurring in an economic context are seen as optimistic predictions about the future that prove wrong, they are considered to be bad. Their ominous character is amplified by the uncertainty stemming from the lack of a consensus on their causes, which make them a major challenge to economic theory. Only rarely bubbles everywhere in Human Affairs 143 and in passing has the question been brought up whether bubbles could also increase social welfare. Bhattacharya and Yu [2008a] asked whether it is possible that nascent, emerging industries need ‘animal spirits’ and overinvestment for innovation. They argue that it is likely that bubbles serve mainly to change the wealth dynamics of society, and through this mechanism affect the investment process. Only little research has been carried out in this concern. The social economist Carlota Perez believes that bubbles inevitably precede each of the ‘techno-economic paradigm shifts’ by which society advances. In her seminal book Technological Revolutions and Financial Capital. The Dynamics of Bubbles and Golden Ages [2002], she uses the term bubbles to describe the financial processes characterized by the instal ation of a new paradigm (or ‘revolution’) and its concentration of investment in the respective new (scientific) enterprise (infrastructure, human resources, etc.). Her analysis suggests that the working of markets cannot by itself explain the recurrence of major crashes and depressions. Instead, the emergence of these phenomena need to be explained by the analysis of the tensions, resistance, obstacles and misalignments that arise from within the wider social and institutional scene. Perez moreover criticizes the majority of Neo-Schumpeterians to have neglected Schumpeter’s legacy in that the accent was almost invariably on the entrepreneur, and has neglected the financial agent, no matter how obviously indispensable this agent may be to innovation. (“In Schumpeter’s basic definition of capitalism as ‘that form of private property economy in which innovations are carried out by means of borrowed money’ […] we find his characteristic separation of borrower and lender, entrepreneur and banker, as the two faces of the innovation coin.”) This view is even more relevant in the diffusion of radical innovation, which is inevitably a question of investment. She develops further her argument by emphasizing the role of the sources of capital that funded the deployment of new technologies. The establishment of the major infrastructures associated with dominant techno-economic paradigms in this view has to be linked to major technological investments, entailing the euphoric and reckless build-up of overcapacities of various kinds. Nevertheless, the lack of attention paid to the sources of capital that funded the deployment of new technologies, diagnosed in 2007 by William Janeway, has not been overcome. More research in this concern is considered necessary. The journalist Daniel Gross, in his 2007 book on Why Bubbles Are Great For The Economy, takes a counterintuitive look at economic bubbles. Common thinking states that excessive investment in fixed assets is bad for investors, for the employees of the bubble companies, and for the economy as such. On the contrary, he argues that during bubbles, investors’ money is used to build infrastructure that can’t possibly repay its upfront costs, but ends up being beneficial for companies and 144 Monika Gisler | Didier Sornette consumers in the long run. To take a recent case, most investors in the ‘dot-com’ episode lost, but their money built the software and infrastructure that runs today’s Internet. In fact, what is likely to be seen as a catastrophe from one point of view, others, such as Perez [2002; 2009] and Gross [2007], see as a collective social gain stemming from bubble behavior (major investments with low short-term returns) in the long run. Innovation processes and the creation of new technology seem inherently associated with bubbles. These authors suggest that crashes/crises/ bursts are unavoidable epochs covering about 10% of the time but, on the other hand, they provide benefit for the remaining 90% of the time. Our own research extends and makes more precise these considerations. The difficulties associated with the characterization of an episode as a bubble have been stressed by Peter Garber in his book Famous First Bubbles [2000], in which he revisited the three bubble episodes, the Dutch tulip mania of 1634– 37, and the Mississippi and South Sea bubbles of 1719–1720. He herein argued that a bubble is “a fuzzy word filled with import but lacking a solid operational definition” [p. 4]. In line with standard financial economic theory, he defined a bubble as an “asset price movement that is unexplainable based on what we call fundamentals” [ ibid. ]. Under this view, bubbles could be positive or negative. Garber is so cautious about defining a bubble that he suggests to see explanations of market anomalous behaviors in terms of bubble as a ‘last resort,’ because they do not explain events thoroughly; they are merely a name attached to a financial phenomenon that has not been sufficiently understood [p. 124]. Accordingly, cal ing a sharp financial expansion that col apses a ‘mania’ or ‘bubble’ is too superficial and should be corrected by a deeper insight. In the three examples he studied, he finds evidence that rational behavior can explain the observed price dynamics. The conclusions of Garber’s study have been questioned, among others by Kindleberger [2005] and Chancellor [2000; on an earlier of Garber’s articles]. The problem of defining what is a bubble is fundamentally influenced by the underlying economic model, which provides the reference point to judge what is ‘normal’. For instance, assuming that markets ought to always be rational and efficient poses a fundamental problem of measuring something that is deemed impossible by construction. The next chapter highlights these difficulties with the example of the tulip mania. 4.3 The Tulip Mania Often cited as a quintessential example of a bubble, the ‘tulip mania’ arose in Hol and in the years 1634–1637. Its essential is that rare bulbs were hard to produce but, once obtained, they were relatively easy to propagate. During this period, tulip bulb prices rose dramatically, with prized specimens allegedly selling for the equivalent of more than what corresponds to the purchasing power of $30,000 in present days. Otherwise sensible merchants, nobles, and artisans supposedly spent all they had (and even borrowed heavily) on tulip bulbs. Starting in 1637, however, prices col apsed, ending the first well-known great bubble episode. Whether or not a tulip bubble was present has ever since played at a central stage of heated debates. The disagreement stems from the clash between those who believe that markets are always rational and efficient and those who call attention to the ubiquity of financial crises. A particularly interesting divergence in the discussions is whether or not the traders and likewise the markets have behaved rationally or irrationally. Garber [2000] argues against the generally accepted view that this was a bubble. He points out that the fact, that rare varieties of tulip bulbs could fetch high prices among professional traders, was not necessarily irrational behavior. On the contrary, the tulips could be used to grow many more valuable hybrids and often earned their purchasers far more than they invested. One of his arguments against a bubble interpretation of the Tulip episode is that no economic distress followed the end of the craze as might have been expected if it was a bubble. This is confirmed by the insights of a recent study of historian Anne Goldgar [2007], who claims that while it is certain that some people lost a lot of money, there is no archival evidence suggesting that any of the involved merchants went bankrupt. The claim that the crisis caused a disruption of the Dutch economy rests solely on some contemporary pamphlets that are highly critical of the trade and use it as a warning against greed, materialism, financial speculation and subsequent social disorder.1 1 Anne Goldgar highlighted a problem that none of the other authors writing on the Tulip mania have ever been concerned with: that the data on which the tales about the Tulip mania are based are more than weak. The most often cited references [Charles Mackay’s 1841 Extraordinary Popular Delusions and the Madness of Crowds as well as the several contributions by the Dutch Historian of Economics Nicolaas W. Posthumus, published in the 1920s] are not contemporary and are thus unreliable. Goldgar stresses that Posthumus’ texts are full of incorrect transcriptions and misprints, which caused a lot of wrong data. Mackay on the other hand took most of his accounts from a contemporary, Johann Beckman, and did not check the sources. The accounts 146 Monika Gisler | Didier Sornette Kindleberger [2005] has energetically discarded this view, pointing out that the Dutch economy slowed in the 1640s before putting on a tremendous spurt again after 1650 (there are no price data immediately after the crash). The decline in the prices of tulip, Kindleberger assures, led to a decline in economic activity. Garber’s second and more relevant argument is that the market for bulbs was not out of line, even though traders were pursuing potentially irrational strategies. The Netherlands, he argues, was a sophisticated trading center, with well-developed commodity markets. The investors were bidding up the price of shares, financed by a captive bank, in the hope that supply would expand later to justify the higher prices. The role of the bubonic plague of 1634–36, decimating Europe at that time, played a crucial role, adds Garber. He shows that the individuals involved in the trade were merely seeking entertainment, after the disastrous bubonic plague, confident that any substantial losses would be written down by the state. He further argues that: “The wonderful tales from the tulipmania are catnip irresistible to those with a taste for crying bubble, even when the stories are so obviously untrue. So perfect are they for didactic use that financial moralizers will always find a ready market for them in a world filled with investors ever fearful of a financial Armageddon” [p. 83]. Kindleberger [2000] in a review of Garber’s book, sees it to be of interest to analysts with a strong commitment to rationality and market efficiency, because it provides a re-examination of the question of whether the use of the term bubble is appropriate or not. A rational investor is one who seeks to optimize his wealth by offsetting risk with reward and using all publicly available information. Were the traders of bulbs behaving rationally? Kindleberger’s answer is no. Nor does Kindleberger believe in efficient markets. Many investors, he claims, do poorly in the market because they chase the latest fashion. This is herd behavior. Only if one is assured that all of us and our actions/reactions are always rational, he concludes, one could explain bubbles just from the speculation of rational investors. Hence, “complexity permits one to say that markets are mostly reliable but occasionally get caught up in untoward activities.” Ross [2005] argues against Kindleberger by stating that modern finance never said, nor required, that individual investors be rational and thus a bubble is necessarily defined by deviations from rationality. What matters, Ross asserts, is that there are a few sharks, or arbitrageurs, who wait for opportunities and then pounce. This makes markets behave ‘rationally’ even if individual participants may be irrational. delivered therein are simply not true. The argument of the Tulip mania being a legend or myth is fueled by such misinterpretations. bubbles everywhere in Human Affairs 147 Chancellor [2000] devotes a paragraph to the interpretation of the tulip mania episode provided by one of Garber’s early articles [Garber, 1989]. He rejects Garber’s hypothesis as an act of ‘historical revisionism’ performed with the goal of supporting the efficient market school of economic thought according to which bubbles or mania cannot exist since market prices always reflect their intrinsic value. The high prices for tulip bulbs could never have reflected the rational expectation of investors, he holds, since it was not known until the twentieth century that the variegate tulip petals were the result of a virus which attacked the bulb. He sees Garber’s assumption as a somewhat wrongful (if not to say ideological) rewriting of