Curriculum Vita‎ > ‎

Computer-network based democracy

Computer-network based democracy:
Scientific communication as a basis for governance (Note 1)


There have been many proposals for electronic democracy. These proposals have merely replaced paper ballots with electronic mail. The ideal, however, is a discoursive democracy, characterized by direct discussion and negotiation among equals. We show how electronic conferencing makes this possible.

A democratic system must incorporate both protections for individuals and mechanisms for coordination. Access to information is crucial for democratic decision making. If people withhold information, however, then even free distribution of available information will not ensure democracy. One problem is that people may be punished if they speak in opposition to power holders. More often, persons censor themselves because they fear that speaking out will damage their career possibilities. Finally, people may prefer to avoid the duplication of effort that occurs when many people write comments simultaneously in response to an event or statement. These problems of protection and coordination must be solved before direct democracy is feasible with large populations.


Stodolsky, D. S. (2002). Computer-network based democracy: Scientific communication as a basis for governance. Proceedings of the 3rd International Workshop on Knowledge Management in e-Government, 7, 127-137.


1. Introduction

Every system of social organization is limited by its communication infrastructure. Representative (elective-indirect) democracy, is based upon the movement of persons. Plebiscitary (direct vote) democracy is based upon the movement of paper documents. Computational (mass discoursive) democracy, is based upon the transmission of electronic signals through computer networks. There have been many proposals for "electronic democracy", that is, instant plebiscites, utilizing information technology's communication capabilities. Such proposals, however, merely move procedures based on the marking of paper ballots into the information technology infrastructure without dealing with contradictions, such as agenda setting, that are inherent in plebiscitary democracy.

While discoursive democracy, permitting direct discussion and negotiation, is usually viewed as ideal, no proposals have been made previously to overcome its limited applicability. Discoursive democracy can be extended to large populations if computational capabilities of information technology elements are fully utilized. Recent developments in computer networking and cryptography have created possibilities for democracy based upon a computational infrastructure. Citizens in such a political system could issue protected communications directly from their personal computers. These communications would be the object of remote criticism, as are scientific articles. Readers would have available automatic article selecting software that could be trained to identify important articles based upon reputations of authors and critics. Automatic negotiating capabilities would guarantee solicitation of alternative viewpoints. Just as public opinion typically guides political actors in a representative democracy, decisions reached discoursively could dominate formal decision taking, making the composition and organization of formal political bodies less important. The simplification of political processes and the greater transparency of decision making in a computational democracy could vastly increase citizen participation.

2. Plan of the Paper

First, ways of strengthen personal integrity using unobservability mechanisms are discussed. Such mechanisms block negative consequences for a source by hiding the person's identity. We outline a way of doing this based upon secure pseudonyms, which support responsible behavior. Such pseudonyms permit the development of reputation. A review process, based upon routinely publicized evaluations of articles, is then introduced. Evaluative judgments of an article by one person can be available to all other potential receivers of that article immediately on a computer network. Evaluations and reputations can then be used to automatically select articles worth reading. Finally, we outline how evaluations can also be used to automatically generate a consensus that selects new authors to write articles.

3. Anonymity: Potential and Problems

One way to protect expression of opinion is to strengthen personal integrity with an anonymity shield. Many text-based conferencing systems (Note 2) permit anonymous contributions. This often leads to irresponsible behavior [9]. "Anonymity breeds irresponsibility," was the conclusion of one system administrator [4]. Abuse of anonymity has also been a long standing problem with scientific journals [2]. The current peer review system has been described as one in which power relations have become dominant [3]. Inappropriately used, anonymity can generate power differentials as well as irresponsible behavior [2]. Another important limitation of protection based upon anonymity is the inability to reward persons individually for specific acts. This is a recognized problem with scientific peer review.

3.1. Pseudonymity

One solution is reputation-preserving anonymity, or pseudonymity, which overcomes many of the problems with traditional journals, while ensuring individual integrity. If people are limited to one and only one pseudonym, responsible behavior can be expected. The person must protect their pseudonym from developing a bad reputation or others will not select articles or use judgments issued under that pseudonym. On the other hand, articles under a pseudonym with a good reputation will be read more often and judgments as to what is worth reading under that pseudonym will have a strong influence on dissemination of articles.

