PARTICIPATION IN AND CONTENTS OF TWO DUTCH POLITICAL PARTY DISCUSSION LISTS CARLO
ON THE INTERNET hagemann
Abstract
This study examines two political discussion lists affiliated with the Dutch political parties D66 and GroenLinks, and the possible contribution of these lists to democracy. Discussion in a democracy, it is argued, should be deliberative: based on rational argumentation, not monopolised by any particular individuals, and related to public affairs. The aim of this study is to assess the deliberativeness of political discussion lists on the Internet. To this end, the degree of contribution from participants to the lists is measured. The findings from this study suggest that, whereas the discussions on the lists as a whole were not monopolised by any individual, both lists had only a small number of very frequent participants. The contributions of members of this in-group were oriented towards one another. Opinions were mainly expressed without argumentation, and when argumentation was given it was predominantly based on common sense or on external sources such as newspapers or teletext. These two discussion lists do not live up to the expectations of furthering democracy. They may, however, eventually serve to fill the gap between the institutionalised public discussion that exists within the party elite and the uninstitutionalised, informal public discussion that transpires in other public and private domains.
Carlo Hagemann is Associate Professor in Department of Communication, University of Nijmegen, email:
c.hagemann@maw.kun.nl.
kO r-s
kO
fN
rvT o o
fN
"o >
A
IO
Introduction
In debates about the vitality of democracy, much attention has been given to the advantages the Internet may have for political discourse. Some people are enthusiastic about the opportunities the Internet offers, whereas others are sceptical (cf. Jankowski and Van Selm 2000). To the critics, politics on the Internet does not enhance democracy, but is merely an extension of a basically commercialised system of political communication. In the field of political communication major changes have taken place in the past years, and the role of the Internet in this change is evident. New initiatives, like websites for political parties and e-govern-ment, have been introduced.
One such initiative is the online or virtual discussion list. These online discussion lists promise a more democratic use of political communication because they provide citizens with access to the democratic process. Discussion lists allow nearly everyone an opportunity to discuss a wide range of issues with other people. Constraints of place or time are absent. The lists have often been compared to the coffee-houses that Habermas saw as an almost ideal public sphere and the critical rational discussion that was supposedly conducted in such establishments (cf. Sassi 2001, Habermas 1990).
In democratic theory, discussion is often seen as an essential component of democracy. To actually support democracy, discussion should be deliberative. The aim of this study is to assess the deliberativeness of political discussion lists on the Internet. To investigate this, a content analysis of two political discussion lists from the Netherlands was carried out. These discussion lists were maintained by two Dutch political parties, the environmentalist-radical left party (GroenLinks) and the social liberal party (D66). At the time of this study, both discussion lists had over 100 participants (GroenLinks = 127, and D66 = 132 in August 2001), and although the lists are primarily directed towards sympathisers of the respective parties, it is not required that discussants be party members or sympathisers in order to participate.
Internet and Democracy
In the past few years, much literature has examined ways in which the Internet might further democracy. When the Internet is discussed in terms of democracy, Habermas (1990; 1996) comes to the fore because he highlighted the importance of discussion in the public sphere. Habermas (1990) argued that the seeds of democracy could be found in these discussions. In the 18th century critical rational discussions were held among members of a privileged citizenry. In the 20th century a manufactured public sphere replaced this ideal public sphere, and instrumental thinking replaced critical reasoning. (Habermas 1990; Poster 1997, 218). Habermas avoids being pessimistic about the demise of a critical public sphere by directing attention to the informal public sphere. Habermas (1996) distinguishes those "spontaneous, unsubverted circuits of communication in a public sphere that are not programmed to reach decisions" from "institutionally structured political will-formation" (Habermas 1996, 485). The contribution of the informal public sphere is that it may bring issues to the fore that have been left out of the discussion in the formal public sphere. This can only be the case in the informal public sphere if
every issue is open for discussion and when the validity of arguments takes precedence over personal status of the participant (cf. Heilbron 1999).
The idea that discussion or deliberation constitutes the essence of democracy has also been advocated by political scientist Dryzek (2000). His criteria for authentic deliberation are less strict than those proposed by Habermas (1996). For him, the only requirement for authentic deliberation is that preferences be presented in a non-coercive fashion. These reflective preferences should influence collective outcomes to constitute authentic democracy (Dryzek 2000, 2). This concept is similar to Dahl's ideas about political participation in polyarchies (Dahl 1989). In Dahl's non-ideal democracies, only some people participate in the political process, and this participation is restricted to elections for the most part. Nevertheless, Dahl feels there is a need for nominal citizen knowledge because there can be no deliberative discussion in the informal public sphere without it. For him, the informal public sphere could never be the arena in which public opinion is formed.
With the Internet, optimistic theorists claim that the age of the city-states has returned and better than ever: citizens are able to gather in a virtual arena to discuss public affairs with each other (cf. Margolis and Resnick 2000). For such discussion, citizens do not have to be in physical proximity. The Internet actually achieves the ideal many-to-many communication of a forum and in some regards even exceeds it: "Compared to a meeting hall, the Internet has fewer barriers of space, time, and cost. Compared to a newspaper, the Internet allows for far greater participation in many-to-many communication" (Klein 2000, 216). The Internet can build forums in which public affairs are discussed (Klein 2000, 214), and thus it operates as an informal public sphere.
