Conference on Rating, Voting, Annotation - Vienna 1997

Vienna, 21-22 April 1997


Comments on

Voting and Rating - Perspectives for Information Collection, Decision Making and Collaborative Rating Using Web4Groups

by Arnold B. Urken 16 April 1997

Note: These comments are not organized in any particular order. I will make overheads with bullets for discussing these ideas. I hope these comments are useful.

1. Direct Participation and Voter Competence: Although the Web4Groups report explicitly rules out discussing teledemocracy from the viewpoint of political theory, it is important to take account of the classic problem of the difference between the will of all and the general will, first articulated by Rousseau. It's not only a problem that is related to sampling; it is one that was addressed by Condorcet in his famous "jury theorem.” Condorcet used the binomial theorem to model the process of finding the "correct choice” in an election in which it is assumed that one binary choice is correct (Rousseau's general will) and the other choice is not.

Condorcet showed that mass democracy can work efficiently even though the competence of the average voter is only 50/50. In other words, the group probability of making a correct choice can be virtually perfect even when the average competence of the voters is mediocre.

There is a related literature that deals with the best way of weighting the opinions of experts when their performance can be measured. Intuitively, one might weight each expert's vote by the ratio between their probabilities of making correct and incorrect choices. Shapley and Grofman (1985) show that better results can be obtained with using the log of expert probabilities of making correct and incorrect choices. Urken (1988) confirmed this result in a Monte Carlo simulation.

Although this literature is concerned with long-run patterns, it provides a basis for thinking about and experimenting with many tasks in which the voting system might be designed to augment or enhance the performance of small or large groups of experts.

2. Motives for participation: While it makes sense to me that people will be motivated to participate by feedback from governmental officials, it also seems worthwhile to investigate the use of immediate feedback from polls and surveys as a motivation for participation. In other words, the poll may have entertainment value and may also be structured to allow individuals to compare (benchmark) their opinions against reference points ( typical individual or some nominal or real group breakdown of the data). If surveys are well designed and include comments, they can serve as an enabling technology for promoting deliberation. And they may also provide a basis for identifying interests that can coalesce, groups of individuals who would have otherwise not realized that they shared any common objectives or values.

3. Asynchronous voting: If voters were allowed to vote over a long period of time (e.g. a week), would the results be easily manipulated those waited to vote last if voting results were known in advance? Would this happen even with ranking rather than one person one vote voting? If voters could change their votes, would manipulation be practically impossible? The answers to these complex questions are not clear.

Making the running vote total public may induce attempts to manipulate the outcome, but it may also motivate participation, particularly if (anonymous or quasi-anonymous) reasons for voting were attached to the vote results.

How would candidates handle such elections? Ambiguity, a common strategy in many political systems, might not work so well.

4. Collecting and owning voting data: Standards for data collection and storage are an interesting area. Why should governments have a monopoly over data collection. In the US, such operations are frequently inefficient, corrupt, and insecure.

Should an individual be able to obtain a receipt for his/her vote and then actively audit the result (i.e. check on the vote to make sure that it has not been deleted and has been counted properly)?

5. Fungible voting: James Coleman and others who have written about this system have not gone beyond idealized scenarios. It would be interesting to conduct experiments in which votes can be given as proxies to representatives.

Although this type of voting system could develop into something that is driven by the sometimes volatile and destabilizing money economy, it does have interesting self-stabilizing properties. In fact, one of the rules that I recall is a redistribution rule that takes some votes from each collective decision and spreads them out over the population to make sure that everyone retains some stake in and derives some benefits from the game.

About the conference
About Web4Groups
Study on voting and rating

Page created by Gernot Tscherteu and Sabine Benzer
Last update: 13 05 1997