List and Goodin: Epistemic democracy

List and Goodin. 2001. Epistemic democracy: Generalizing the Condorcet jury theorem. Journal of Political Philosophy 9.

In Brief

List and Goodin introduce the Condorcet Jury Theorem, which argues that under certain plausible conditions, proper voting rules do track the truth--that is, they produce the "correct" outcome. Condorcet's logic, then, can be extended beyond a binary choice; since the logic holds, then if the assumptions are true, then the conclusions of the theorem will be true. (This article leaves unanswered whether the assumptions are true.) This article is most useful as an introduction to the CJT, though it also includes an interesting discussion of plurality rule. See also Wikipedia's article on Condorcet's jury theorem.

Take-home Point:

The Condorcet Jury Theorem (CJT) can be used to explain the truth-tracking properties of voting. If individuals are (on average) more likely to make correct decisions than incorrect ones, then large groups of them will be very likely to make correct decisions via majority/plurality voting mechanisms.

Main Argument

People discuss democracy in two ways. 'Epistemic democrats", like Rousseau, wish for a governing mechanism that will "track the truth"--that is, satisfy some objective standard. 'Procedural democrats', like Dahl and Schumpeter, are more interested in the rules of the game, regardless of the outcome. When it comes to a majoritarian choice between two alternatives, epistemic and procedural democrats have no need to disagree; Condorcet's jury theorem long ago showed that, as long as the average voter has at least a 50% chance of making the (epistemically) "correct" decision, majoritarianism works. The authors' main contribution is to argue that most other democratic procedures (with more than two options) also "work" under the same assumption, and they all work equally (more or less) well. Thus, plurality rule, the Borda count, etc all work as "truth trackers." We need not worry about epistemic debates, then, when deciding what sort of democracy to have.

The CJT Explained:

"If each member of a jury is more likely to be right than wrong, then the majority of the jury, too, is more likely to be right than wrong; and the probability that the right outcome is supported by a majority of the jury is a swiftly increasing function of the size of the jury, converging to 1 as the size of the jury tends to infinity" (page 283).

• The CJT holds up even when jury member capabilities are heterogeneous--i.e. some are smart and some are dumb--provided that the mean probability of choosing correctly across the jury is greater than 1/2.
• The CJT is very straightforward when applied to binary choices, but becomes more complex as the arena of possible choices is expanded.
• The CJT applies to plurality voting (and other voting rules with k>2) as well as majority voting. The former case is interesting because the CJT holds up even when the mean voter is less than 1/2 likely to be correct, so long as they are more likely to choose the correct choice than any one incorrect alternative. Larger juries are required for this result to be robust, however (see Table 1).

Difficulties of Applying the CJT

This article shows only that Condorcet's logic holds. Thus, if the assumptions are true, the conclusions are true. This article does not claim that the assumptions are true, though, only that the logic is correct. So we cannot apply the theorem of one of these three objections to the assumptions is correct:

1. There is an objective standard of "correctness"
2. Competence must be better than random (i.e. 51% for a binary choice)
3. Individuals make independent judgments--one individual's choice does not influence another's. (This gets thrown off if we use polls or opinion leaders as information cues. And if there is one thing we know from voting studies, it is that people influence one another; see the Columbia voting studies and books like Lupia and McCubbins 1998.)

Comments and Criticism

• Big threat': This article requires that there be an independent standard of correctness. By what criteria are political decisions deemed "correct"? This problem is avoided in juries, since guilt/innocence is fairly straightforward, but how does one judge which political leader, health care system, etc... is the "correct" choice?
• What if there is no "correct" choice, or what if people disagree as to which standard to use? The authors show that plurality voting rules work fine for "truth tracking" when everybody has a marginally higher probability of choosing the "correct" option. But what if some people prefer one option, and others prefer another option?
• Political choices are rarely binary. Even when they are presented in binary fashion, the process used to narrow the alternatives down to two will likely be biased (see footnote 31 on page 284). Even in the case of plurality voting, there is usually some limitation of available options that may be biased (do write-ins alleviate this problem?).
• Is there a good reason to assume that people are, on average, more likely to be right than wrong? Is there a way to measure this in a manner that can be extended to the types of decisions commonly made in the voting booth?
• Would it be better (in epistemic terms) to have 435 people with, say, 60% mean competency make decisions by majority/plurality rule, or 1,000,000 people with 50.001% mean competency? If there is little difference, what does this say about representation as a means of political decision making?