Although the link between discussion lists and governance may seem distant, discussion lists may play a role in the democratic process. Online discussion lists can put issues to the fore and play a role in opinion formation. These lists are especially suited for these tasks because they have three specific characteristics. In unmoderated discussion lists, any issue is open for discussion. No censors or editors block an issue from the discussion. This can give the discussion an unexpected turn and produce new insights that have not yet reached institutionalised politics. The second characteristic is that online discussions lack face-to-face contact. Because identities may be concealed, and because people do not necessarily know each other, the power of argumentation can outweigh the power of personal status.
A third characteristic that may make online discussions especially appropriate for deliberation is that the Internet offers a technique, which goes beyond traditional communication patterns. Where the press excels in one-to-many communication, coffee-houses in one-to-one communication, and city-states in many-to-many communication, the Internet facilitates a new sort of conversation: a "multilogue" (Shank 1993). The multilogue has, like discussions generally, a "many-to-many" setting. However, it goes beyond this, in the sense that it enables a multitude of interventions. To put it more concretely: whereas, ideally speaking, in discussions people allow each other to finish their contributions before responding, in the multilogue this is not the case. In a reply to a message, the responder is able to break in after every phrase and comment on it. These comments can be to the point, but can touch on a totally different subject as well and thus introduce a new "thread in the discussion." In a conventional discussion setting, such behav-
iour would lead to chaos, whereas on the Internet a multilogue, with its multiple layers, does not hinder such an exchange. Such interruptions and comments seem natural and constructive. The Internet thus offers new, and formerly unknown, opportunities for political discussion. The question remains whether these opportunities are actually realised.
Empirical Evidence from Other Studies
A considerable number of investigations have been carried out on political information on the Internet, varying from content analyses of websites and banners, to surveys concerning the effect of the Internet on the dominance of traditional parties (cf. McMillan 2000). Two empirical studies that address the question whether the Internet fulfils the democratic ideal, are those by Schneider (1996; 1997) and by Wilhelm (1999).
Schneider's (1996; 1997) starting point is Habermas' concept of public sphere and its emphasis critical rational discussion. Four dimensions identify this discussion: equality, diversity, reciprocity and quality (Schneider 1996, 379). With equality is meant the rate of participation in the public sphere. Diversity denotes the patterns of discussion and participation, and reciprocity is concerned with the opportunities to gain knowledge of the perspective of others and the degree to which these opportunities are realised (cf. Heilbron 1999; Ahuja and Carley 1998). Schneider links the dimension of quality to the criterion of critical-rational argumentation. The defining characteristics of critical-rational argumentation are that the merits of the argument prevail instead of the merits of the proponent. For an argument to be evaluated on the basis of its merits it must be of a certain quality (Schneider 1997, 39). How this quality is defined is not elaborated in detail by Schneider. In his study he operationalises quality as "the proportion of issues that are concerned with the abortion issue" (Schneider 1997, 75), in other words, the proportion of issues that are on topic.
Schneider investigated an Internet (Usenet) newsgroup focusing on the issue of abortion. He found that despite the democratic aspirations, newsgroups on the Internet seemed to score low on the aspect of equality. The discussion on the Internet was largely confined to a relatively small group of very active discussants (Schneider 1996). Schneider (1997) saw the quality of the discussion as one of the negative points, whereas diversity and reciprocity were not problematic in Internet discussions. Many diverging issues were discussed and participants actually interacted with each other.
Wilhelm's (1999) investigation addressed the deliberativeness of online political discussion. In this study Wilhelm refers to Fishkin (in Wilhelm 1999, 159) who characterises the contemporary political scene as democracy without much deliberation. In Fishkin's view a democracy is deliberative providing three conditions are met: "political messages of substance can be exchanged at length; there is opportunity to reflect on these messages; and the messages can be processed interactively, with opinions being tested against rival arguments" (Fishkin in Wilhelm 1999, 159-160). Wilhelm translated these conditions into three themes: a) the transaction between providing and seeking information, b) the exchange of opinions and the incorporation of others' viewpoints, and c) the in-group homogeneity of political opinion (Wilhelm 1999, 160-161).
Wilhelm examined these themes by means of a content analysis conducted on 57 political newsgroups on Usenet and on 14 political discussion lists on America Online's "Washington Connection." Wilhelm concluded that the newsgroups and "Washington Connection" were not very deliberative. Participants used the newsgroups mainly to amplify their own views instead of listening to others, posing questions or giving answers. Monologues overwhelmed the discussion. On the other hand, argumentation was provided in most of the contributions. Contributions of an "ad hominem" nature constituted only a small minority.