what is known already. In our work on financial bubbles, we have found that bubbles are not necessarily followed by crashes. Instead, bubbles are non-sustainable transient regimes that end at a tipping point, beyond which a new regime is established. The new phase can be a crash followed by revulsion, or simply a plateau or slow decrease of the market [Sornette, 2003b]. Therefore, Garber’s first argument that the absence of bankruptcies and of an economic recession following the Tulip mania bursts is a testimony of the absence of a bubble becomes pointless. On the other hand, the argument that markets are always rational, efficient and col ate all available information is difficult and perhaps impossible to debunk since any behavior could in principle be attributed to some rational speculative strategies. This raises the issue of falsification of the rational expectation and efficient market hypothesis, which remains unsettled. Indeed, the problem of testing the efficient market hypothesis (EMH) or the rationality assumption is that there are not strict tests for them only. Any test comes as a joint test of the hypothesis and of a specific mathematical formulation. This is referred to as the ‘joint hypothesis’ problem in financial economics. The joint hypothesis problem means that market efficiency as such can never be rejected, because any test of the EMH is a joint test of an equilibrium returns model and rational expectation. To make the EMH operational, one must specify additional structure, such as investors’ preferences, the amount of available information and how it is accessible to investors, and so on. But then a test of market efficiency becomes a test of several additional hypotheses as well. The rejection of such a joint hypothesis does not provide any clues on which aspect of the joint hypothesis is inconsistent with the data. In order to make progress, other approaches may be considered. For instance, Sornette et al. [2007, 2008] have developed a working methodology to validate/falsify a given model and/or hypothesis, based on testing the degree to which predictions of the model are born out by crucial empirical tests, and thus increase/decrease our trust in the model. 4.4 The Apollo Space Program2 The Apollo program was one of the most ambitious and costly single project ever undertaken by the United States in peacetime. With the first men landing on the Moon in 1969, it was the general belief at the time that thirty years later at the transition to the third Millennium, mankind would have established permanent stations on the Moon and on Mars, with space travel expected to become almost routine and open to commercial exploitation for the public. Today, we know better. In our study of the Apollo program [Gisler and Sornette, 2009], we stressed that this (too) optimistic view exemplifies the bubble spirit that is typical rather than exceptional. Enthusiasm was always present to push for the endeavor, risk always a topic but never an issue. Major risks have been accepted by individuals, first of all by some of the pioneers (engineers, astronauts). They did not shy away from taking all possible types of risks, even at the costs of their own life or health. The cold war and the space race were indeed important factors in the formation of the Apol o program; however, we argue that there were other equally important and perhaps even more important factors at play. Support for this hypothesis can be found in the fact that the Cold War did not end with the termination of the Apol o project; neither did col aboration between the US and the USSR start after the col apse of the latter in 1989, on the contrary, the SOJUS project started as early as 1967. Our interest hence focused on the internal perspectives that played a role in the development of the Apollo program. It is important to note that the program enjoyed high visibil-ity and strong interest from a large fraction of the population, including its financial and technical components. We argue that the Apollo program developed as a bubble, first mounted up by a special interest group, later inflating to a very large size, through general positive feedback mechanisms. Indeed, the Apol o program enjoyed a tremendous support, financially as well as societal. The Apollo program was originally conceived early in 1960, during the Eisen-hower administration, as a follow-up to America’s Mercury program. The goal was to develop the basic technology for manned spaceflight and investigate human’s ability to survive and perform in space. At its peak, the Apollo program employed 400’000 people, and required the support of hundreds of universities and 20’000 distinct industrial companies. Enthusiasm was high in terms of funding as well as in terms of human effort. Its objective was twofold and its expectation was soaring: the immediate goal, as proclaimed by President John F. Kennedy before Congress in 1961, was to land men on the Moon. A second and 2 This section derives from our paper on the Apollo program; see Gisler and Sornette, 2009. bubbles everywhere in Human Affairs 149 far broader objective was to make the United States preeminent in space, taking a leading role in space achievement and ensuring to prove the nation’s ability to explore the riches of the solar system and beyond. Kennedy did not immediately come to a decision on the status of the Apollo program once he became president. He knew little about the technical details of the space program, and, what’s more, he was put off by the massive financial commitment required by a manned Moon landing. Despite public support, Kennedy had expressed concerns about the program and the funds that it absorbed. His plans were abruptly changed by two unexpected events in mid April 1961: The first man in Earth orbit by the soviets [Yuri Gagarin] and the CIA-backed Cuban exiles invasion failure at the Bay of Pigs. Kennedy, as a result, seized on Apollo as the ideal focus for American efforts in space. He ensured continuing funding, shielding space spending from the 1963 tax cut and diverting money from other NASA projects. Even though Kennedy did not say it directly, he had significant concerns with respect to the risks associated with putting humans into space and the corresponding public consequences of possible failures. The reward for the United States when people were willing to take risks and to explore through manned space flight was obvious to Kennedy; it was the public he had to convince and make sure that it understands it. After Johnson became president in 1963, his continuing support of the program al owed it to succeed in 1969. The U.S. commitment to space progressively captured the American imagination and attracted overwhelming support. No high official at the time seemed deeply concerned about either the difficulties or the expenses. This was essential y to the credit of Johnson’s effectiveness in building a nation’s consensus for a space program. Observing some of the most powerful people endorsing the idea of a space program made it easier for the rest to follow. Johnson increased NASA’s spending from 150 million to over 4 billion USD, to develop the required technology and the science needed to build a human presence in space. Johnson argued that the space program as expensive as it was could be “justified as a solid investment which will give ample returns in security, prestige, knowledge, and material benefits.” [L. B. Johnson cited by Dallek 1997:73]. It was of little use that economists argued against it. Pol s carried out in 1964 showed that he had made his point to the public, who overwhelmingly supported him. The fact, that the rational cost-benefit analysis put forward by economists was ignored, provides a vivid illustration of the bubble spirit mentioned previously in our introduction. Financial support was not a given thing, though, but had to be negotiated on a yearly basis during the entire project [fiscal years 1961–1969]. The Apollo share of the total NASA Budget increased over the years from 10% [1962] to 70% in 1966, when it reached its peak [Ertel and Morse, 1969–1978]. In 1963, the final 150 Monika Gisler | Didier Sornette fiscal budget for Apollo was still 0.62 billion USD or 17% of the entire NASA budget. In 1964, it increased to 2.27 billion USD or 57% of the overall NASA budget, even though the aim of putting men on the Moon was still far beyond reach. Public complains about the costs in 1963 did not put an end to the support by Congress; it merely led to a momentary halt of the budget increase. In 1965, the total sum of 4.27 billion USD went to NASA, and thereof 2.61 billion USD to the Apollo program. The funding reached its zenith in 1966 with an overall NASA budget of 4.51 billion USD, of which 2.97 billion USD or 67% went to the Apollo program. After the initial rise of efforts embodied in the Apollo program, space exploration reached equilibrium, accompanied by drastic budget reductions; the fiscal budget in 1971 was 0.91 billion USD for Apollo or 36% of the overall NASA budget, and in 1973 it was down to a minuscule 3% fraction only. The last Apollo mission landing astronauts on the Moon was in 1972, when Apollo 17 concluded the Apollo mission. Thereafter, the United States did not undertake any other Moon flights. Enthusiasm for the program was usually high, even though not necessarily un-critical. Launches from Cape Canaveral drew hundreds of thousands of excited spectators. In 1962, the volume of sales of space toys and kits in department stores after each launch was proof of the public attraction to anything related to space. As early as 1963, however, criticism was beginning to grow in the press. Whereas in 1962, the editors used superlatives when talking of space exploration, in 1963, fervor was somewhat low. The lavish amount of money being poured into NASA was being questioned, should it go up continuously. A number of writers criticized the program as a cynical mix of public relations and profit-seeking, a massive drain of tax funds away from serious domestic ills of the decade, a technological high card in international tensions during the cold war era. All the same, the pendulum again swung towards the space program. After this temporary drop of enthusiasm in 1963, polls in the spring and fall of 1964 showed 64–69 percent of the public were favorably disposed to landing an American on the Moon, with 78 percent saying the Apollo program should be maintained at its current pace or speed up. Financial support was not a given thing, though, but had to be negotiated on a yearly basis. Polls performed in summer 1965 showed another slight decrease of enthusiasm, as a third of the nation now favored cutting the space budget, while only 16 percent wanted to increase it. Over the next three and a half years, support for cutting space spending went up to 40 percent, with those preferring an increase dropping to 14 percent. At the end of 1967, The New York Times reported that a poll conducted in six American cities showed that five other public issues held priority over efforts on outer space. The following year Newsweek echoed the bubbles everywhere in Human Affairs 151 Time’s findings, stating that the United States space program was in decline. At the same time, Congress was strongly leaning towards a reduction of NASA’s budget. A White House survey of congressional leaders at the end of 1966 revealed pronounced sentiment for keeping Apollo on track but, simultaneously, for cutting NASA spending by skimping on post-Apollo outlays. The efforts of 1966/67 paid their tributes: When Apollo 11 made it to the Moon on July 19, 1969, an estimated 600 million people—one fifth of the world’s population—witnessed it on television and radio. Some observers designated the day as a turning point in history. After Apollo 11 had landed on the Moon, lunar scientists as well as astronauts became highly optimistic about the outcome of scientific research associated with orbiting flights and exploration of the Moon. Astronauts had demonstrated that men were able to function as explorers in the lunar environment. They were viewed by the advocates for manned space flight as ample justification for the enormous investment they required. Hopes of the scientists for resolving major questions about the origin and evolution of the Moon reached a peak of optimism at the beginning of 1970. With astonishing rapidity, however, the raison d’être of the Apollo program was undergoing a metamorphosis. By the spring of 1970, it was obvious that the intellectual rationale for Apollo could not justify the full program