Chapter 9 advances the first stage of an argument which is completed in chapter 10. I’ll just concentrate on the part of the argument that is presented in chapter 9, but it is worth seeing where it is going. The burden of the two chapters is to show that feasible democratic procedures have substantially more than random probability of finding the correct answers to what society should do. Chapter 9 only argues that an imaginary situation which Estlund calls a model epistemic deliberation would have a substantially better than random chance of getting the right answers, but he also says that this model is not an ideal for which it would be sensible to strive. Rather, it illustrates how an ideal democracy could be justified epistemically, and the subsequent chapter will argue that feasible democratic procedures for which it would be sensible to strive are relevantly like this “ideal”. The point is to “defeat a certain kind of sceptic, the one who denies that any (nonutopian) democratic arrangements could tend to perform better than random”.
How does the argument of this chapter work, then?
Estlund eschews providing a determinate theory of justice by which to evaluate procedures because, as he says, any justification of democracy that presupposes a controversial theory of justice will inherit the controversy concerning that theory of justice. It might seem very odd to claim epistemic value for a procedure despite not having specified the knowledge for which it is supposed to be epsitemically valuable but in fact this is quite a common situation in epistemology — we claim great success for scientific method, despite the fact that what we know about science we know only by using the scientific method. And, he thinks, we have independent access to some of the content of a theory of justice or the common good — an account of particular public bads. So the method is to offer a theory of primary bads, and argue that model epistemic deliberation will do much better than random at avoiding these.
A list of primary bads must meet several conditions. The issues should be among the most important issues that are subject to political decision; it should contain enough variety that we “have some reason to extrapolate from performance on this set to a great number of issues not on the list”; the list should avoid unnecessary length, complexity or empirical inscrutability; and it should meet the requirements of public reason: “it should be seen to be capable of serving its function from within a wide variety of reasonable or qualified points of view”.
Estlund’s list of primary bads: war, famine, economic collapse, political collapse, epidemic, and genocide.
The list meets all the conditions — avoiding each of these things is among the most important things a government can do, and failing to avoid them when they are avoidable is a terrible failure. They are tractable, meet the public reason requirement, and it seems possible to extrapolate from them to other bads, including some that would not meet the public reason requirement.
What, then, is the model epistemic deliberation that will do substantially better than random at avoiding this list of public bads? He offers a list which is an answer to the question of what set of conditions will make it plausible that, on the sorts of issues that political communities will face, deliberation would have a significant tendency to make decisions that are, on independent grounds, morally right. Here are the answers to tahat question:
Everyone has full and equal access to the forum
Everyone has the same chance to speak as everyone else
People only say thing that they believe will help others to appreciate the reaons to hold one view or another among those that are in question
Anyone whose interests are at stake in the decision is either present or represented by an effective spokesperson
Everyone has as much time to speak as they wish
Everyone has equal bargaining power
Everyone equally credits and attends to the contribution of others
Everyone recognizes (or tends to recognize) a good reason when they see one
Participants strive to address the “devil’s advocate”
Estlund makes no direct argument for the superiority of this model over randomness for the avoidance of any of the bads on his list. But he argues that deliberation under these conditions would have considerable epistemic value. It enables the dispersal of knowledge and the laundering of reasons; it is a procedure that will tend to reject demonstrable falsehood, bad reasoning, and bad reasons, and to filter in good reasons and truths. Sure, people have flaws, blindspots and pathologies, but these are not bound to be “too severe for the purposes of epistemic proceduralism conceived of as an aspirational theory’. And that, Estlund says, is all that he needs.
Estlund considers an objection and an alternative.
The objection: The outcome of a procedure is collective judgment, formed out of the aggregation of individual judgments. But if so, the process of judgment is liable to fail to adhere to certain standards of judgment like logical consistency. Consider the tenure case:
A university committee has to decide whether to give tenure. The requirement is excellence in both teaching and research. One committee member thinks she is excellent in teaching but not research, another that she is excellent in research but not teaching, and a third that she is excellent in both. A majority considers her excellent in teaching, a majority that she is an excellent in research, but a minority considers that she should get tenure. This looks like logical inconsistency in the judging agent.
Estlund responds that the mistake here is to think that there is a judging agent: what are rendered in collective deliberations are not the judgments of a single judging agent, but the collective decision, not judgment, of a collection of individual judging agents.
The alternative: Why should the franchise in the model epistemic procedure by wide, rather than narrow? After all, we might think that we could achieve the epistemic benefits of the procedure by choosing a large enough group of deliberators by lottery, or by using some sort of qualification procedure. Estlund argues against the use of a qualification procedure by appeal the unavailability of invidious comparisons. He argues against the use of lottery as follows: it is not infeasible to have all adults free to vote in periodic elections, and so this is a way to partially harness the epistemic benefits of the whole large group. But, if it were pragmatically possible to have “even elections decided by only a small randomly chosen set of voters, after plenty of discussion by all…. [there is} nothing in this arrangement that is particularly offensive or contrary to the moral spirit of democracy.
A coda. The chapter helped me to see that Estlund’s theory of legitimacy and authority is not, as I have sometimes wondered, a consequentialist theory. He’s right that consequentialism in this context is slippery (or, perhaps, just a term of art). But on a narrow understanding of consequentialism Estlund’s theory does not meet the conditions. It does not allow, for example, that the government may do just anything in order to avoid primary bads. Nor does it offer a fully specified theory of the good or justice against which governments (and democratic procedures) should be judged. And finally, as he says, “The general acceptability condition, which says that political justification must be acceptable to all qualified points of view, entails that if there is a true view of what the best consequences are but this is not available to all qualified points of view it is not available in political justification”. So, I think it is fair to say that his view is not consequentialist in any interesting sense, even though it depends on the consequences of the procedure.
A couple of comments
First comment is to emphasize that I agree with Estlund that, given his epistemic account, there is nothing wrong with having the electorate selected by lottery. It is implausible to me that universal franchise is necessary in order to achieve the epistemic benefits he seeks, and it is also implausible to me that knowing one will have a vote is necessary to prompt deliberation (there is not much difference between knowing one has a chance of getting a vote, and knowing that one has a much smaller chance of affecting the outcome with one’s vote). But I wonder why we need to allow even universal participation in deliberation? Deliberation within a randomly selected and manageably sized but large enough o be inclusive group seems more likely to have epistemic benefits than universal deliberation.
Second comment: exclusion from deliberation or voting by means of a lottery does not involve any invidious comparisons. But it does seem contrary to the moral spirit of democracy. Democracy is, in part, justified because we have an obligation to allow other people to play a full role in determining how our collective affairs will go, precisely because they have a stake in ho things will go. It is true that democratic procedures offer them only a very small chance of being decisive, but it offers them some chance, and the ability to offer their own judgment in a communal process, and know that it is as likely to count for something as anyone else’s. I think this is one, but not the only, value that plays a part in justifying democracy. I have long been unsure whether Estlund disagrees, or whether the same basic thought underlies the bar on invidious comparisons (or, more precisely, the conception of legitimacy that underlies that bar — the conceptual space is just not clear to me). The comment about lottery suggests this is a real disagreement (but I may be wrong about that!)