An often raised objection to this approach is that even if an author can not be traced through the system because of formal protections, the writing style and similar factors can be used to identify an author. In the system described here, judgments similar to votes play an important role, and these can provide complete protection even in relatively small groups. The system automatically collects judgments of articles after they are read and transmits them in a standard format, a type of electronic voting slip. Completely protected judgments can then be used by others to automatically select articles worth reading. We assume that it is not feasible to read all articles.

3.2. Public versus Protected Communications

An analogy between the proposed system and anonymous voting in parliamentary meetings can be extended, although it is not completely correct. In such meetings, communication occurs on two levels, public and anonymous. Substantive motions, procedural motions, and discussions occur on the public level: There is no attempt to hide source identity. Voting, however, is typically anonymous. Similarly, with the proposed system articles and judgments directed against them can be considered as operating on two different levels. Especially in a smaller group, it may be difficult to effectively hide the source of an article because of stylistic features. Judgments in a standardized form, however, have no such features and therefore offer significantly greater protection for source identity. Like anonymous votes, they can play a crucial role in decision making.

However, unlike anonymous votes, pseudonymous judgments can play an important role in reputation development. For instance, a person could develop a good reputation purely on the basis of judgmental responses: Those responses are highly protected. This would ensure that any articles later contributed would immediately come to the attention of other group members, thereby maximizing the article's influence potential. With the pseudonym system, there is also much greater flexibility in the treatment of articles. One extreme could be to have them signed by their authors using their public names. Another would be to have articles sent first to human or machine editors, who would remove stylistic features that could reveal authorship, translate them to another language, or even rewrite them for clarity. Pseudonymous communication, then, offers many more possibilities for finding an effective balance between protection and accountability for authors. It also offers significantly more information to readers --actually their programmed article sorting systems--, who must decide which articles to read.

4. Basic Organization of the System

In its simplest form, the system includes untraceable mail and digital signature capabilities. By untraceable mail we mean that a message can not be traced back to its sender by physical means or by analysis of the information transmitted with the message. An ideal broadcast system would have such a physical characteristic. In practical systems, a ring topology network can transmit untraceable mail at 25% efficiency as compared to normal mail. The author's identity can be unconditionally secure, that is, resistant to infinite computational power. Then, finding the source of the message requires cooperation of all parties except for the one being traced [1]. A less secure, but readily available, system is the public-access telephone network. Many data networks can also provide adequate security for short connection times.

The digital pseudonym (Note 3) is required to be untraceable and unforgeable. A one-to-one mapping between persons and pseudonyms is required. This can be implemented with public-key cryptography (Note 4) using an independent registrar or is-a-person organization. This organization is the only one that can engage in pseudonym creation. Its interaction with a potential user permits the authorization for creation of a pseudonym to be issued to the user. The user then, at a later time, returns by untraceable mail the actual pseudonym to be used. This pseudonym serves as that individual's public key in a digital signature system. Messages decrypted with that key could only have been sent by that individual. Persons must be physically identified to obtain an authorization, thus each person can acquire one and only one unforgeable pseudonym.

Readers would train their computer systems as to which pseudonyms merited attention merely by giving evaluative responses to articles. Both the reputation of the author of a article and the reputations of previous readers of that article (assuming they offered judgmental responses to it) would be used to automatically rank the article in priority. This is meant to duplicate, in a more rigorous manner, the way we use recommendations of friends and colleagues to decide what is worth reading.

A common misconception about using pseudonyms is that the benefits would be short-lived since once a pseudonym's reputation had been established it would function just as a real name, prejudicing reader acceptance of articles and reactions to them. It is true that only an anonymous message system guarantees the evaluation of message content without any influence of previous messages from that author. However, such a system also offers the reader little basis for selecting articles to be read. In a properly functioning pseudonym-based system, reputation information gives an unbiased estimate of whether a new article is worth reading. This estimate is based partly upon the content of previous articles from the same author. In summary, a system based upon anonymous messages treats each message equally, while a system based upon pseudonymous messages treats each author equally.