in the absence of enthusiastic public support, and that was waning. In November 1969, Apollo 12 astronauts achieved a second lunar landing and made two Moonwalks. Once again, there were live pictures, but the reactions lacked of enthusiasm, the public was progressively more disenchanted with the space program. The voice of Apollo’s critics began to swell. The national polls in the summer of 1969 found that 53 percent of the country was opposed to a manned mission to Mars. And a poll taken in 1973 showed that only foreign aid had less support than space exploration. In the short run, project Apollo was an American triumph. In the long run, the costs, close to 25 billion USD in total in 1960s dol ars, were large and might have made a difference in other programs. In the context of bubbles associated with innovation ventures and the creation of new technology, the Apollo project demonstrates the large risks that have been undertaken individually, politically and financially, leading to a collective (individual, public and political) over-enthusiasm, which played a very significant role in the development and completion of the process as such. The qualifier ‘over’ emphasizes that the enthusiasm did not out-live by much the first Moon landing, and a general positive sentiment in favor of the Moon exploration started to fade shortly after the first step on the Moon. The evidence gathered here supports the view that the Apollo program was a genuine bubble, with little long-term fundamental support from society. It led to innumerable technological innovations, and scientific advances, but many of them at a cost 152 Monika Gisler | Didier Sornette documented to be disproportionate compared with the returns. These returns may turn out to be positive in the long run as many of its fruits remain to be fully appreciated and exploited. Enthusiasm was constantly, even though not univocally present to push for the endeavor, risk always a topic but never an issue. The program was first nucleated by a special interest group, which inflated to a very large size only through the general positive feedback mechanisms. As a result, the Apollo program enjoyed a tremendous support, financially as well as societal. At the same time, Congress approved funding until its peak in 1966. The ‘upheaval’ by the public in 1963 had no direct manifestation, at the time Congress was strongly willing to increase NASA’s budget. After 1966 though, support by Congress slowly decreased until it was cut off almost entirely under Nixon, terminating it in 1972. We have called the Apol o program one of the most exceptional and costly project ever undertaken by the United States at peacetime. However, just as Apol o had come out of nowhere, and held center stage for a decade, it vanished from the public consciousness, as if it had never happened. It thus was an exceptional niche, not having been revisited to any significant degree ever since. 4.5 Outcome The example of the Apol o Space Program has revealed that, in situations where new technology or scientific options are on the horizon, i.e. when a new horizon opens up, and individual or groups believe to be ready for it, then they do it, whatever the risks and the consequences are. The factors combined in this endeavor then weave a network of reinforcing feedbacks that lead to widespread over-enthusiasm and extraordinary commitment by those involved in the project. Our concept of social bubbles tries to grasp such episodes. When several arrangements, such as technological, economics, and political, become intertwined into a self-reinforcing spiral, i.e. when social networks and their financial and societal dynamics are at stake, such situation is adequately refereed to as a social bubble. The central factors can be decomposed into: i) idea (scientific/technological) ii) credit creation via public/private investment iii) enthusiasm iv) saturation of the idea (exploration comes to an end)/open critic v) program termination. The appearance of a new idea is followed by upward engagement (financially, man power, time load), connected to high expectations and utopian goals that push a bubbles everywhere in Human Affairs 153 project for an extended period. Collective over-enthusiasm, as well as unreasonable investments and efforts derived through excessive expectations of positive outcomes, lie at the bottom of such processes. People want to get the maximum, but have no such opportunity most of the time. Only during these times do they dare explore new opportunities, many of them unreasonable and hopeless in a non-bubble context, with rare emergences of lucky achievements. And even when investments flop with the crash of the bubble, businesses and consumers can find themselves with a usable commercial and industrial infrastructure at large, which they can progressively put to new uses. Resources created during bubbles do not disappear in the long run, when its investors go bust. It more often than not gets reused, by entrepreneurs with new business plans, lower cost bases, better capital structures [Gross, 2007]. Our overall goal is thus to explore and test the evidence that bubbles lead to a lot of destruction of value but also to the exploration and discovery of exceptional niches. The investigation of other examples of major inventions should strengthen this point: the Human Genome project [Gisler, et al. , 2010], the cloning of mammals (Dolly, the sheep), the ITC bubble culminating in 2000 [Sornette, 2008], or the adventure of nanotechnology (impossibility to keep up with the pace initially announced very optimistically). One thing in common to all this topics is the presence of extremely high expectations towards the outcome of the proposed research and/or innovation project. The large enthusiasm at the inception of the project prompted the undertaking of very risky decisions, which opened the road towards tremendously high innovations. Some of them led to dynamic societal progress and structural changes, others captured the imagination of large societal groups and proceeded along a roller-coaster of rising expectations, steep growth and spectacular downturns, while their potential future benefits are still uncertain. They all constitute an essential element in the dynamics of important inventions or innovations, and are thus crucial for society. New Models for Generating Power Law Distributions 5 Günter Haag 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 13 16 15 5.1 Introduction The Pareto distribution is fascinating scientists since many years, not only because of its seemingly very simple mathematical structure but also because scaling laws are some particularly simple cases of more general relations. It should be emphasized that scaling never appears by accident [Barenblatt, 2003] but reveal an important property of the system under consideration, namely self-similarity. This means that the structure of the system reproduces itself on different scales. In physics many systems can be found which exhibit scaling behavior over many orders. In social sciences, e.g. for the city size distribution, a potential law can be found with acceptable accuracy only for a few rank-size-decades often within a population interval between about 1 Million to about 1000 inhabitants. The mechanisms behind the urbanization process, namely the transformation of rural into urbanized landscape is of significant interest among researchers of different disciplines. Since Auerbach [1913], Gibrart [1931] and Zipf [1949] possible explanations of how the observed size distributions are generated are discussed. Different aspects of city growth have been treated on the basis of different frameworks [Beckmann, 1958; Roehner and Wiese, 1982; White, 1977; Wong and Fotheringham, 1990; Guerin-Pace, 1993; Curry, 1964, to mention a few]. However, since the rank-size distribution reflects just one dimension of a highly dimensional dynamic process of urban growth it is not surprising that all the different frameworks come out with size distributions more or less similar with the observed ones. The explanation is thus shifted from the size-distribution issue to the constraints which have to be defined to force the theory to conformity with the process [Pumain, 1982; Pumain, 2006]. A description of the size distribution by mathematical models which attempt to replicate data suffer from the fact that they are easily adjusted to the observations, because of the simple structure of the distribution. This problem was already encountered by Quant [1963] and Robson [1973] and is referred to as the “over-identification of models problem.” This has the following consequence: The quest about an appropriate dynamic modeling of the urbanization process and its projection to the size-distribution can only be decided by a comparison of the complex underlying empirical and theoretical interaction processes and not via looking at the size-distribution. This means that we do not need a further theory for the explanation of the size-distribution but rather a good framework for the dynamic urbanization process of highly interacting units. In the following some new modeling ideas are considered with respect to this issue. 158 Günter Haag FIGURe 5.1 Hierarchical Differentiation in City Sizes ��������� ����� �������������� ������� ���� ������ ���� ���� ���� ������� ���� �������� ����������� ���� ������� ���������� ������ ������ ��� ������ ������������ ����� ����� � �� ��� ���� ����� ���� Sources: europe: Moriconi-ebrard F., 1994, GeOPOlIS/India: Census of India 2001/USA: United States Census 2000/ France: INSee, Recensement de la Population 1999/South Africa: Statistics South Africa, Census 2001, base CVM. The power-law distribution or more general the term scaling describes a very simple situation namely the existence of a relationship between nk =n 1k-q (1) the population n of a city and its rank k. The rank-size-coefficient q and n the k 1 population of the biggest city are constants. Scaling laws reveal the phenomenon of self-similarity. This property requires that certain characteristics of the system reproduce itself on different scales. The stability of the rank size distribution of cities over the last century and the question of the general appearance of such distributions in most countries of the world are an important issue of research. During the last twenty years many computer experiments have been performed to illustrate under what conditions power laws for city size distributions can be found [Pumain, 2006]. Batty [2006] demonstrated in a number of computer experiments that the distribution generated from the Gibrat process with diffu- New Models for Generating Power law Distributions 159 sion are somewhat flat and as the level of diffusion is reduced, the hierarchical structure begins to disappear.  nkt k  1 0 . 8 (2) i n t 1  (1  ) it i n t     n 4 it The underlying process assumed (2) describes the dynamics of a number of cities in a grid at place I at time t with size n and a randomly chosen growth it rate ε according to Gibrat. In order to model the interaction between the it cities some simple diffusion to adjacent cities is added via a fixed proportion λ of the population diffuses to its nearest cities. In case of the introduction of agglomeration economies into the model by adding a term 0 = 0.2 reflecting current city size the sharpening of the distribution of city sizes according Figure 5.2 comes better in line with real experiments. FIGURe 5.2 City Size Distribution for the Model (2) at t = 1000 and t = 10.000 ��� ��� ��� ��� �������������� ��� ��� � � � � �������� Source: batty [2006] To introduce a different form of hierarchy Batty introduced much more explicit networks of interaction. In each time period, for a city which is already linked to other cities a random link is added an the new population size n of city I is it+ 1 modeled via n it  1  n  ijt i  tk 1 j (3) 160 Günter Haag In (3) n is the total number of links from city I to j and the random addition of ijt a link is described by δ . It is assumed that a link is added or not depends on the ikt size of the city n and its distance d to other cities via a gravitational function ijt ij    , 1 if rnd (  )  Kn exp(  d ) ijt 1 ijt 1 ijt ij (4) where the term rnd( ε ) determines a random choice. The parameter β reflects ijt+ 1 the frictional effects of distance. However, this model (4) is related to a scale free network [Barabasi and Albert, 1999]. FIGURe 5. 