Wilhelm and Schneider both concluded that political communication via the Internet did not fully meet the expectations of discursive or deliberative democracy. Both authors claimed that active citizenship and critical rational political discussion remained an activity of an already engaged elite. The same elite that participates in party politics is active in political discussions on the Internet. The frequently mentioned possibilities that the Internet may provide for deliberation and discussion, were not achieved. For the dimension of reciprocity, which both authors addressed, the results differed. Schneider's (1997, 105) conclusion was that on the discussion list he analysed, the discussants acted very reciprocally. Wilhelm was more pessimistic about the reciprocity in the discussion: "Rather than listening to others, more often than not persons opposed to a seed message used it to amplify their own views" (Wilhelm 1999, 171). His pessimism was partly due to the stricter criteria he maintains for reciprocity. Where Schneider saw reciprocity as the mere fact of responding, Wilhelm was more interested in the content of the responses, whether it be a statement, a question or an answer. Only the latter two were in his view, really reciprocal.
The studies addressed the question of quality in different ways. Schneider concentrated on a single feature of the discussion: whether discussants stay on topic. Schneider concluded that quality was lacking (Schneider 1997, 105). Wilhelm operationalised quality as a function of the process, defining quality as giving arguments. He was more positive about the quality of the discussion lists, as almost three out of four contributions provided reasons to justify their statements (Wilhelm 1999, 173).
More studies of Internet discussion have been conducted. Jankowski and Van Selm (2000) investigated a political discussion involving senior citizens. On www.seniorweb.nl a direct online contact between politicians and seniors was available. Results from an online questionnaire, face-to-face interviews and content analysis suggested that, as in Schneider's research, a few participants monopolised the discussion. It was striking was that although senior citizens wrote messages to politicians, the politicians seldom responded directly to such queries. A majority of the politicians' messages were directed towards institutionalised organisations that participated and that the politicians had experiences with in "real life."
Research Questions
In the investigation carried out for this paper, the starting point is that a political discussion, in order to further democracy, has to be a free and open discussion on public affairs, based on rational argumentation. The central question is whether the two political newsgroups under study actually fulfil this deliberative role and
to what degree they do so. An investigation of deliberativeness implies that the following concepts have to be taken into account: openness and freedom of discussion, the degree to which public affairs are discussed, and the degree of rational argumentation.
In developing these aspects, the concepts of equality, diversity, reciprocity and quality were mainly borrowed from Schneider (1997); the concepts of "exchange of opinion," "providing ideas versus seeking information," and the "critical rational dimension" are from Wilhelm's (1999) study.
The following research questions are central to this study:
•	In what respect do the discussion lists from D66 and GroenLinks constitute a deliberative discussion? This question assesses the degree to which the discussion is monopolised by certain members or certain groups of members. Earlier studies (Schneider 1996, 1997) have shown that participation in Usenet discussion groups is heavily concentrated. The expectation is that this will be the case for these listserv lists as well and that though discussion may be free, it is not open to all participants. It is expected that this holds even more for the most frequent participants because they may have other motives to discuss other than deliberation.
•	To what degree do the discussion lists actually contain discussions? Central to this question is the degree of reciprocity in the discussion, and whether the "discussion" presents true pros and cons, or is merely a "get together" of like-minded people who agree on most of the public issues. Another perspective from which reciprocity is approached, is that of the multilogue. The question is whether the possibility of direct response is used. It is expected that the discussion lists show a large degree of reciprocity, simply because the Internet offers this possibility by facilitating a direct way of responding. Debate will be favoured over taking stances and amplifying one's own views.
•	To what degree do the discussion lists contain discussions about public affairs? This question addresses the contents of the discussion. To analyse the content of the discussion, the number and sort of issues discussed are investigated. Two groups of subjects are distinguished: public affairs versus all other subjects. As the discussion lists are explicitly presented as political, it is to be expected that the issues actually concern public affairs. A special interest in party-related affairs is to be expected inasmuch as the lists are explicitly announced as party-related.
•	To what degree are these discussions based on rational argumentation? Wilhelm (1999) concluded that newsgroups worked as a forum for amplifying views and not for deliberation. To determine whether this is the case for the listserv-groups under study, an assessment is made of the degree to which opinions are supported with some form of argumentation. It is beyond the scope of this study to assess the rationality of the argumentation. Any form of argumentation with some degree of reference to scientific, historical or philosophical grounding is seen as an effort towards rational argumentation. The expectation is that the D66 and GroenLinks lists contain a large amount of argumentation, mainly because discussion lists attract people who enjoy debating. The amount of sloganeering is expected to be small. This may be the case more for D66 than for GroenLinks because the former is generally known as a party of intellectuals, whereas GroenLinks has roots in communism, pacifism and protest campaigns.
Method
The data in this analysis of two listserv discussion lists stem from the SURFnet archive and were transferred to text files to make analysis possible. SURFnet is an institution that connects the networks of universities, colleges, research centres, academic hospitals and scientific libraries in the Netherlands to one another and to other networks in Europe and elsewhere (http://www.surfnet.nl). The discussion list archive consists of e-mailed texts and a table of contents, ordered by threads. A thread is a series of postings on the same subject. Threads are characterised by the same subject title, possibly preceded by "Re:" These tables of contents were used for an overview of the number of contributors and of the themes discussed in the discussion lists. In the content analysis, the recording unit was the individual posting. Quotations from former contributions were excluded from the analysis, but could provide the context for the analysed part. The unit of analysis was the discussion list as a whole.