5. Differential Competence and Reputation Management

The possibility of differing levels of competence in different subject areas can be accounted for by allowing persons to have a different pseudonym in each separate conference or journal. By use of a credential mechanism [1], expertise developed in one conference can be transferred to another without any loss of security -- that is, without release of information that would permit the association of the different pseudonyms. If one of the names used was the name by which the person was known publicly, this mechanism could be used to show the person's reputation freely.

An essential feature of the credential mechanism is the ability to move reputational information from one name to another untraceable. Given this ability, even positive identification of the author of a given message would not compromise the overall functioning of the system, since all pseudonyms could be changed without a loss of reputational information. For instance, such a procedure could operate very similarly to the double-blind peer review used by many scientific journals. While an article was being evaluated, authorship could remain hidden. Later, the author could claim the article and even the referees could identify themselves publicly. If this was followed by an immediate change of pseudonyms, then the next article from that same author could be evaluated in an unbiased manner, since the association between the article with the publicly identified author and the new article would be untraceable. The credit for producing that publicly identified article would be available to the author, however, thus ensuring that the new article was widely distributed. We assume here that an author could not be identified by stylistic characteristics. In any case, such attempts at identification would be risky, since an imposter could easily adopt the style of another person. A limited number of authors with "good" reputations might also make it possible to "guess" someone's identity. This is a general problem, which can only be solved by ensuring that the details associated with a given identity do not reduce the pool of possible individuals to such a degree that protection is lost. Changing all pseudonyms simultaneously, whenever any need changing, is a way to minimize this problem.

6. Review Messages

When an individual reads an article and makes a judgment of it, that judgment can be signed by the reader and broadcast to other potential readers. These judgments of an article can be used by those who have not yet read an article to rank it in priority. The standard format of the review message permits the user's program to compute the probable importance of a given article and automatically schedule articles for attention. A Bayesian estimation model can be used to combine the information about the author with the judgments of the previous readers. The user's own judgment upon reading an article can then be used as a basis for revising the probabilities in the model parameters that describe each person's judgmental competence and competence as an author.

6. 1. Evaluative Dimensions

Another elaboration of the basic system permits judgments to be given on multiple dimensions. These judgments establish different types of formal relations between articles. For instance, a scientific paper after having been judged relevant is most likely to be accepted for publication if it meets three criteria [2]. First, it should be sound. The author must have employed reliable data, drawn valid conclusions, and committed no flaws in logic. It should also be original. Finally, it should be significant, meaning that it should contain some new perspective or observation of potential importance. Judgments on these dimensions could be combined to decide whether an article is worth reading.

While evaluative messages could give quantitative responses on various dimensions, this is not a requirement for system operation. Both the dimensions or categories for evaluation and the scaling of such evaluations would follow from agreements between the users of the system. Such agreements would permit more or less effective sorting of articles by computer software. With very high article volume, multi-dimensional and carefully scaled responses would be beneficial. However, if powerful natural language understanding software was available, then unstructured responses could be entirely adequate.

The need for evaluative information becomes much greater with the complex and opaque multi-media or hypertext documents now being developed [6]. With parts of such documents, or with very short articles and more conversational interaction, often associated with voice messages, the types of judgments could be quite different [5]. The degree of impact on the priority relations among articles waiting to be read would be the crucial measure of quality for review messages.

If a reader finds an article to be lacking on a certain dimension, an article may be requested to supplement the judgment given. A structured form of argumentation can then take place. Some authors would be attracted into the controversy and strive to gain credibility by issuing evaluations of a given article. Others would prefer to wait until the situation had stabilized, as calculated by Bayesian estimation, before reading any of the articles [5]. This latter strategy might be called the text or reference book approach to conferencing. The first might be called the meeting approach.

7. Coordination mechanism with automatic negotiation

Ensuring distribution of alternative viewpoints is important in a democratic system, since effective decisions require the widest possible range of information. An invitational mechanism that can select a single competent spokesperson for each different viewpoint can support what has been termed a consensus journal [8].

Consensus journals have the economy of invitational journals and the objectivity of journals based upon the peer review. That is, all articles are published and the reader benefits from article selection based upon impartial refereeing. An additional benefit of consensus journals is that the negotiation process, that typically occurs prior to publication, is automated, thus saving efforts of participants. Readers submit reviews that evaluate articles on agreed dimensions. A statistical procedure is used to identify the most knowledgeable representative of each consensus position and these persons are invited to submit articles that justify the review judgments they have submitted.