3 Rank-Size distribution of the Nodal Network Distribution ��� �� ���������� � � �� ��� ���� ���� Source: batty [2006] Figure 5.3 shows the Zipf plot where in addition to (4) an algorithm is added for the birth of new nodes. This population distribution is close to the pure Zipf scaling. It confirms that this model of preferential attachment based on Gibrat does generate a distribution compared to the simpler non-network cases. However, empirical data of migration flows, traffic flows or flows of commodities in a network of cities clearly demonstrate that the underlying assumptions concerning the spatial interaction of cities are quite different from the above discussed issues. 1) There are no diffusion terms λ in the population dynamics as introduced in (2). The exchange of population between cities is based on migration. New Models for Generating Power law Distributions 161 2) The nodes (cities) are not randomly linked. The flows are rather determined by economic reasons of transport, labor market conditions and general economic impacts. 3) Agglomeration effects as introduced in (2) lead to the amplification of the population growth with increasing population size. This leads in the long run to a steady increase of the slope of the rank-size distribution and not to an almost stable rank-size coefficient q. A further critique on current modeling efforts is found in Pumain and Haag [1994] in the sense that the geographic specificity of a set of cities must be integrated in mathematical models. The above obstacles can largely circumvented by incorporating a stochastic approach. The master equation framework provides a flexible tool for passing from individual decisions of different agents (micro level) to a macroscopic consideration of a system of cities. Therefore, it seems to be advisable to start with an integrated dynamic urban model where al parameters and functions involved have a clear meaning and can be related to empirical observations. The simulation of this integrated urban macro model then should lead under certain conditions to the observed hierarchical organization characterized by the rank-size distribution. Of course, the self-organization process, and therefore the development of the skewness and complexity of the distribution, is related to the model parameters and the flows within the network of settlements (flows of people, materials, information, etc.). The interactions among the different elementary units of the system (the different settlements, agents, economic effects, etc.) depend in general on the spatial distribution pattern of the elementary units. Self-accelerating processes and saturation processes have to be considered. The elementary decision process of an individual to migrate depends on those impacts, as well as on psychological, social and economic conditions. The occurrence and temporal stability of the rank-size distribution is then obtained as the result of a dynamic self-organization process of the urban network [Haag, 1993]. Under these conditions, the rank-size distribution of settlements can be considered as a stable attractor in urban growth [Haag and Max, 1995]. 5.2 The Dynamics of Urban Pattern Formation The Rank-Size as a Self-Organization Process The master equation is the starting point [Weidlich and Haag, 1983; Haag 1984; Weidlich, 1984] for the modeling of the urban system. Since settlements can not be considered as isolated systems, rather as elements of an urban network, flows between the urban elements (nodes) are of fundamental importance. 162 Günter Haag In addition the settlement/hinterland interaction and the emigration and immigration rates as well as birth/death events have to be considered in this stochastic framework. Let the urban system be composed of L settlements and the hinterland. Then [Haag et al., 1992] the most probable development of the city size can be described via L L i nt  (1 1 i  t) int 0  ifjtnjt exp[ iut jut]  0  jfit int exp u[ jt iut] ih W  h W i j1 j1 (5) where nit is the population size of city i at time t and ρ its natural growth rate. it The two sums describe the in-migration and out-migration of population. The migration flow between the cities I, j is driven by differences of attractiveness ( u – u ) of the different settlements I, j. The parameter v is a measure of the it jt 0 mobility of the population and the barrier effects (distance deterrence function) are modeled via the symmetric matrix f ( d ) = f ( d ). Emigration and ijt ij jit ji immigration effects are modeled via W , W . Evidently (5) is a set of L coupled ih hi nonlinear first-order differential equations for the mean city sizes. The change in population size of city I is due to migration events between cities as well as to interactions with the hinterland and birth-death events. Specifications of this model were discussed in the case of interregional migrations [Weidlich and Haag, 1988]. A hierarchical ordering of the settlements is assumed such that n is the largest 1 t city, and n is the population of the i-th ranked place. Now the nonlinear it transformation  q ( t) k n (6) kt  n 1 t k is introduced, where q ( t) are denoted as rank-size coefficients. Insertion of (6) in k (5) yields after some minor manipulations an equation of motion for the rank-size coefficients [for details see Haag 1994]: dq ( t ) k    f ( q ( t), f ,  , u ,. .) ijt 0 t (7) dt The rank-size distribution would be a stable attractor of the spatial system of settlements if lim kq( t) q for k2,.., L (8) t However, this seems to be a rather restrictive condition. New Models for Generating Power law Distributions 163 Equation (5) or (7) becomes fully explicit by insertion of the analytical form (assumption A or B) of the attractiveness u ( n ): k k Assumption A 2 ukt   kt   nkt   nkt (9) In assumption A the attractiveness u ( n ) of city k in dependence of its population k k size n is represented as a truncated Taylor series up to second order. Positive k agglomeration effects, κn increase its attractiveness with increasing size, negative kt externalities are represented by the term σ n² kt . This means it is assumed that an optimal city size exists, which maximizes the cities attractiveness. All other socioeconomic influencing factors (not size dependent) merge into the term δkt . Assumption B u     log n (10) kt kt kt In assumption B, a convex decreasing marginal settlement attractiveness is assumed, approximated by the term κ log nkt . This assumption seems to be rather plausible, compared to assumption A. In Figure 5.4 the two assumptions A, B are shown for a subset of the French city system consisting of the 78 largest urban units. This contains all cities with more than 50.000 inhabitants in 1962. The parameters pertain to this urban system was obtained by means of a nonlinear estimation procedure. As starting condition of the simulation of the urban process an almost homogeneous initial distribution (same city size for all cities) has been chosen. Furthermore, the system of settlements has been treated as closed; this means that emigration and immigration processes have been neglected. Therefore, the total population remains constant during this simulation, but a redistribution of population among the cities due to migration occurs. In Figure 5.5 assumption A is used in the dynamic simulation of (5). Under those conditions the urban system remains close to the initially chosen homogeneous distribution for a prolonged time period (almost 300 time units). When the hierarchy starts to develop the evolution accelerates and within a short period of about 40 time units as an intermediate state a rank-size distribution appears which finally ends up in two clusters of cities, namely a small cluster of dominating cities (Paris, Lion, Marseille) of almost the same size and a big cluster of smaller cities. A reason for this agglomeratory phase transition is caused by the shape of the attractiveness function favouring an optimal city size. 164 Günter Haag FIGURe 5.4 Size-Dependence of the Cities Attractiveness for the French Urban System in 1962 � � � � ����� � �������������� ���� ��������� � ����� �� � � � � � �� ���������������������������� Source: Haag [1994] In Figure 5.6 assumption B is used in the simulation. The hierarchical ordering process starts immediately and within a much shorter time interval of approximately 100 time units a Pareto distribution is approached. In this case, the rank-size distribution is the result of a dynamic self-organization process with the rank-size coefficient q as a stable attractor. In other words, the slope of the distribution is completely determined by the parameters of the settlement model (5) and therefore linked via the individual choice processes behind (5) to the micro level. The slope of the rank size distribution according Figure 5.6 corresponds to the observed distribution [Figure 5.1], if the agglomeration parameter is closed to its critical value κ ≈ 1/2. For κ > κ = 0,5 the homogeneous population distribution s s (the initial distribution) becomes unstable [Weidlich and Haag, 1987]. Therefore, the obtained distribution belongs to the case of SOC (self organized criticality). Agent-Based Modeling and SOC The model of the settlement system (5) can easily be improved by the introduction of subpopulations ( e.g. age groups, sex, different other population groups). New Models for Generating Power law Distributions 165 FIGURe 5.5 Simulation of an Urban System Using Assumption A ��� ����� ��� ��� ��� ��� � ��� ��� � ��� ����������������� ��� ��� ��� ��� ��� ���� κ=0.596; σ=0.188; v=0.001; N=25.6*106; l=78 FIGURe 5.6 Simulation of an Urban System Using Assumption B ��� ����� �� �� �� ��� � ��� � ��� ����������������� ��� ��� ��� ��� ��� ���� κ=0.596; σ=0.188; v=0.001; N=25.6*106; l=78 166 Günter Haag The interaction between the subpopulations may lead to more complicated dynamic structures such as limit cycles or even chaos [Haag, 1989]. If we introduce as many subpopulations as agents in the spatial system, equations for the modeling of a agent-based spatial simulation model are obtained L L                  i x t  (  1  ) x  1  f x exp u[ ( x) ( )]  exp[ ( ) ( )] it it 0 ijt jt i j u x   f x u x  0 jit it j i u x (11) j1  j 1 Where the different individuals are denoted by γ and  xxγ the probability that it agent γ decides for alternative I (settlement i) at time t. The different agents are  coupled together via their utility functions uγ ( x, κ γ), depending on the decisions i of all other agents and a set of control parameters κ. u ( x )  C  i   ln d ( ) i k  rank (12) k lim COORD(n)  COORD If one assumes that the attractiveness (12) of a city depends on the distances dγ to the hierarchically ordered centres κ (to be close to the centre is more i| k=rank attractive) again a Pareto distribution is obtained [Vogel, 2007] [Figure 5.7]. FIGURe 5.7 Size-Distribution of Settlement Places for a 1000x1000 Grid, for κ= 1 ������ ��� �� �������������� ����� � � ���� ��� ����������������������� �� � ������ ������ ������ ������ ������ ����� ���� ��� ���������������������� Deviations from the scale-free distribution for small values [see Figure 5.7] are related to the quadratic grid structure. However, only for a decreasing marginal New Models for Generating Power law Distributions 167 attractiveness with distance ( e.g. a logarithmic dependence) a scale free distribution was obtained. This finding corresponds to the results of subsection “The rank-size as a self-organization process. ” 5.3 Conclusions and Summary In Pumain [2006] a comprehensive overview of assumptions and models linked with the hierarchical differentiation of space are discussed. It is shown, as already mentioned that many different, partly not compatible assumptions may lead to systems exhibit rank-size behavior. Which of those many theories seems to be the most adequate one with respect to the many different aspects of urban systems and the behavior of its agents? Because of the dynamic spatial network structure of the urban system and the various organizational levels, nonlinear interactions, feedback effects, millions of involved agents, the social and political structure of the system we are dealing with, the urban system and its mathematical representation is certainly complex. As described in the early book of Weidlich and Haag [1983] these urban phenomena belong to the field of socio-political decisions of individual agents on the micro level and the consequential col ective, economic and abstract structures on the macro level of society. On the level of individual agents a complex mixture of fluctuating rational considerations, professional activities and emotional preferences and motivations finally merge into one of relatively few well