Operationalisation
The research questions in this study require an elaborate operationalisation. To assess the freedom and openness of the discussion, insight into the patterns of participation is needed. The central concept studied here was that of concentration and monopolisation of the discussion by a small group of frequent participants. A discussion exhibits a large degree of concentration when a small number of participants are responsible for a large proportion of the messages. To assess this degree of concentration, data were collected from postings over the period of one year. An entropy coefficient1 of the discussion was calculated. In addition to these frequency counts, a network analysis of the patterns of participation was carried out in order to gain insight into the structure of the discussion between the participants. To conduct this network analysis, a crucial assumption was made. Although postings on a listserv list are principally directed towards all members, one can identify to what other member a response is directed by determining who wrote the quoted passage. The degree of concentration was then defined as the degree to which active participants respond to each other, as opposed to responding to less active participants. The subject titles were not sufficient. To determine the degree of concentration one needed to inspect the contents of the postings as well. This was done over the smaller time span of one month, August 2001, inasmuch as this sort of analysis would be more intensive than simply looking at the tables of contents. This month was selected because of the rather large number of postings that appeared on both lists (D66: N=189; GroenLinks: N=101). A random sample taken from one year or longer would not have been sufficient; the structure of a discussion list, with its responses and its threads, means that it needs to be studied as a whole.
A second question was whether a discussion list dealt with postings about public affairs. To answer this question, the individual topics discussed on the list were analysed. Individual messages were categorised according to the topics discussed. In the results section an indication is given on the topics that are identified as public affairs and those that are not.
A third element in the analysis was that of reciprocity as a part of the discussion. A discussion in which participants actually respond to each other has a greater capacity of contributing to deliberative democracy than a discussion in which the
participants only offer information and opinions without any real connection to the other participants (cf. Wilhelm 1999). Reciprocity was operationalised in three ways. The first, most global way, was comparing the number of responses with the number of seeds and stand-alone postings. This analysis is based on the tables of contents. The second way concerned the texts of the postings, looking at the kind of response present in the postings. Explicit "agrees" or "disagrees" were counted as an indicator of reciprocity, whereas statements and isolated opinions were not. This sequence of answers, "agrees" and "disagrees", and of seeds and responses could only be determined by looking at the individual postings. For this part of the analysis, the data from August 2001 were used. The third operationalisation of reciprocity is also based on the content of postings from this period. This was derived from the concept of "multilogue" that Shank (1993) introduced. In a multilogue anyone can participate in a discussion by "interrupting" another participant. How this works becomes visible when scrutinising e-mails in these discussion lists. Respondents are free to "interrupt" the line of thought put forward by the original discussant and give their views. This interruption of the discussion can be interpreted as an indication of reciprocity inasmuch as it is a very intense way of debating. In this study, a discussion was recorded as having a large degree of reciprocity if such interruptions took place, whereas postings where a response was given at the end or the beginning of the posting, or where no material from a previous discussant was quoted, were viewed as having a low degree of reciprocity.
As mentioned earlier, an assessment of the rationality of argumentation in the various discussions was beyond the scope of this study. Rather, any effort to support an opinion or suggestion, whether it was historical, ideological, psychological or philosophical in nature, counted as an argument. The validity of the reasoning was not assessed. In this respect, argumentation was basically the opposite of sloganeering. A first step in assessing argumentation was to divide contributions into two groups. One group consisted of opinions and suggestions (the latter being defined as the expression of a temporary idea, stated less strongly than an opinion), the other group consisted of all other postings, mostly consisting of information or questions. Opinions and suggestions were expected to contain argumentation. One posting could express an opinion as well as make a suggestion as well as provide information. Though only a few cases contained more than one of these categories in the same posting, these multiple responses still accounted for a total greater than 100%.
Results
A first glance at the D66 and GroenLinks discussion lists suggests that a lively discussion was present on both lists during the period of analysis from February 2000 to January, 2001. Although both lists had the same number of participants during the period of analysis (N = 132 on February 1, 2001), the D66 list was much more active than the GroenLinks list. The table of contents of the archive showed 3086 individual postings to the D66 list from February 2000 to February 2001. The GroenLinks archive had 996 postings during the same period. During this twelvemonth period, the D66 list contained 1236 threads of discussion. whereas the GroenLinks list had 313 threads, a mean of 2.5 contributions per thread for D66 en 3.2 for GroenLinks.
Concentration and Monopolisation
Concentration and monopolisation deals with the distribution of postings among participants. Concentration would be minimal if all participants posted the same number of messages. On a non-regulated discussion list with over 100 participants, this is not to be expected. However, there is a difference between a discussion list that is dominated by a small group of very active participants and one with participants who are more equal in number of contributions. To gain insight into the degree of concentration on both discussion lists, the concept of equality, as developed by Schneider (1996), was used. Figure 1 shows the level of concentration of postings by contributors in the D66 list.