Invitational journals can be distinguished from typical scholarly journals by the sequence of events that results in publication of an article. The sequence of events with a typical journal starts with the writing of an article. The article is then transmitted to an editor and refereed. After a successful review, often contingent upon negotiated revisions, the article is published and read. With invitational journals, however, events are reversed. The tentative decision to publish an author is made first, often based upon the reading of previous work by that author. Then negotiation between the editor and author occurs, or there is informal refereeing of a proposal, which if successful, results in the writing of an article. The great advantage of this second sequence -- read, negotiate, write -- is that almost every article written gets published. The disadvantage is that selection of authors is somewhat arbitrary and there is no way an unknown author can get published. Here we outline a method of scientific communication that has the economy of invitational journals and the objectivity of journals based upon the peer review. These self-edited journals are called consensus journals in order to distinguish them from conventional invitational journals.

Any reader of an article in a consensus journal can act as a referee. Assume, for simplicity, that referees send reviews to a mediator. At a deadline, the mediator performs calculations and issues invitations to the referees who have been selected as new authors (Figure 1). These calculations are implicit negotiations, that is, they predict which persons would have been selected to respond to the reviewed article if referees has actually negotiated and reached a consensus.

 ------  R: Review   -------  M: Invitation   ------- 
| Read |----------->| Calc. |--------------->| Write |--------->
 ------              -------                  -------  R:Article
                                                 | R: Renege
 R = Referee    Calc. = Calculate consensus      |
 M = Mediator                                    V

Figure 1. Simplified cycle of operation for a consensus journal

The simplified cycle of operation for a consensus journal shows actions in boxes and messages as arrows. In this simplified cycle, referees invited to publish (and justify the reviews they have submitted) have a choice of submitting their article by a deadline or reneging on the promise implied by their review. This simplified cycle of operation assumes, additionally, that consensus positions can be calculated and that published articles are retained indefinitely. Eliminating these assumptions requires a more articulated cycle of operation and additional message types [8].

8. Summary and Implications

The security of the system reduces the effect of power relations on the interchange of information [7]. The invitational mechanism selects alternative viewpoints for publication automatically, ensuring different viewpoints are presented. The overall system integrates the reliability of the scientific journal with the rapid response of informal dialogue, thereby creating a powerful coordination technology.

Traditionally, knowledge management in the public sector has been simplified by the preprocessing and integration of divergent points of view by political parties and political representatives. While the concept of representation has always been weak theoretically, the mechanism has functioned reasonably well in translating the basic demands of the citizen into public policy. However, modern developments have undermined the effectiveness and legitimacy of the representative system of government. The much greater complexity of modern society has made it difficult for a single party to satisfy the highly diverse demands of an increasingly individualized citizen.

The complexity and interrelationships among events impacting the citizen has created the feeling that a single vote, no matter how well respected, can do little to influence the multitude of events affecting the individual. Increasingly, the idea of forcing all public decision making through a single representative body is seen as inadequate. When we add to this general malaise the corrupt practices that inevitably surround these centers of power, widespread delegitimation and public alienation are the result. Meaningful involvement of the public today must permit direct influence on the issues of importance to a particular citizen. The complexity of modern society requires a diversity of forums for the myriad of issues that confront the public. New forums must be created upon demand, if the public is to be heard on the issues of the day, which arise with increasing frequency as technological change accelerates.

In many societies, the consolidation of the media system into powerful economic units or its domination by political actors has prevented alternative voices from being heard. The resulting limitation of discussion to accepted lines and the inability of novel inputs to reach an audience means a chronic conservatism and a failure to deal with technological or social changes until they result in a crisis. In the crisis management mode, democracy is typically seen as inappropriate or too slow. The proposed system makes it possible for any individual in the society to issue an early warning by stimulating a debate directed toward taking effective action. Such a debate would attract both the diverse opinions in the public arena and the expertise needed to make an informed decision. The rapid exchanges possible with computer-based conferencing permit crisis management to be avoided. Alternatively, this can be seen as a democratization of the (public) crisis management process.

