"This also helps us think more clearly about the possibility conditions for highly successful problem solving in democracies. We summarize these in the most cursory fashion (we hope to expand on this in further work). First, in contrast to existing epistemic accounts, cognitive accounts suggest that individuals need to be able to expose their different points of view to each other, rather than the polling of individuals in strict isolation from each other required by Condorcet’s Jury Theorem. This goes hand-in-hand with a different account of problem solving — rather than asking whether people can determine whether a given decision will be correct or incorrect, as Condorcet does, it asks when individuals will be able to discover hitherto-unperceived solutions within a complex landscape. Second, individuals need to be at least “weak learners” in the terms of statistical learning theory (Schapire and Freund, 2012). Individuals who are fundamentally obtuse, profoundly blinded by ideology, or whimsically perverse will detract from collective learning rather than help it. Third and related, as Mercier and Sperber suggest, cognitive democracy requires that individuals participating in democratic argument have some core commitment to the truth, even if they disagree strongly about what the truth is. People need not be as disinterested as Gardner (this volume) would like them to be, but neither should they be so warped by self-interest that they cannot see the truth, or allow themselves to care for it. Fourth, even if people disagree on how to solve a problem, they agree on what the problems are that need to be solved in the first place, and have some minimal common empirical standards (see also McAfee, this volume).
Clearly, these conditions are falsified in our everyday political experience."
(p. 18 of the below mentioned document; via @henryfarrell)