Figure 1: Level of Concentration of Postings by Participants on the D66 Ddiscussion List (February 2000 - January 2001 )
Contributors (in cumulative order)
In this figure, the contributions of the various participants are presented in terms of percentages. This is done in a cumulative form, starting with the least active participants. The curved line represents the distribution of participation. The straight line represents what would be a situation of perfect equality where every participant contributed the same number of postings to the discussion. What becomes clear from the figure is that the top quartile of D66 participants is responsible for 89.8% of the contributions. For GroenLinks, this is approximately the same. The top quartile of GroenLinks participants is responsible for 81.4% of the contributions.2
To determine the degree of concentration in a number, an entropy coefficient is calculated. In a perfectly equal distribution the coefficient would reach the maximal score of 1. The entropy coefficient for D66 is .73 and for GroenLinks .77 (see Table 1). These coefficients are comparable to the coefficient of .67 Schneider calculated in his research of the newsgroup on abortion. The discussions on both Dutch lists, especially the GroenLinks list, tend even more toward equality. For D66 as well as for GroenLinks a considerable degree of the concentration can be attrib-
uted to two participants. For the D66 list, these two participants were responsible for 20% of the postings. If these very frequent discussants are omitted from the calculation, the entropy coefficient increases from .73 to .76 indicating a broader spread in participation. If the same is done for GroenLinks, the overall entropy coefficient of .77 increases to .79. On both lists these two front-runners, are followed by other frequent participants. If these are removed from the calculation of the entropy coefficient, the figure increases to .80 for D66 and .87 for GroenLinks.
Table 1: Entropy Coefficients for Participation (February 2000 - January 2001)
Entropy for:	D66	GroenLinks
Total list	0.726	0.771
List excl. top 2	0.762	0.800
List excl. top 6	0.802	0.865
Top 25%	0.677	0.798
Top 50%	0.791	0.804
List excl. single posters	0.779	0.800
Although the existence of two very frequent authors on both lists may suggest otherwise, on neither of the lists is the discussion monopolised in an extreme way. Complete equality is not to be expected where the discussion is unregulated.
Another perspective on the discussion can be gained by concentrating on its structure. The question is, who discusses with whom. If the most frequent discussants mainly discuss with each other, the debate is more monopolised than suggested by the overall figures. If less frequent participants are also addressed, one can speak of a more open discussion.
An analysis of the discussion structure requires a closer look at the postings to see to whom a response is directed. For an answer to the question whether responses of frequent authors are mainly directed to other frequent authors, the population of participants was split in two: the six most active participants, henceforth called the 'in-group', and the less active participants, called the 'out-group'. The degree of monopolisation was then defined in terms of the degree to which members of the in-group respond to other members of the in-group and the degree to which they extend the discussion to members of the out-group as well.
Tables 2 and 3 give an overview of the discussion structure for the in-group and the out-group for D66 and GroenLinks lists respectively. The tables show a great deal of activity by the six most frequent participants for both lists. On the D66 list, 69.8% (138/189) of all the postings came from the in-group. The largest group of messages from the D66 in-group (41.7%) was not in response to any other participant, but a stand-alone posting, or a seed to a thread. Responses to other in-group members accounted for 32.6% of the in-group postings. A response to out-group members was slightly less common (25.8%). On the other hand, out-group members were highly directed towards the in-group: 50.9% of the out-group postings were a response to a posting by an in-group member whereas only 19.3% of the out-group postings were in response to another participant of the out-group. Discussion on the D66 list is not restricted to the in-group, although this group plays a predominant role.
Table 2: Number of Postings to and from In-group and Out-group D66 (August 2001)			
		Posted by:	
	In-g roup N = 132	Out-Group N = 57	Total N = 189
In-g roup Response to: Out-group Seed/stand alone	32.6% 25.8% 41.7%	50.9% 19.3% 29.8%	38.1% 23.8% 38.1%
Total	100.0%	100.0%	100.0%
Chi-square = 5,663; p<0.1; df=2			
Table 3: Number of Postings to and from In-group and Out-group GroenLinks (August 2001)			
		Posted by:	
	In-group N=80	Out-Group N=21	Total N=101
In-group Response to: Out-group Seed/Stand alone	70.0% 11.3% 18.8%	42.9% 14.3% 42.9%	64.4% 11.9% 23.8%
Total	100.0%	100.0%	100.0%
Chi-square=6,101 ; p<0.05; df=2			
Table 3 shows the figures for GroenLinks. It can be seen that the in-group on this list monopolises the discussion more than the in-group on the D66 list does. A large majority (79.2%) of the entire postings stemmed from the in-group. Other members of the in-group were addressed by 70% of these postings. Many of the postings of the out-group were also directed to the in-group (42.9%). The picture of the GroenLinks discussion is that the in-group plays a major part in the discussion, more than could be seen from the overall entropy coefficients.