Effective knowledge management is becoming the key to legitimation. The citizen must see that the governance system provides the knowledge needed for decision making. Further, citizens must see that the knowledge they provide to the governance process has a fair chance of influencing decisions. An effective method of doing this is to focus the effort of the citizen directly on the issues of the day. This is achievable with today's computer networking technology. The failure to apply this possibility will rightly be seen by the public as an attempt to maintain a monopoly on power by an elite. In an age where technological developments make weapons of mass destruction available to smaller and smaller groups, the consequences of a failure to redistribute power through a reinvigoration of democracy could be disastrous.

9. References

[1] Chaum, D. 1985: Security without identification: Transaction systems to make big brother obsolete. Communications of the ACM, 28, 1030-1044.

[2] Garfield, E. 1988: Refereeing and peer review: Part 1. In E. Garfield, Essays of an information scientist: Towards scientography (Vol. 9, pp. 230-238). Philadelphia, PA: ISI Press.

[3] Michel, F. C. 1982: Solving the problem of refereeing. Physics Today, 35(12), 9;82.

[4] Spragge, J. G. 1987: If computer users are a community, the conference must be the town meeting. The second Guelph symposium on computer conferencing, (pp. 91-104). Ontario, Canada: Guelph University, Department of Rural Extension Studies.

[5] Stodolsky, D. 1984, December: Self-management of criticism in dialog: Dynamic regulation through automatic mediation. Paper presented at the symposium Communicating and Contracts between people in the Computerized Society, Gothenburg University, Sweden.

[6] Stodolsky, D. 1987, April: Telematic journals [Abstract]. Proceedings of the First STIMDI Conference on Man-computer Interaction. Stockholm, Sweden: STIMDI (Sveriges tvaervetenskapliga intressefoerening foer maenniska-datorinteraktion).

[7] Stodolsky, D. 1994: Telematic journals and organizational control: Integrity, authority, and self-regulation. Interpersonal Computing and Technology, 2(1).

[8] Stodolsky, D. S. 1995: Consensus journals: Invitational journals based upon peer review. The Information Society, 11(4). [1994 version in N. P. Gleditsch, P. H. Enckell, & J. Burchardt (Eds.). Det videnskabelige tidsskrift (The scientific journal) (pp. 151-160). Copenhagen: Nordic Council of Ministers. (Tema NORD 1994: 574)]

[9] Wilkerson, I. 1987, April 18: Ethnic jokes in campus computer prompt debate. The New York Times.

10. Notes

1 Presented in part at the IFIP WG9.2 meeting, Copenhagen, Denmark (May 11-12, 1996). Revision and extension of: Stodolsky, D. S. (1994). Computer-network based democracy: Scientific communication as a basis for governance. In K. Bjerg & K. Borreby (Eds.), Home-oriented informatics, telematics & automation (pp. 51-58). Copenhagen: Oikos. {Stodolsky, D. S. (1992). Computernetvaerk-baseret demokrati: Videnskabelig kommunikation som grundlag for den demokratiske beslutningsproces [Computer-network based democracy: Scientific communication as a foundation for the democratic decision process.] For medlemmernes skyld? [For the members' sake?] (pp. 139-158). Esbjerg, Denmark: Foreningen for studier i andelsbevaegelse og kooperation [Danish Society for Cooperative Studies].} Parts of this paper appeared earlier in: Stodolsky, D. (1990). Protecting expression in teleconferencing: Pseudonym-based peer review journals. Canadian Journal of Educational Communication, 19(1). ([1989, May 9]. Communication Research and Theory Network [CRTNET], No. 175 [Semi-final draft available by electronic mail fromLISTSERV@PSUVM.BITNET at University Park, PA: The Pennsylvania State University, Department of Speech Communication and COMSERVE@Vm.ecs.rpi.eduat Troy, NY: Rensselaer Polytechnic Institute, Department of Language, Literature, and Communication])

2 Text-based or asynchronous (store-and-forward) conferencing systems are distinguished from real-time audio conferencing which requires simultaneous presence.

3 The pseudonym is a binary number of over two hundred digits.

4 Public-key cryptography uses two different keys, one for encoding and one for decoding. The public-key can be widely distributed without risk of revealing the private-key that is used to decode messages and sign documents. This system makes key distribution practical when there are large numbers of users.