Epistemic democracy

Most defences of democracy are procedural: it is fair because everyone has an equal say, whatever the outcome. Epistemic democracy adds that democratic procedures are also good at tracking the truth or producing good outcomes. If many people with partial knowledge are combined in the right way, the group can be wiser than any member, including experts.

The idea has ancient roots. Aristotle suggested that the many, when they come together, may be better than the few best, as a feast to which many contribute is better than one provided by a single host. Modern epistemic democrats give the idea formal and empirical content, and use it to answer critics who propose rule by the knowledgeable.

The Condorcet jury theorem

In his Essai of 1785, the Marquis de Condorcet showed that if each voter is more likely than not to choose the correct option between two, votes independently and has the same competence, then the probability that the majority is correct increases with group size and approaches certainty as the group grows. The converse is equally important: if voters are on average less than 50 per cent reliable, the majority becomes almost certainly wrong.

Later work generalised the theorem. Bernard Grofman, Guillermo Owen and Scott Feld showed that it holds when competence varies across voters, as long as average competence exceeds one half. Christian List and Robert Goodin extended it to choices among more than two options under plurality rule. Franz Dietrich and Kai Spiekermann's survey identifies the independence assumption as the most demanding: shared information sources, opinion leaders and common errors make votes correlated.

Diversity trumps ability

Lu Hong and Scott Page (2004) modelled problem solvers as agents with different perspectives and heuristics searching a solution landscape. They found conditions under which a team of randomly selected agents outperforms a team of the individually best agents, because the best agents tend to be similar and get stuck at the same local optima. Page popularised the claim that cognitive diversity can matter more than individual ability.

Democratic reason

Hélène Landemore's Democratic Reason (2013) combined the jury theorem, the diversity result and deliberation into an argument that inclusive democracy, especially when it uses deliberation among many and diverse people, has greater collective intelligence than rule by the few, even the wise few. Her later work favours open, randomly selected mini-publics as practical vehicles.

Epistemic proceduralism

Joshua Cohen described an epistemic conception of democracy in which voting expresses judgements about the common good rather than private preferences. David Estlund's epistemic proceduralism holds that democratic decisions are legitimate because democracy is better than random at producing good outcomes, and because it treats citizens as equals, without claiming that it always gets things right. This lets him reject epistocracy: even if some people know more, no one can claim authority on that basis in a way that all reasonable citizens could accept.

• Juries: the theorem's original application; unanimity and majority jury rules have been analysed for error rates.
• Prediction markets and forecasting tournaments aggregate dispersed judgements and often outperform individual experts, though they rely on incentives rather than votes.
• Citizens' assemblies in Ireland, France and Belgium are defended partly on epistemic grounds: diverse randomly selected citizens, informed by experts, can produce careful recommendations.
• Crowdsourcing: Wikipedia, open-source software and vTaiwan's opinion-clustering processes rely on many contributors correcting each other.
• Expert panels sometimes use structured aggregation methods (Delphi, weighted averaging) that apply similar insights.

• Voter ignorance: decades of survey research document low levels of political knowledge. Jason Brennan argues in Against Democracy (2016) that this undermines the competence premise and favours epistocracy.
• Correlated errors: voters share media, parties and social networks, so their errors are not independent; misinformation can push average competence below one half on some questions.
• Is there a correct answer? Many political questions involve value conflicts rather than facts, so it is unclear what "truth-tracking" means.
• Misuse of mathematics: Abigail Thompson argued that the Hong–Page theorem depends on artificial assumptions and that its popular interpretation, "diversity trumps ability", is not supported by the mathematics; defenders reply that the model makes a narrower point about search and diversity.
• Deliberation's mixed record: deliberation can correct errors but can also polarise like-minded groups or be dominated by confident speakers.
• Instrumentalisation: grounding democracy in outcomes risks making it hostage to performance; if another system did better, would it then be preferable?

Epistemic democracy has moved from a fringe position to a major strand of democratic theory. Its formal results are valid under their assumptions, but those assumptions (independence, competence above chance, a correct answer) are often violated in politics. The stronger version of the argument is comparative and institutional: well-designed inclusive procedures, combining diversity, information and deliberation, can match or outperform closed expert bodies on many questions. Designers apply this through mini-publics, structured forecasting and aggregation platforms rather than through unaided mass voting.

• Franz Dietrich and Kai Spiekermann, "Jury Theorems", Stanford Encyclopedia of Philosophy
• List and Goodin, "Epistemic Democracy: Generalizing the Condorcet Jury Theorem" (2001)
• Hong and Page, "Groups of diverse problem solvers can outperform groups of high-ability problem solvers" (2004)
• Abigail Thompson, "Does Diversity Trump Ability?" (2014)
• Joshua Cohen, "An Epistemic Conception of Democracy" (1986)
• Hélène Landemore, Democratic Reason (Princeton University Press, 2013)
• David Estlund, Democratic Authority (Princeton University Press, 2008)