Reciprocity of the Discussion
A central element in discussions about discursive or deliberative democracy is the degree of reciprocity in the deliberation (cf. Dryzek 2000, 169). In this study, reciprocity is approached as a process of discussants responding to each other. The structure of the responses in the discussion was analysed in a global way by looking at the number of postings that were part of a thread, as opposed to the number of postings that were never responded to. In addition to this, the content of the postings was scrutinised with special attention paid to the degree of agreement among discussants, coded as "agrees" and "disagrees."
The number of postings that is part of a thread varies greatly during the periods of analysis. During the period February 2000 - January 2001 on the D66 and GroenLinks list only about half of the postings were part of discussion threads. For D66 this was 57%, for GroenLinks 47% of all postings. In August 2001, the threads were more dominant, as 85% of the GroenLinks postings, and 82% of those on the D66 list were part of a thread. Because of this large difference it is apparently not possible to assess the amount of reciprocity by merely looking at the postings and the threads.
Another look at reciprocity can be gained by assessing the degree to which the lists actually contain discussions or debates. A large number of the explicit indica-
tions of agreement or disagreement between the discussants can be considered as a first indicator for reciprocity. On both lists, the contributions apparently show neither agreement nor disagreement. Disagreement is explicitly expressed in 13.2% of the postings on the D66 list and 20.8% of those on the GroenLinks list. Explicit agreement is found even less often, 7.9% and 3% respectively. A great part of the discussion on the lists consists of passing on information about current events or current discussions. Clippings from external sources like (Internet-) newspapers and copies of teletext information were used for this purpose.
A final way of assessing the reciprocity of the discussion is deducted from the concept of "multiloguing" (Shank 1993). Participants on the GroenLinks list use this possibility of interrupting others significantly more than participants on the D66 list (22.7% versus 13.1%) do. The participants on the D66 list, in contrast, were more inclined to respond without quoting text than were the GroenLinks discussants (18.9% versus 9.9%).
Discussing Public Affairs
So far the discussion lists have been scrutinised by the quantity of messages in order to assess the equality and reciprocity evident on the two discussion lists. In this section the substance of the discussions is analysed in order to determine the way in which these discussion lists constitute a forum for discussing public affairs.
The main subjects that were addressed on the discussion lists during August 2001 were the Israel-Palestine conflict and the alleged fraud case of a GroenLinks Member of Parliament . The latter influenced the discussions not only on the GroenLinks list itself. The racism discussion on the D66 list was a direct result of this event, largely because of the ethnic origin of the MP. Another issue that drew attention on both lists was the Israel-Palestine conflict. Another subject that could be said to be at the heart of public affairs discussion was the discussion about aspects of democratic governance - procedures, legitimacy, and elections. An issue that only appeared on the D66 list involved gay marriages and whether a civil servant should be allowed to refuse to perform such marriage services. An excerpt from this discussion is illustrated below:
In my opinion the following question is central: What is predominant, what has priority: The disqualification of a conscientiously objecting civil servant who cannot do her job anymore or the exclusion of the gay couple that cannot be married by "every" civil servant? I think the interests of the civil servant are predominant because the gay couple has an alternative: choosing another civil servant (who is just as unknown to them.)3
An issue that only appeared on the GroenLinks list involved the anti-globalisation movement. After a participant asked for information about this movement, a discussion ensued as to whether the anti-globalisation movement should be supported by GroenLinks:
Author # 1: Who can tell me more about the aim of the anti-globalisation movement? I also want to know what position GroenLinks takes. I read in the Saturday newspaper that the Green Party in Germany is split on the issue of supporting to this new protest movement.
Author # 2: I don't understand this reservation either We all have the same political preference, don't we? We are leftist.
The attention for public affairs is, opposite to what may be expected on these kinds of discussion lists, not restricted to party-related subjects. Although participants mention their own party in about 10% of the postings, this did not dominate the discussion. References to other parties were almost absent on the GroenLinks list. On the D66 list other parties were mentioned in a considerable amount (12.1%) of the postings. The above-mentioned fraud issue accounted for the greatest part of these references to other parties. When participants discussed their own party, they did not agree or disagree more than on other subjects, although they sometimes took a critical stand towards politicians:
Why don't we hear a thing from our yuppie celebrity, the polyglot Lousewies van der Laan [chairwoman of D66 in the European Parliament]? It is about social money from the European Social Fund, isn't it? Or is she - after she, in a television interview, bravely banished all women from the future parliamentary party - still busy trying to emasculate the Pope in Rome?
The majority of these issues can be considered related to public affairs, and on both of the discussion lists there is, indeed, much discussion about such matters. Some are party related, but the majority is on a more general level. The discussion on the fraud case may be seen as a party issue, although it was widely covered by the media.