demarcated resultant attitudes. The same fluctuating micro thoughts, emotions and experiences finally merging into an attitude may from time to time also lead to the transition from one attitude to another. The manifold of attitudes is an open one. Changes in the attitude space of an individual may lead to the decision to change the place of home or work, or both, or… Therefore, the dynamics of interacting populations on the macro level is linked via the nested decision behavior of its individuals to the micro level. Which one from the many different, in general not distinguishable attitudes may finally enforce the decision to move and how much of this decision is related to uncertainty and hazard? The evolution of a socio-economic system such as a system of settlements is not an autonomous process, but the result of human decisions, occurring over time as a broad stream of concurrent, unrelated and interrelated, individual or corporate choices, embedded in a spatial environment which is shaped by the choice processes of the individuals and hence also changing over time. In quantitative sociology a probabilistic description of the motion of macro variables e.g. the population sizes of cities proves to be adequate even when the details of the micro fluctuations of the system are unknown. In Haag [1989] a general coherent 168 Günter Haag and closed framework for the dynamic modeling of decision processes is presented, based on the master equation. Therefore, the mathematical description, the master equation, leads to the development of a highly dimensional probability distribution namely that a certain population distribution is realized. The most probable distribution of settlement sizes provides the input to construct the rank-size distribution. In so far the comparison of the computed rank-size distribution of an urban settlement model of a spatial system with the empirical rank-size distribution of that particular system can be used as a necessary but not sufficient test for the underlying model assumptions. One finding of such a comparison of the distributions based on the outcome of the simulation models (5) and (11) supports the assumption B of decreasing marginal attractiveness of a settlement rather than of a quadratic dependence of the attractiveness on size. The evolution of a network of settlements based on migration and the hypothesis of marginal decreasing attractiveness of settlements with size demonstrate that the rank-size distribution behaves as a stable attractor in urban growth. The distance (travel time) between the cities influence the spatial structure and the net growth effects beside the natural growth rates of the settlements. The mobility of the population is responsible for the speed of adjustment to any disequilibrium situation. The dynamic response of the rank-size coefficient is therefore influenced by the mobility. The steepness of the rank-size distribution, the rank-size coefficient in this self-aggrandizing urban model is finally determined by the spatial agglomeration parameter, namely the tendency of the population to cluster together in cities. For the existence of any rank-size distribution the homogeneous population distribution must become unstable. This requires that the agglomeration parameter is above a critical value and a phase transition appears. RISC-Processes and Their Weak Societal Protection Networks 6 Karl H. Müller 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 13 16 15 This article provides an overview on the special relations between RISC-processes and their societal control potentials, be it at the national or at the global level. At the outset, the intricate relations between RISC-processes, controls or governance and forecasting will be discussed in greater detail and the classical equivalence of explanation control and prediction will be effectively abolished.1 Finally, the second part of the article points to inherent vulnerabilities of globalized RISC-societies which lie clearly beyond any societal control. These concluding considerations point to the uneasy fact that the infrastructural constitution of RISC-societies which has been discussed at greater length in the second chapter of this book has an in-built Achilles heel or a necessary blind spot which makes today’s globalized societal ensemble open to very large-scale breakdowns. 6.1 Towards a Typology of RISC-Processes and Control Potentials In standard textbooks on Philosophy of Science or on scientific modeling one finds a statement that to explain a process P means also to be able to predict P and to control this particular phenomenon P under consideration: Ex (P) P  r (P) C (P)  This classical equivalence is no longer valid in the case of RISC-processes. Table 6.1 has been constructed in a way as to demonstrate that one finds a variety of RISC-processes with varying degrees of control, of forecasting capabilities and of prevention and damage control. In particular, one is confronted with RISC-processes which are perfectly predictable, although there is currently a lack of control and, consequently, an absence of prevention and damage control. A paradigmatic RISC-process in question are the distribution of incoming materials and meteorites from outer space where rare events can be predicted with high accuracy, although the controls on the underlying dynamics or prevention and damage controls are, at least at the present time, beyond societal capacities. Similarly, one finds RISC-processes whose underlying dynamics cannot be controlled, they cannot be predicted accurately but, nevertheless, there is a large potential for societal prevention and damage control which minimizes the impact of rare events. Again, the paradigmatic instance here comes from earthquakes which are beyond control in their tectonic dynamics and beyond forecasting 1 On this equivalence, see, for example, Casti, 1989a and 1992. 172 Karl H. Müller capacities, except for probability mappings of sites with high, medium and low probabilities for earthquakes. However, regulations for buildings and construction materials can be implemented in a way as to guarantee a minimal impact even in the case of very strong earthquakes. Table 6.1 introduces a wide variety of types of RISC-processes and their different recombinations of control dynamics, predictability and societal prevention and damage control. TAble 6.1 RISC-Processes and Control Potentials Prevention and Damage Control Not Possible Limited Degree High Degree P* NP* P* NP* P* NP* No Control Type I Type II Type III Type IV Type V Type VI on the RISC- mechanism limited Type VII Type VIII Type IX Type X Type XI Type XII Control of the RISC- mechanism Efficient Type XIII Type XIV Type XV Type XVI Type XVII Type Control of XVIII the RISC- mechanism P*: Accurate predictions possible NP*: No accurate prediction possible The most surprising fact about the eighteen types of Table 6.1 is that many of these types can be empirically observed. The types with the rarest occurrences are centered in the third line of Table 6.1 where it is assumed that an efficient control on the underlying dynamics has been established. Here, it can be argued that the types XIII to XVI are, most probably, empty because an efficient control of the underlying dynamics is incompatible with no or limited controls with respect top prevention or damages. Another striking feature emerges from Table 6.1, however. RISC-processes can change from one type to another. The history of financial markets shows, for example, that one can observe in the long run an oscillation between type X and type XII, i.e. , between regimes of high degrees of prevention and damage controls like the period from 1945 to the 1970s and periods of limited and insufficient degrees of control like in the period of the super-bubble from the 1980s onwards, leading to the crash of 2007ff. 6.2 The Infrastructural Constitution of Contemporary Societies as MR-Ensembles (Metabolism-Repair) To assess the potential impact of RISC-processes in a somewhat unconventional manner, a new multi-component framework for the Great Transformation I2 will be utilized3 where a multi-component ensemble is characterized by two main attributes, namely by metabolism and by repair. Not surprisingly, the resulting configuration can be described as MR-networks. The following main ingredients become necessary for an appropriate MR-specification. Starting from a national or, more appropriate, from a global level, one can construct a self-organizing complex of five interacting market networks, consisting of agriculture (M ), industry (M ), firm-related services (M ), household-related 1 2 3 services (M ) and a domain for waste-disposal and recycling (M ). Each of these 4 5 five network segments fulfills the following conditions. − First, inputs from other market segments or from the environment are transformed to new outputs, i.e. , to goods and services. − Second, the output from M will be purchased from other market segments i or from the market environment. − Third, a non-negative share of the monetary income from M is transferred i to the R-segments. It becomes quintessential, to characterize the concept of a market environment in a more precise manner. The first essential environmental complex for M consists in a repair-and maintenance segment which may be composed of five distinct components, namely of the three infrastructural networks for energy (R ), 1 information (R ) and transport (R ) as well as a household segment (R ) and a 2 3 4 sector for the education and training networks (R ). It should be easy, even at 5 first sight, to identify input-output relations between each of the market segments and the five repair and maintenance ensembles.4 The second environmental domain consists, not quite unexpected, of natural resources, land or, more generally, of the ecological settings as well as of the waste production, emissions, etc. which are produced in the course of the basic market metabolism. 2 On the first Great Transformation, see the second chapter of this book. 3 See, for example, Rosen, 1991 or Casti, 1986, 1988, 1989a,b, and 1992. 4 One could think on the relations between monetary flows between (M – M ) → energy 1 5 networks (R ), information networks (R ) or transport networks (R ) or between financial 1 2 3 contributions from (M – M ) → households or the national education and training systems 1 5 (R , R ). 4 5 174 Karl H. Müller Formally, the following three conditions must be fulfilled: − Condition : Each market segment receives at least one input from other 1 market segments or from the R-sector. − Condition : Each market segment produces at least one output. 2 − Condition : Each market segment has an output link with at least one of the 3 R-sectors. In the case of the ten-component MR-network specified above which, one must add, corresponds to the infrastructural constitution of RISC-societies, introduced in the second chapter of this book, conditions 1 to 3 are fulfilled even in a highly trivial manner. The basic formalism for MR-configurations assumes two types of metabolic processes, namely the transformation of natural resources into goods and services as well as the transformation of goods and services into monetary income. Formally, each of the five market segments transforms natural inputs Ω from the environment into monetary income flows Γ. f:Ω→Γ This market metabolism is taking place in two steps. First, as the production of goods and services Ξ g: Ω→ Ξ and, second, as a selling and distribution chain of the format h: Ξ → Γ To safeguard this market metabolism from disturbances, a repair system must be conceptualized as well which has two essential functions. On the one hand, the repair system must be able to adjust and regulate the market metabolism f R : г→ H(Ω, Γ) r On the other hand, the intensity of the repair and adaptation process can be formalized as ß :: H(Ω, Γ) → H(Γ, H(Ω, Γ)) r To set the basic MR-formalism into a “working mode,” the essential connections and exchanges between these ten network components have to be laid out in greater detail. With respect to the five market segments, the metabolic transformations can be analyzed in a conventional manner, relying, for example, on input-output tables and the like. The interesting and challenging point from the specifications so RISC-Processes and Their Weak Societal Protection Networks 175 far has to do with the role of the environment which enters into this scheme in an internal manner—the exchanges between sectors 1 to 4 and sector 5 as well as in an external way—the exchanges between all five