Yes, We Can (Make It Up on Volume): Answers to Critics

ABSTRACT

The idea that the crowd could ever be intelligent is a counterintuitive one. Our modern, Western faith in experts and bureaucracies is rooted in the notion that political competence is the purview of the select few. Here, as in my book Democratic Reason, I defend the opposite view: that the diverse many are often smarter than a group of select elites because of the different cognitive tools, perspectives, heuristics, and knowledge they bring to political problem solving and prediction. In this essay I defend my epistemic argument against proceduralist democrats; the value of model thinking against empiricists; the bracketing of fundamental value diversity against critics who see such diversity as an essential feature of politics; the intelligence of the masses in the face of voter ignorance and systematic biases; and the normative priority of democracy over market mechanisms. I also consider challenges to my use of Hong and Page's formal results, the epistemically proper selection method for representatives, and the role of deliberation in problem solving. I finally chart three avenues for further research.

Notes

1. The risky part having to do with the fact that, through the generosity of Critical Review, I can't even appeal to a word limit to save myself from answering some objections.

2. I count this objection as external since my critique of the empirical literature on voter ignorance is part of a sub-argument within the book that I do not see as essential to my core analytical claim.

3. Their thesis, which I became cognizant of only after my own book was already published, is hence quite close to what I have in mind when I argue for the epistemic superiority of democracy (except that Knight and Johnson's argument is broader and not restricted to epistemic issues, which is perhaps why they ultimately reject the label “epistemic”): Democracy is what we should use when we need to make a collective decision and we do not know what to do. We first talk the issue through, lay our options on the table, and then take a vote on one of the proposed solutions. Some of these democratic decisions may well be to delegate the solution of the identified problem to a subset of the group, whether experts or volunteers, or even to let the market do its thing, for example if the question is one of prediction or allocation of goods that lends itself to that decentralized solution.

4. I thus mention in the book Ségolène Royal's Deweyan presidential campaign on the theme of citizens’ expertise, of which one radical proposal was that interest rates should be set through some kind of democratic authority rather than an independent central bank.

5. See also Kitcher 2011 (especially ch. 2), which, however, does not have a very specific answer either.

6. See http://www.goodjudgmentproject.com/.

7. Having said this, the conception of representative democracy defended in the book is itself rather radical in important ways, since on my reading the best version of representative democracy would be one in which representatives are randomly selected rather than elected.

8. Which itself follows a wave of publications on the strengthening of the oligarchic element in American politics (e.g., Bartels 2010; Hacker and Pierson 2010; Schlozman, Verba, and Brady 2012).

9. They are already wearing out in circles of activists. Iceland has experimented with quasi-random samples of its population. France has a very popular populist activist, Etienne Chouard, who since the 2005 referendum in France has drawn millions of fans to his ideal of a direct democracy with lottocratic elements.

10. Procedural, I should say, is not a great term. I myself use it in the book; but really it is confusing. Both epistemic democrats and “proceduralists” such as Muirhead are proceduralists, though in different ways. Epistemic democrats like myself believe in the epistemic value of the procedure of democracy, separate from (and potentially in addition to) its intrinsic value, while so-called proceduralists such as Muirhead care about the procedure for what it expresses, embodies, symbolizes, or “reflects.”

11. They continue: “Put most starkly, we defend particular conceptions of freedom and equality insofar as they are required for democratic decision-making rather than defending democracy because it is instrumental to attaining the goals of freedom and equality” (Knight and Johnson 2011, 20n50).

12. On my epistemic account, the fact that China and Singapore do well is not a reason to embrace autocratic rule over democratic rule.

13. Indeed, as Estlund (1997 and 2008) argues, it is not entirely clear why political equality should commit one to majority rule rather than to a number of other procedures—coin-flipping and Queen-for-a-Day among them—as political equality seems equally distributed under those alternatives as well. I don't know that any proceduralist has ever been able to answer that objection satisfactorily.

14. I leave aside the question of how bad the wrongs would have to be, since that issue might perhaps be solved by adding a set of fundamental human and liberal rights to the constraints on political legitimacy.

15. And this in turn most likely requires a better understanding of what the very concept of political legitimacy demands—whether it always and everywhere demands a particularly democratic legitimacy, or if there are (sometimes, in some places) other sorts of political legitimacy.

16. The only thing that changes—the only thing I vary, if you will—is the number of decision makers. In that comparison, I set out to “prove” (in a non-empirical sense) that provided certain conditions are met and given the nature of political choices, the model democracy will do better than the model oligarchy. That's the gist of the Numbers Trump Ability Theorem.

17. See instead the excellent work by Michael Weisberg (2013) on this. Scott E. Page (forthcoming) also has some very interesting thoughts on this; see his forthcoming Model Thinking. For an intriguing take on the role of models in political theory, see Johnson 2014.

18. See Rawls 1971 for the classic definition of ideal theory.

19. So to be clear, my model of democracy is not an ideal in the normative sense, that is, a picture of democracy as it should be. Instead, it tries to capture some minimal essential features of democracy—a diverse, numerous set of decision makers trying to solve problems through various deliberative and aggregative mechanisms that weight individuals’ input equally. Saying that this is “ideal” (again, in the normative sense) is a little like saying that the decision makers in Olson's collective-action dilemma are a model for how men should behave, simply by virtue of the fact that they are abstracted and idealized representations rather than empirical, historical figures.