Rational Argumentation
For a discussion to contribute to democratic practices, many theorists regard it as essential that discussion be rational or at least contains rational elements (cf. Sassi 2000; Habermas 1990; 1996). A discussion that consists primarily of sloganeering or merely the stating of positions does not contribute to a rational determination of opinion. It is beyond the scope of this paper to assess the validity of the argumentation given. Instead, for this study effort was expended to ascertain whether opinions uttered in the discussions were supported by argumentation. On the GroenLinks list a total of 33% of the messages contained an opinion (22%) or a suggestion (12%), but in only 9% of those messages was some form of argumentation given. Such argumentation was based on a variety of references: historical, philosophical or scientific findings, religious references, logical thinking, and external sources like newspaper clippings. On the D66 list a total of 40% of the messages contained opinions (23%) or suggestions (18%). Argumentation for these opinions, however, was given in only 15% of these postings.
It can be concluded that a fair amount of the discussion rests on opinions without argumentation. This does not mean that the discussion list consists of no more than slogans and unsubstantiated opinions. Many postings ask for information, for suggestions or for opinions, and many postings give information. What can be said, though, is that the opinions that are expressed are not firmly based on rational argumentation.
Discussion and Conclusions
The Dutch political parties D66 and GroenLinks have active discussion lists on their respective websites. During the period of study, between 120 and 150 participants contributed to the discussions on these lists. A large number of these persons were willing to participate actively in the various discussions held on the lists. The view commonly expressed about discussion lists, which most participants lurk
rather than contribute, was not justified for these two lists during the period of study. Nor is the view upheld by this study that discussion lists are burdened with flame wars and spam. The question remains, however, whether such discussion lists contribute to deliberative democracy. One condition is that such lists should be "free and open" - that they are accessible to everyone who wants to participate, and that there are no principal constraints regarding the standpoints or subject discussed, providing discussants remain on-topic. In addition, it is expected that the discussions reflect a certain amount of reciprocity and are based on rational argumentation.
One feature of a free and open discussion is that there is an equal distribution of contributions among participants - that the discussion is not monopolised by a small group of participants. The concentration of the contributions to the discussions on these two lists, as a whole, was not higher than that found in Schneider's study of the discussion list on abortion (Schneider, 1996). However, on both the D66 and GroenLinks lists more than 20% of the contributions were posted by two participants. In addition to those mega-discussants, there was a more evenly distributed group of frequent participants. These participants were, to a large extent, reacting to postings from each other. An in-group was, in other words, present on both discussion lists, albeit in greater degree for GroenLinks than for D66. On the other hand, it was not the case that some participants were systematically neglected. The equality criterion may be fulfilled for the list as a whole, but not without qualification. A small in-group leads the discussion whereas a larger out-group at best, reads the contributions.
Despite the fact that there was interaction between participants on the lists, it would be incorrect to draw the conclusion that there was a great amount of reciprocity on the lists. There was limited expression of disagreement, and even less of explicit agreement; there was also only limited use of multiloguing. Perhaps, however, the operationalisation of reciprocity should be reconsidered; a few suggestions are made in this direction below.
Public affairs were, indeed, discussed on the discussion lists. The issues that were debated also appeared on the political agenda. Although every participant would stress his favourite topics, it could be said that the discussions generally were related to current affairs. There were few abusive messages or flames. What may seem striking is that contributions on these political party discussion lists seldom mentioned the name of the host party. Most of the discussions were related to current affairs that were also covered by the media. In this respect, the role of political parties seems to be taken for granted.
The rational character of both lists was limited. Opinions were usually expressed without argumentation. If argumentation was given, it is mostly based on sound thinking, or on external information like newspaper clippings or teletext messages. One can say that discussion on current public issues is actually present on the discussion lists. Whether this discussion is an asset to deliberative democracy is less evident. It may be considered small at first glance, mainly because opinions are not very often based on argumentation, and this argumentation is mainly based on sound thinking, as opposed to contributions grounded in historical, scientific or philosophical ideas. However, the equality of participants on the discussion lists is, despite some concentration of a small in-group of participants, not particularly limited as compared to earlier studies (Schneider 1996). It is striking that women are nearly absent on this political forum.
The real restriction to being an asset to deliberative democracy may be the distance to actual governance. Political discussions may be considered without value if it does not contribute to formation of political will. For D66 a member of parliament participated in the discussions on a regular basis. For GroenLinks this is not the case. Substantial influence on politics, by influencing existing or new policies, goes further than that, of course. It may be that the discussion lists serve mainly as a "laboratory" or exercise area for more influential discussion settings.
It would be presumptuous to conclude on the basis of this investigation that political discussion lists like those used by D66 and GroenLinks contribute to deliberative democracy. The theoretical line from Habermas (1990; 1996) to the actual discussion lists that is drawn by Sassi (2001) merits qualification. An ideal public sphere has never existed and is not likely to be built on the Internet. The notion of an informal public sphere and how this contributes to democratic will formation, remains implicit. The distance between non-politicians and government is too large to expect an influence from these discussions on the actual formation of political will. On the other hand, Dryzek (2000) considers deliberation important in itself, regardless of the content. Slogans or even gossip may play an important role, if it brings new issues to the fore. The problem with this definition is that professional politicians are more easily persuaded by an elaborate rational argumentation than by slogans or gossip.