sectors and the natural environment proper. With the inclusion of these dual exchanges one fulfills one of the core demands for an environmental and entropy-based economic analysis, set up by first and prominently by Nicholas Georgescu-Roegen. “Numerous elements of any production process are not commodities proper—tired workers, worn-out tools, and waste are normal outputs, while free goods are normal inputs.” [Georgescu-Roegen 1976:41]. With respect to the relations between the R -segments and the M -sectors, a j i seemingly difficult problem arises since these repair and adaptation mechanisms must be included within the two metabolic transformations g: Ω → Ξ and h: Ξ → Γ. At this point it must be sufficient to state that input-output exchanges can be observed between all five market with each of the five repair segments. Finally, the R-segments themselves are highly interconnected as well which can be easily seen from the multiplicity of exchanges and flows between two R-components like the ones between households and state, between households and the national system of education and training or between the state apparatus and R&D, etc. Thus, a densely connected MR-web can be identified for this specific societal configuration in which each of the ten segments is linked to the remaining nine domains in a multiplicity of ways. 6.3 The Potential for Very Large-Scale Involutions of RISC-Societies As a “Zero-Hypothesis”, a conjecture, born out of recent versions of modernization theories and Fukuyama’s “End of History” [1992], will be formulated which will act as an intuitively plausible developmental vision for densely connected networks within contemporary RISC-societies. Robustness-Theorem (RISC-Society Version): Due to the high network densities within and between the M-segments and the R-components, MR-networks are characterized by a very high degree of robustness to external or internal disturbances. Thus, the MR-configuration of contemporary RISC-societies has the quality of an coevolutionary stable complex. In light of the “Zero-Hypothesis,” two theorems will be proposed which run counter to the vision of coevolutionary stability, though. In order to get a proper understanding of these theorems, two new concepts must be introduced. 176 Karl H. Müller First, the notions of a re-establishable and non-re-establishable component refer to the following configuration. A network element M is re-establishable if and only if i there is an input relation to another network component M (j≠i) and the R , the j i repair component for M , is not entirely dependent on M . Otherwise, a network i i component must be qualified as non-re-establishable. A central component within an MR-ensemble is characterized, then, by two requirements. On the one hand, it must be a non-re-establishable element and on the other hand, the breakdown of the central component leads to an overall breakdown of the MR-ensemble as well. Under these conditions, the two theorems can be formulated as follows. − Theorem : An MR-network in all its possible connection patterns possesses 1 at least one non-re-establishable element. − Theorem : If an MR-configuration has only a single non-re-establishable 2 component, then this component will be the central one. Both theorems offer a counter-intuitive picture on developmental processes in highly connected networks and their evolutionary stable character. Two points must be stressed emphatically. The first consequence from the two theorems lies in a counter-intuitive insight on network densities. Growing interdependencies and network connectivities are not a safeguard from catastrophic disruptions. In other words, a densely connected MR-configuration is, contrary to the modernization-based “Zero-Hypothesis,” not coevolutionary stable. On the contrary, densely interconnected networks may even possess relatively small non-re-establishable units which, following Theorem , become the central ones for the entire ensemble.5 2 The second interesting implication has to do with the micro-constitution of the overall MR-configuration. Since each of the ten MR-components can be conceptualized, once again, as an MR-ensemble itself, consisting of smaller MR-units which, at the level of firms, are MR-systems themselves ...,6 a growing awareness should set in that contemporary RISC-societies are inherently unstable. It might well be the case that relatively small MR-units acquire the capacity to disrupt the entire MR-ensemble in an al -encompassing manner especially because the MR-network connectivities have become so dense. 5 An immediate counter-argument lies in the closed specification framework, developed so far. But this argument does not hold upon closer inspection since an appropriate metabolism-repair ensemble can be constructed for a national economy by taking into account its import-export relations and by postulating, then, the two theorems for an open economy-context. 6 Quite generally, M-R systems can be regarded as “self-similar” configurations, applicable to very different network levels, ranging from the global to national, regional or even to the firm levels themselves. Moreover, the core-domain for M-R ensembles lies in biological areas, namely in cells and cell-formations. RISC-Processes and Their Weak Societal Protection Networks 177 Consequently, the MR-theorems offer a radically alternative view on robustness and coevolutionary unstable configurations, beautifully summarized in the subsequent quotation from John L. Casti. “In order to be ‘resilient’ to unforeseen disturbances one would desire a system to consist of a large number of re-establishable components. On the other hand, the above results show that if only a small number of components are non-re-establishable, then there is a high likelihood that one of them will be a central component whose failure will destroy the entire industry. Thus, a system with a large number of re-establishable components will be able to survive many types of shocks and surprises, but there will be certain types of disturbances that will effectively cripple the whole system ... [Casti, 1989b:26] ... This last result has obvious implications for policies devoted to keeping every component of a system alive ...” [Casti, 1992:198]. Thus, network formations of the MR-type have an involution potential which cannot be diminished—it just can be shifted from one network type of a large number of non-re-establishable and isolated components—for example the capitalist world system in the 18th century—to today’s formations of a very large number of re-establishable components and a very small number of non-re-establishable, but central segments. Centered on any major global RISC-induced rare event, the following implications are very difficult to avoid. First, contemporary configurations within the economic sphere exhibit a small number of non-re-establishable segments in infrastructural areas like electricity, gas, oil or water which have been labeled as the MR-infrastructure. These components alone or in conjunction qualify for obvious reasons as a central component since they act as necessary pre-requirement for a smooth metabolic exchange and transformation within the main economic sectors or clusters. Second, RISC-induced problems become a very serious issue for the MR-infrastructure. There is clearly an above average amount of repair and adaptation work needed within the MR-infrastructure in order to avoid a partial or prolonged breakdown of the overall MR-ensemble. Why? Simply because the delivery chains within the MR-infrastructure have become highly sensitive to time and, thus, to failures in relation to time. Third, the increasing network densities through new production regimes like just in time-proliferation or lean, non-degenerate organizations, the reliance on multiple delivery chains or on firm networks have increased the robustness of the re-establishable segments with respect to a wide range of systemic failures. It is interesting and disturbing to note however, that RISC-induced failures of large magnitudes reveal clearly the vulnerable sides of the new production regimes both in their overall dependency on the MR-infrastructure as well as on a fail-free network of customers and clients in case of universal, global and non-time transferable problems, demanding effective solutions. 178 Karl H. Müller Fourth, contingency planning within an MR-ensemble would require, among other things, a complete revision of the organizational changes introduced over the last fifty years. Thus, successful contingency planning, too, is a very unlikely occurrence given the path dependencies and lock-ins with respect to changes of long-term developmental drifts within the MR-ensembles, regional, national or global. 6.4 ICT-Based Knowledge Pools as PM-Configurations (Program-Maintenance-Ensembles) From a RISC-perspective, it becomes highly instructive, once again, to point to a specific domain of vulnerability which lies in the diffusion of today’s complex information networks. The machine code basis of contemporary knowledge pools can be described and analyzed as a multi-component configuration, consisting, on the one hand, of program segments (P) and a maintenance part M which sets the common standards and measures like time, sizes. More specifically, M is responsible for a proper coordination and standardization of the program outputs. The subsequent specifications are aimed at the new machine code-bases in the regional, national or global knowledge pools which can be labeled as pools of machine-based programs. In order to facilitate the subsequent definitions, this specific pool will be qualified as program pool, for short. In a trivial manner, the program pool can be separated into various segments. In the present case, the program pool will be divided into those ten segments that have been identified for contemporary RISC-societies already. Thus, the overall knowledge base consists of a program pool for each of the five market segments and for each of the five repair components.7 For a single program pool component P , the i following points become of relevance. − First, inputs from other program domains like new program tools, new program languages, etc. will be transformed into new outputs within a specific program domain. − Second, the output of P is reproduced at least in some degree by other pro-i gram pools P as well. j − Third, a part of the output of P is connected with the M-segment. i M is a to be conceptualized as a very small segment of the program pool, organized and defined by all those program components necessary for the organization 7 To be more precise, the program pool is composed of ten program components and of ten practically identical PM-elements. RISC-Processes and Their Weak Societal Protection Networks 179 and synchronization of standards like time, space, lengths, weights, money etc. within the overall program pool. In a formal way, the following three conditions must be fulfilled for the interactions between program pool components and the M-element. − Condition : Each program segment must receive at least one input from the 1 internal or external program environment. − Condition : Each program segment P produces at least one output. 2 i − Condition : Each program segment is linked at least with one of its outputs 3 to the time maintenance segment M. It seems hardly necessary to stress the trivial fulfillment of each of the three conditions in the case of a PM-configuration. The basic formalism for the PM configurations postulates, once again, two types of metabolic processes, namely the transformation of external inputs into program tasks as well as the transformation of program tasks into an recognizable surface output. Formally, each of the ten program pools transforms external inputs Ω from the environment into an externally accessible program output Γ. f: Ω → Γ This program metabolism is taking place in two steps. First, as the production of internal program tasks Ξ g: Ω → Ξ and, second, as a task completion chain of the format h: Ξ → Γ To safeguard this program metabolism from disturbances, a maintenance system must be in operation which has two essential functions. On the one hand, the maintenance system must be able to adjust and regulate the program transformation f R : Γ → H(Ω, Γ) r On the other hand, the intensity of the maintenance adaptation can be formalized as ß : : H(Ω, Γ) → H(Γ, H(Ω, Γ)) r To set the basic PM-formalism into a working mode, the essential connections and exchanges between these program components have to be laid out in greater detail. With respect to the ten program pool segments, the metabolic transformations can be analyzed in a straightforward way in terms of program reproductions and program connections. With respect to the relations between the maintenance 180 Karl H. Müller domains and the program pools, the maintenance areas must be included within the two program transformations g: Ω → Ξ and h: Ξ → Γ. At this point it must be sufficient to state that the maintenance segment is included in a mission critical manner within the input-output transformations of the ten program pools. Finally, the PM-segment turns out to be highly standardized and uniform, being composed of synchronized elements distributed in an identical fashion throughout the knowledge bases. The PM-part has a unique format for the global knowledge society. In other words, time has become embedded in an identical fashion throughout the global TM-bases. 