20. I'm now working with French economist Sasha Bourgeois-Gironde to develop a more complex model in which elements of strategic behavior and value diversity are reintroduced. See also a paper by Stephen Stich (2014a), which aims to offer a version of the epistemic argument for democracy that accommodates fundamental value diversity in ways my own approach does not.

21. This objection is raised almost in passing by Sandy Levinson's otherwise generously supportive essay, when he wonders about “the actual degree to which a given (would-be) polity in fact shares a sufficient ‘consensus of values’ to make it likely that it will be able even to survive, let alone to flourish” (Levinson 2014, 99).

22. For Goodin (2003, 86-87), “bargaining over beliefs” is not so much an “agreement to disagree” as a way to “settle on some course of action, together with some rationale as to how it is supposed to work to produce the desired results. In the course of that, they agree to treat certain beliefs ‘as if they were true.’ But they definitely do so in the subjunctive case—in the tentative and hypothetical way in which propositions being tested are treated in scientific experiments.”

23. The paper claims to show, among other things, that “democracy increases future GDP by encouraging investment, increasing schooling, inducing economic reforms, improving public good provision, and reducing social unrest” (Acemoglu et al. 2014). Most strikingly, the authors claim to quantify the impact of democracy on GDP: “Our central estimates suggest that a country that switches from nondemocracy to democracy achieves about 20 percent higher GDP per capita in the long run (or roughly in the next 30 years). Our results indicate no differential effect of democracy on economic growth by the initial level of economic development, though there is some evidence that democracy is more conducive to higher GDP in countries that start out with higher levels of education.”

24. Against a whole literature trying to demonstrate that the average voter is mediocre based on empirical measures of their “input” to democracy, Page and Shapiro famously showed that if you look at the output, then cautious optimism is allowed.

25. For a more detailed critique of Caplan, see Elster and Landemore 2008.

26. This could happen, for example, because inter-elite competition (e.g., in political parties) weeds out all incompetent candidates, so that only good candidates win a major party's nomination in the first place. This (if we also assume value homogeneity, so that candidate “competency” is the only dimension on which to choose) would guarantee broadly good outcomes even in the face of the most incompetent public you could possibly imagine (only because the set of candidates from which this hypothetical electorate would choose is a set with no weak candidates).

27. Since this critical argument by itself does not justify according priority to more centralized institutions, Knight and Johnson also consider three other decentralized institutional mechanisms (Coasian bargaining, community, and technocratic incentive-compatibility schemes). They advance analogous reasons for withholding priority from them (singly or jointly) as well.

28. Knight and Johnson (2011, 52) actually go through the trouble of comparing the performance of democracy and the market to conclude that compared to democratic decision making, the market does worse “in terms of the criteria by which [they] justify the priority of democracy.” They also provide what seems to me a compelling refutation of the literature that aims to present the market as chronologically prior to centralized institutions and sustainable without them. They show that the most robust account of institutional emergence (Coasian bargaining) “actually undermines just the sort of confidence required to sustain the privilege frequently accorded to markets in discussion of institutional emergence and change” (ibid., 69), and that “absent some effort to alter the context in which institutional arrangements are initially selected and subsequently operate, markets will operate, at best, quite imperfectly and, thus, are not a good candidate for an ex-ante first-order priority vis-à-vis other institutional forms” (ibid., 70).

29. Surprisingly, Quirk (2014, 147) admits that he is not all that opposed to schemes involving direct participation by amateur citizens, since he is well aware of the success of the British Columbia Citizens’ Assembly, about which he favorably cites the classic study by Mark E. Warren and Hilary Pearse (2008).

30. If anything, I hope this shows that Hong and Page are themselves quite interested in applying their results to real-world contexts and that I am not as “in the dark” with respect to the concrete implications of their results as Quirk suggests.

31. The two unproblematic assumptions are that the number of members of the pool from which the problem solvers are selected must be large enough, and that the problems must be hard enough that there isn't an obvious expert capable of solving them.

32. Page (2007, 160) goes on to specify that “this condition does not say that given any solution, some problem solver exists who immediately can jump to the global optimum. That assumption would be much stronger and would rarely be the case. Our assumption says, instead, that some problem solver exists who can find an improvement. That improvement may be small.”

33. Individuals need not know the exact values; they need only be close enough. Importantly, the assumption of an oracle does not mean that the problems are trivial and the solutions obvious from the outset (which would violate one of the other assumptions). We would not need deliberation if that were the case. What the oracle assumption means is simply that once a better solution has been brought up or devised, it will appear as such to everyone.

34. The classical example of something equivalent in psychology is the Wason task. Typically people get the answer wrong 90 percent of the time. Once someone explains the solution to them, however, they are capable of recognizing its truth in 95 percent of the time, which is why groups are much better at solving it than individuals (if one person in the group figures out the solution, she is quickly able to convince all the others of the right answer).

35. This is, I think, the objection specifically raised by Jamie Kelly.

36. For example the eight categories of intelligence proposed by Howard Gardner (1983): visual-spatial, verbal-linguistic, logical-mathematical, bodily-kinesthetic, interpersonal, intrapersonal, musical, and naturalistic.

37. See Philip Tetlock's “The Good Judgment Project” at http://www.goodjudgmentproject.com/.