In addition to that, one has to admit that the concept of reciprocity has been operationalised in this study in a very narrow manner. This resulted in the conclusion that the discussion lists lacked reciprocity. However, on face value, the lists show a very lively pattern of interaction. On reflection, reciprocity should be studied in a more qualitative way, concentrating on the question "who says what to whom." This means that a more in-depth analysis of the question-and-answer/ agreement-disagreement structure would have to be conducted.
In future studies of discussion lists attention should also be given to the participants themselves and their motives for participation. In addition to that it has to be assessed how the participants process information gained from the discussion, and in what other discussions or party activities they participate in addition to online discussion lists? Another crucial question to be answered is, what goals the political parties pursue by hosting discussion lists and how the occasional participating MP processes the input he or she gets from the opinions contributed on the list.
Notes:
1. Formula for calculating degree of entropy in discussion lists (cf. Schneider 1997, 83):
N
a s, log 2 (1/ s, )
— i=i
Entropy =
log 2 N
where Si is the percentage of total participation of the ith participant.
The entropy coefficient reaches its maximum of 1 when there is perfect equality. This is the situation where all participants post the same number of messages. The minimal score of 0 would be reached if one participant were responsible for all the postings, and the rest did not post at all. This score cannot be reached in the current study, because only active participants (at least one posting) are part of the analysis. For D66 the minimal score would be 0.116; for GroenLinks it would be 0.129.
2. A figure for the level of concentration of postings for GroenLinks is not included, because the
figures would be similar. This figure can be seen at the personal Web site of the author http://
baserv.uci.kun.nl/~carhage .
3. All quotations from the discussion lists have been translated from Dutch to English by the
author.
References:
Ahuja, Manju K. and Kathleen M. Carley. 1998. Network Structure in Virtual Organizations. Journal of Computer Mediated Communication 3. Available at: http://www.ascusc.org/jcmc/vol3/ issue4/ahuja.html.
Dahl, Robert A. 1989. Democracy and its Critics. New Haven, Connecticut: Yale University Press.
Dryzek, John S. 2000. Deliberative Democracy and Beyond. Liberals, Critics, Contestations. Oxford: Oxford University Press.
Garton, Laura, Caroline Haythornthwaite and Barry Wellman. 1997. Studying Online Social Networks. Journal of Computer Mediated Communication 3. Available at: http:// www.ascusc.org/jcmc/vol3/issue1/garton.html.
Habermas, Jürgen. 1990. Strukturwandel der Öffentlichkeit . Frankfurt: Suhrkamp.
Habermas, Jürgen. 1996. Between Facts and Norms. Contributions to a Discourse Theory of Law and Democracy. W. Rehg Trans. Cambridge, Ma.: MIT Press.
Heilbron, Bart 1999. Internet als Openbaarheid. Vier Dimensies van de Publieke Sfeer binnen Drie Nederlandse Nieuwsgroepen. Unpublished master's thesis, University of Nijmegen.
Jankowski, Nicholas W. and Martine van Selm. 2000. The promise and Practice of Public Debate in Cyberspace. In K. L. Hacker and J. van Dijk (eds.), Digital Democracy: Issues of Theory and Practice, 149-165. London: Sage.
Klein, Hans K. 2000. Tocqueville in Cyberspace: Using the Internet for Citizen Associations. Information Society 15, 213-220.
Margolis, Michael and David Resnick. 2000. Politics as Usual. The Cyberspace "Revolution." Thousand Oaks: Sage.
McMillan, Sally J. 2000. The Microscope and the Moving Target: The Challenge of Applying Content Analysis to the World Wide Web. Journalism & Mass Communication Quarterly 77, 80-98.
Nixon, Paul and Hans Johansson. 1999. Transparency through Technology: The Internet and Political Parties. In B.N. Hague and B.D. Loader (eds.), Digital Democracy Discourse and Decision Making in the Information Age, 135-153. London: Routledge.
Poster, Mark 1997. Cyberdemocracy: Internet and the Public Sphere. In D. Holmes (ed.), Virtual Politics: Identity and Community in Cyberspace, 212-228. London: Sage.
Richard, Elizabeth. 1999. Tools of Governance. In B.N. Hague and B.D. Loader (eds.), Digital Democracy Discourse and Decision Making in the Information Age, 73-115. London: Routledge.
Sassi, Sinikka. 2001. The Transformation of the Public Sphere. In B. Axford and R. Huggins (eds.), New Media and Politics, 89-108. London: Sage.
Schneider, Steven M. 1996. Creating a Democratic Public Sphere through Political Discussion. A Case Study of Abortion Conversation on the Internet. Social Science Computer Review 14, 373-393.
Schneider, Steven M. 1997. Expanding the Public Sphere through Computer-Mediated
Communication: Political Discussion about Abortion in a Usenet Newsgroup. Available at: http://www.sunyit.edu/~steve.
Shank, Gary 1993. Abductive Multiloguing. The Semiotic Dynamics of Navigating the Net. The Arachnet Electronic Journal on Virtual Culture 1. Available at: http://www.infomotions.com/ serials/aejvc/aejvc-v1 n01 -shank-abductive.txt.
Webster, Frank. 1995. Theories of the Information Society. London: Routledge.