6.5 The Potential for RISC-Induced Involutions in the Program Pools Seen in this perspective, one is led to formulate another “Zero-Hypothesis” for contemporary ICT-based knowledge pools which may be seen as a corol ary to the modernization vision in the societal part. Robustness-Theorem (ICT-networks): Due to dense linkages, high replication rates and a huge amount of redundancies, ICT-knowledge pools are highly robust to external and internal disturbances. Thus, the machine-layer of the knowledge pools can be qualified as coevolutionary stable. Once again, two counter-intuitive theorems can be laid down which run opposite to this code-based stability vision. − Theorem : A PM-complex in all of its possible connectivity patterns possesses 1 a non-reproducible element. − Theorem : If a PM-complex has only one non-reproducible component, 2 then this element becomes the central one. Both theorems open up a self-similar pattern for the co-involution of machine-based knowledge pools, matching the pattern already identified for societal formations as a whole. Four special points are worth being emphasized. The first one is self-similar to an argument, developed for the societal network side already. An intensification of code densities and wide program distributions does not lead by itself to an overall stabilization in the machine-based knowledge pools. On the contrary, high reproduction rates of PM-components aggravates and intensifies the resulting repair and coordination efforts. Second, in all these instances of PM-transformations, changes in the societal networks require corresponding non-time transferable and effective adaptations in the machine code bases, too. Thus, many of the new societal coordination problems will turn out to be of a non-transferable nature since any change in RISC-Processes and Their Weak Societal Protection Networks 181 well-embedded standards like a currency change on a massive scale imposes a fixed temporal sequence of changes and adaptations which have to be undertaken by virtually all societal network actors. Third, RISC-induced problems should be considered as the first and probably a very spectacular case in a series of definitely new societal coordination problems, prompted by the growing dependencies on and the increasing embeddedness of the machine code program bases. These non-transferable coordination problems will require a new set of time-dependent or temporal organizational arrangements, capable of coping with non-transferable coordination challenges and with the necessity for, effective problem dissolutions. Fourth, these new coordination problems will lead to a radical re-definition and re-shaping of the notion of comparative regional or national advantages since flexibility and high adaptability in dealing with PM-transformations will become one of the major regional or national advantages within the RISC-societies of the future. 6.6 The Deceit of Reason The introduction to Part I already indicated that the linkage densities during the third stage of RISC-societies have increased dramatically and that the current RISC-phase is characterized, more and more, by the emergence of scale free networks which become embedded in a cumulative mode in the already existing network formations both in the economic and in the infrastructural domains. In combination with the two theorems presented in this article, an interesting trade-off can be observed which justifiably can run under the heading of Hegel’s deceit of reason. Figure 6.1 points to the disaster potential for RISC-societies during the second stage in which several random networks in transportation like railroads or in the automotive sector were built. Due to the overall network-structures, the effects of a rare event turned out to be local only. However, the local effects were usually heavy because the networks themselves had only a very limited substitution potential. 182 Karl H. Müller FIGURe 6.1 Local Disasters in Random Networks Turning to the current stage of RISC-societies, scale-free networks are characterized by a high substitution potential so that a local network failure, due to the overall network structure, can be compensated internally. However, scale-free networks have, although at very low probability levels, a potential for global failures if, for example, two or more central nodes are affected simultaneously. Figure 6.2 points to the critical configuration of scale-free networks with three central nodes. FIGURe 6.2 The Potential for Global Disasters in Scale-free Networks RISC-Processes and Their Weak Societal Protection Networks 183 Thus, seen from a long-term perspective, RISC-societies are drifting from a pattern of local disasters without substitution potentials during Stage II to a state with a considerable substitution potential at the local level and, at the same time, a non-negative probability for overall failures and global breakdowns. 6.7 Further Outlooks To conclude, both the network formations and the machine-code centered knowledge bases of contemporary RISC-societies are inherently unstable and vulnerable, due to their composition and due to the necessary existence of non-reproducible components within these networks or knowledge bases. Viewed in a global manner, intensifying the linkage densities of the regional, national or global RISC-configurations reproduces the potential for very large-scale failures and breakdowns, too. Zipf ’s Law in Labor Status Transitions: New Insights from Austrian Labor Market Data 7 Michael Schreiber 6 5 7 1 4 8 3 9 20 10 18 2 19 11 17 12 14 13 16 15 Motivated by discussions about competitive strategies of Europe that expect member states to implement flexicurity for employers and employees we present recent findings of research into RISC-processes and labor markets. We studied the transitions in the employment status in Austria for a period of six months in 2009 by analyzing monthly data according to three divisions among target groups: age, gender and education. It turned out that frequencies of changes in employment status followed a power law during these six months. Apparently, labor markets can be characterized as complex, adaptive and self-organizing systems, too. Moreover, the complexity of the status change networks was shown to be reducible by cut-off values that enable schematic classifications of the different groupings. 7.1 RISC-Processes and Labor Markets Rare incidents with strong consequences have been traditionally linked to financial markets or to the domain of income and wealth distribution. So far, labor markets have eschewed the attention of RISC-researchers. At first sight, labor markets stand in very close relations with labor market policies and with steering efforts by the state apparatus and by public labor market services which aim to push labor markets towards high levels of employment and, correspondingly, low levels of unemployment. Thus, labor markets seem to be situated very far from complex self-organizing systems and from an evolutionary perspective which are the usual arena for RISC-processes to emerge. At second glance, however, labor markets share an impressive amount of characteristics of a complex self-organizing system and should exhibit, thus, a variety of RISC-processes as well. In a first move we want to provide a short list of attributes for labor markets or, alternatively, for employment systems which belong to the domain of complex and self-organizing systems. Multiple decompositions: First, labor markets can be decomposed in a multiplicity of ways. Take, for example, a decomposition of labor markets into sectors. Then one can focus on the evolution of these employment sectors and describe this sectoral evolution as being headed in an irreversible movement to a complete marginalization of two traditional main sectors, namely of agriculture and industry, and to an al -encompassing service segment. Additionally, this sectoral evolution can be accompanied by a variety of different trends, ranging from a dualistic split between high-qualified and low-qualified activities to a more homogeneous rise of high quality activities especially in the services. 188 Michael Schreiber A second decomposition of labor markets can be undertaken in terms of employment clusters. Cluster arrangements are in the long run consistent with a multiplicity of professional shifts, ranging cross-nationally from a sharp rise of scientific-technical professions to a modest increase only, from a rapid expansion of state employees to an overall stagnation or even decline, from a severe downward movement in the number of skilled industrial workers to a slow decline. Moreover, the evolution of economic sectors and of clusters share a clear relation of complementarity since changes in the cluster-composition can be accompanied cross-nationally both by increases or decreases in specific sectors. A shift from an iron and steel-cluster to a cluster of constructing iron and steel-plants industries may imply both a decrease or an increase in the corresponding economic sectors, depending entirely on the performance of the other sector-components outside the specific clusters. Furthermore, even a single decomposition like the sectoral approach can be separated into a manifold of ways, from a three sector scheme or from the nine sector OECD-classification to the nineteen sectors of input/output-analysis or to the twenty-five economic sectors of the Austrian micro-census. In addition, two types of sectoral partitionings can be found. In the first instance, sectoral schemes fol ow a sub-class relation where, for example, the three sector classification can be subdivided in a way that the three sectoral building blocks of an employment system SE (primary, secondary, tertiary sector) can be subdivided into more distinct sectors so that the tertiary sector can be divided into banking and insurance, retail trade, personal services and the like. In this case, the fol owing subset-relation holds– SE  SE i,k, l i,k i.e. , the building blocks of the economic system E under the decomposition can k i be separated into different sets of sub-units . In the case of non-subset-divisions, 1 strict independence prevails. Take for example a sectoral division where a heavy emphasis is placed on natural resources and where a sector of natural resources and energy is defined by comprising mining and quarrying, crude petroleum and natural gas as well as, finally, electricity, gas and water, then the resulting new sector will have the following non-subset relation with the three sector scheme or the nine sectors of the OECD– SE  SE i,k,l i,k Since the sector of natural resources and energy comprises sector 2, parts of sector 3 and parts of sector 4 in the OECD-categorization, this particular building block can neither be reduced nor decomposed into the OECD scheme. Zipf’s law in labor Status Transitions 189 Thus, the building blocks for labor markets exhibit an astonishing openness and, above al , independence with respect to their basic constitutive elements. As a consequence, processes of structural adjustments, learning and adaptations occur in a multiplicity of ways for which, by necessity, a multitude of complementary and, thus, irreducible decompositions can be established. Aside from the multiplicity of decompositions labor markets exhibit all sorts of other characteristics of complex and self-organizing systems. For example, labor markets can be described with the following concepts from complexity and self-organization theory. − Multi-Component Systems: The overall direction of labor markets are determined by numerous interactions of many dispersed units acting in parallel. − No central steering unit: Labor markets proceed without efficient controls on interactions—controls are maintained in an inefficient manner and incapable of reaching central target values like a minimal value of unemployment within a short or even medium time range. − Multi-level-organization: Labor markets can be divided into many levels of organization and interactions, ranging from local, regional, national to global dimensions. Units at any given level typically serve as building blocks for constructing units at higher levels. − Permanent adaptation: At any level, these building blocks of labor markets are recombined and revised continually, leading, thus, to a permanent adaptation process far from equilibrium. − Coupled fitness landscapes: These adaptation processes are performed within coupled fitness landscapes, leading to an embedded process of feed-forward and feed-back between different levels of labor markets. − Economic niches: At any level, the arena in which labor markets operate is structured by niches and, consequently, by entrance and exit barriers. Moreover, there are no universal super-competitive units that can fill all niches. − Socio-technological systems as niche creator: Niches in labor markets are continually created by new socio-technological systems with new products and services and the very act of filling a niche provides new niches. − Technologies as niche annihilator: In turn, long established niches in labor markets are continually closed by the introduction of new technologies, leading, thus, to permanent processes of creative destructions. − Far from optimum: Labor markets operate, because of the permanent creation of new niches, far from an optimum or from a global attractor. − Power of the Local: The spatial distribution patterns of labor markets are not homogeneous, but highly concentrated in comparatively few areas. 190 Michael Schreiber − Temporality: The permanent adaptation processes are characterized, finally, by time lags and an uneven temporal distribution in the duration of adaptation patterns. In this manner, an evolutionary and self-organizing vision of labor markets and their trajectories has been formulated which serves as one of the background justifications for a deeper search for power laws and for Zipf-distributions in labor markets. 7.2 Probing Flexicurity A second motivation to look into possible power law-distribution in labor markets comes from recent political discussions at the European level on flexicurity. Flexicurity replaces traditional employment strategies from protecting existing jobs that had been popular in Europe in recent decades but were perceived as part of the problem. “Institutional changes affecting Europe’s labor markets over the last 25 years are a central reason for Europe’s poor labor market performance.”1 The flexicurity model started, first, in Denmark under a social democratic government, headed by Poul Nyrup Rasmussen during the 1990s. In its simplest form, it can be described as a recombination from flexible hiring and firing practices (flexibility for employers) with high benefits for the unemployed (security for the employees). In December 2007, the European Council adopted eight common principles of flexicurity in the following manner: 1) Flexicurity is designed to implement the main principles of the Lisbon Strategy. 2) Flexicurity, in addition to being committed to life-long learning, active labor market policies and a modern social welfare system, sees the need for flexible contractual arrangements. 3) Flexicurity needs to adapt to the different circumstances in each Member State. 4) Flexicurity needs to support open and inclusive labor markets which help to reintroduce inactive employees back into employment. 5) Flexicurity needs to involve the smooth transition between jobs by constantly up-grading employees’ skills and providing the necessary social protection in transition periods. 6) Flexicurity should promote both gender equality as well as considering means to reconcile work|life balance issues. 7) Flexicurity needs the support of the social partners. 1 See Siebert [1997] p. 39. Zipf’s law in labor Status Transitions 191 8) Flexicurity needs to involve a cost-effective distribution of resources which public budgets can sustain.2 From the description of labor markets as self-organizing and from the European discussion on flexicurity as a target for employment systems in general it becomes a highly relevant empirical question whether labor markets like the Austrian labor market exhibit processes and distributions which are characteristic for complex and self-organizing ensembles. 7.3 Employment Status Transition Frequencies The following sections present a particular heuristic interaction prototype developed to monitor frequencies of changes in employment status. In this article, we will outline the methods used and share some preliminary results. Our pilot study processed frequencies of employment status changes obtained from a database provided by the Austrian government for six month in 2009 [June, 2009, July, August, September, October, November, 2009]. We compared frequencies of employment status changes among groups segmented by sex, age and education. Three age groups were distinguished: • → below 25 years • → from 25 to 44 years • → over 45 years Segmentation according to education is problematic because of privacy concerns. We had to rely on synthetic data computed by official agencies which provide only a formal classification into seven levels of qualification in their database. Table 7.1 presents the most important twenty transitions within the Austrian labor market. 2 These principles have been retrieved from http://www.eurofound.europa.eu/areas/industri-alrelations/dictionary/definitions/FLEXICURITY.htm 192 Michael Schreiber TAble 7.1 Top Twenty Transitions Frequency Link Source Target 572 067 XO → Ol other out of labor force other out of labor force 332 952 Ub → XO employed other out of labor force 325 254 XO → Ub other out of labor force employed 201 005 NU → Al non subsidized employed unemployed 188 274 Al → NU unemployed non subsidized employed 185 424 XO → Al other out of labor force unemployed 147 074 Al → XO unemployed other out of labor force 131 273 ND → G* changed employer changed employer 109 284 Al → QU unemployed qualification 73 499 QU → Al qualification unemployed 47 548 Gb → Ub marginal employment employed 44 862 KG → Ol child benefits other out of labor force 41 558 Ub → Ub employed employed 37 964 Ub → Gb employed marginal employment 31 299 Ub → Re employed pension 29 748 Ub → KG employed child benefits 28 872 Re → Ol pension other out of labor force 25 421 KG → Ub child benefits employed 19 647 QU → Ub qualification employed 18 367 XO → QU other out of labor force qualification These top transitions are positioned in the following adjacency matrix together with less frequent changes of employment status. The headings of rows give the initial employment status of transitions. The headings of columns indicate the final states of these changes. Zipf’s law in labor Status Transitions 193 TAble 7.2 Adjacency Matrix of Transition Frequencies AL GB GU KG ND NU AL 14473 5699 188274 GB GU 4049 KG 8961 ND 131273 NU 201005 OL 185424 6296 PZ 3863 QU 73499 RE 2318 SB 6180 UB 37964 29748 OL PZ QU RE SB UB AL 147074 2259 109284 6481 5892 GB 47548 GU KG 44862 25421 ND NU OL 572067 13621 18367 10899 15764 325254 PZ 8909 12294 QU 16489 1964 19647 RE 28872 6941 SB 14938 40 972 16443 UB 332952 9811 1571 31299 16939 41558 We simplified this adjacency matrix and its corresponding graph below by modifications of identifiers. The label XO is replaced by OL as both indicate an other out of labor status. This replacement generates a self loop at node OL. The labels XB and B_ are replaced by SB because both are labels for self-employment. These replacements result in a self loop at node SB. Finally we replace “G*” by ND and obtain an isolated self loop at node ND which indicates a change of employer. 194 Michael Schreiber Sorting by net differences between the counts of source labels taken to be negative and counts of destination labels we condense the processed collection of transitions into the following table. The isolated self loop at the node ND reduces to a net value of 0. All other nodes have either positive or negative net counts. A marked decrease of non subsidized employment NU is partially covered by increases of subsidized employment GU. Self employment SB becomes more frequent which helps to compensate decreasing marginal employment GB and the counterintuitive difference between an increase in unemployment AL which is larger than the decrease in employment UB. FIGURe 7.1 Simplified Network of Transitions QU GB NU UB AL ND GU SB OL PZ RE KG Zipf’s law in labor Status Transitions 195 TAble 7.3 Net and Total Counts of Employment Labels Label Translation Net Total OL other out of labor force 18471 2313855 QU qualification 17663 240861 RE pension 10548 86810 GU subsidized employment 10424 18522 AL unemployed 5863 964735 SB self-employed 2958 80104 PZ military service 625 50757 ND changing employer 0 262546 UB employed -6736 996948 GB marginal employment -9584 85512 NU non subsidized employment -12731 389279 KG child benefits -37501 120987 Note that these net counts of status labels do not match the differences between status counts in June and November that are returned by a direct query for the 38 categories of employment status. The results of counting differences in unemployment by occurrences AL labels in transitions is 5863 for instance whereas the difference computed from direct queries is 29698. The following table groups differences consolidated according to three levels of classification. These nested classifications are given in the cells of the first column. The third level has been dropped unless it is different from the classification already specified at the second level of distinction. The individual labels encompassed by those nested classifications are given in the second column. The differences between the total of counts for June taken as negative and the total counts for November are computed in the final column. The top and bottom ends of this sorted tabulation are dominated by other states escaping these records in particular data not available labeled by KD and the default status generated for intervals without social security coverage LL. The category encompassing labels for secured out of gainful employment includes military service and PZ but is dominated by receivers of pensions RE. 196 Michael Schreiber TAble 7.4 Differences Between Direct Status Counts for June and November Classification Labels June November Difference other undetermined TO, KD 321212 450712 129500 labor Service Record Al 239967 269665 29698 unemployment labor Service Record D2, SC 60571 78861 18290 qualification other secured out of W1, W2, eD, eO, 2047333 2056291 8958 employment KG, KO, PZ, Re, lS, SG employment employed FU 6281 7915 1634 fragmented other marginal G1 142372 143632 1260 employment employment self lW, S1 412513 413344 831 employed employment employed Fb, Fl, FA, FF, FS 32924 25239 -7685 subsidized other pre AO 45388 37483 -7905 unemployment record employment employed be, le, AA, FD, 3152503 3129130 -23373 non subsidized SO other other out of AU, MK, MP, MS, 2413731 2243365 -170366 employment SV, ll Some of these differences could be discussed in more detail by queries of the mon_erwerb_chg database but this would force us to omit or reconstruct the distinction between levels of education which depends on statistical reconstructions of fused anonymized records. We thus proceed by using heuristic frequency cutoffs instead to generate simplified views of this transition system according to our selected distinctions of focus: gender, age and education. 7.4 Zipf Compressed Views for Nested Audiences Compressibility can be regarded as a proxy for entropy measures that indicate the amount of information in given sets of data elements. This idea can be applied to our data about employment status transitions by introducing a lossy compression according to relative frequency. Relative frequency is defined as the number of occurrences of a particular label n(t) divided by the total count of all occurrences of all labels W. f(t)=n(t)/W The sum of relative frequencies S(f ) can be required to match a desired constant of quality Q in the unit interval. 0C>0 Conversely, a similar compression can be defined by setting a threshold value T for the number of different kinds of transitions that are to be considered in the compressed view. f(1