I recognize that David says at the top of 81 that “we should not say that procedures… are not fair simply because they are not fully anonymous. Rather, procedures can mix fairness and other principles, remaining fair, among other things. But the fairness of a procedure is, on my view, to be understood by reference to anonymity, the blindness to personal features.” I guess I don’t understand this passage. If fairness is full anonymity, then aren’t deviations from full anonymity deviations from fairness? Isn’t there a loss of fairness, on this understanding of fairness, whenever we move away from full anonymity? Why should equal treatment in the design of a decision-procedure require complete insensitivity to preferences, as opposed to equal sensitivity?

Despite all of that I find myself persuaded by the conclusion of the argument, on p.83, where David confronts the modified version of fair proceduralism. At its simplest, the argument is that we can’t avoid all substantive disagreement by retreating to equal treatment in the design of the decision-making process, because (a) we’ll need to argue for a particular dimension in which people need to be treated equally (e.g. preferences or votes?), and (b) we’ll want to argue for positive influence, not just equal lack of influence. The bare idea of “equality in the procedure” isn’t enough, unless one is willing to accept randomization, which achieves equality in every dimension. So the real question is about what substantive values one thinks it appropriate to admit into the design of the decision procedure. Is that right David? And do you disagree with this p.83 argument, Jonathan, or do you think it blurs an important difference between kinds of values, by whatever name we want to call them?

(In the below, read “substantive” and “procedure-independent” as the same thing, and read “purely procedural” and “intrinsically procedural” as the same. Also, I assume that if a standard is not purely procedural, it is substantive, at least in part, i.e. that there’s no logical space for standards that are neither purely procedural nor substantive. I believe this usage is consistent with the usage in the chapter.)

One pass at the distinction the chapter seems to be assuming is something like the following:

SP1: a standard is purely “procedural” [or “intrinsically procedural”] iff we can evaluate compliance with the standard without knowing any facts about the outcome it generates.

Accordingly, aggregative standards are substantive because, so goes the argument, they require facts about the relationship between the procedure and the outcome — a certain voting mechanism not only has to be correlated with the outcome, but it also has to cause the outcome in a certain way.

But that can’t be right. For one thing, it’s in tension with the claims at the end of the chapter, that full anonymity is a procedural standard. Full anonymity is defined on the results, and specifically their failure to change if we change features of the people.

This tension is particularly apparent in the coin flip example. On pg. 82, it’s claimed that a coin flip is purely procedural. But on pg. 75-6, it’s claimed that a coin-flip procedure that actually insists that the outcome be a result of the coin flip (rather than merely correlated) is not purely procedural, because it too demands that the coin flip cause the outcome.

I fear I’m misunderstanding the point of the coin-flip. But it seems fairly clear: pg. 75: coin-flip rule with an idea of retrospective fairness demands that outcome be responsive to coin-flip. Pg. 76: “The … responsiveness… is part of a procedure-independent standard.” Ergo, responsive coin flip is not purely procedural. But, pg. 79: “full anonymity … is the logical limit of procedural fairness.” Pg. 82: a coin-flip is “blind to all features of people.” By the definition of full anonymity, a coin-flip is thus fully anonymous, and ergo, the coin-flip is at the logical limit of procedural fairness.

But how could a procedure (coin-flip) be both at the logical limit of procedural fairness and procedure-independent? There are three possibilities:
1. The chapter contains a contradiction. Rule that out by the principle of charity.
2. Even at the logical limit of procedural fairness, there’s no such thing. Possible, but if an eliminativist argument was being offered, presumably the chapter would say so.
3. There is something going on that allows us to distinguish the coin-flip of pg. 75 from the coin-flip of pg. 82.

Let’s consider #3. What kind of a distinction can distinguish between the coin flip generally (purely procedural) and the coin flip which only counts as acceptable if it causes the outcome, not (pg. 75) where the uncle would steal the watch anyway (partially substantive)?

Well… surely if using a coin-flip as the standard is to have any content at all, it must be necessarily related to the outcome — the correlation between the coin flip and the outcome can’t be indeterminate or purely contingent. If it were, it would be no standard at all, because we’d have no facts by which we could judge whether the procedure was satisfied (except, I suppose, facts about whether the coin-flip was held at all, regardless of its relationship to the outcome — but that bears no resemblance to any kind of evaluative standard for a process that anyone sane would ever propose, so we can safely discard it). We’d be back to the eliminativist claim that there is no such thing as a procedural standard.

This suggests SP2: a standard is purely “procedural” iff we can evaluate compliance with the standard by examining the extent to which the procedure is necessarily correlated with the outcome, without examining the extent to which or manner in which the procedure causes the outcome.

That might do the work the two coin-flips demand, but it seems to entail all sorts of contestable metaphysical commitments about causation and so forth. And it’s not intuitively compelling.

On the other hand, suppose I assert SP3: a standard is purely “procedural” iff we can evaluate compliance with the standard by examining only facts about the procedure and its relationship with the outcome, including causal facts, and need not examine any other facts about the outcome.

SP3 is intuitively more compelling than SP2, and it doesn’t require me to claim that causation is something over and above necessary connection. SP3 evaluates aggregation, anonymity, and a whole host of other intuitively procedural standards (like full participation) as “procedural.” And yet it excludes the sorts of things that we usually consider substantive — whether the ultimate decision maximizes utility, whether it is consistent with revealed truth, whether it is just in a non-procedural sense, etc.

Isn’t SP3 a better way to distinguish between substantive and procedure? And if so, can’t procedural standards do a lot more work?

One easy case is a jury trial. Suppose everyone in the relevant qualified population (the judge, all lawyers, the community at large, etc.) knows that the defendant is guilty (say he robbed a bank on FBI payday). Further suppose that everyone — even the defendant (though of course he nonetheless prefers to avoid punishment)– agrees that he morally deserves punishment. Summarily clapping him in gaol has probability 1 of being correct. On the other hand, giving him a trial has probability 1-epsilon of being correct, where epsilon represents, say, the chance that one juror will come to trial drunk, will get bribed, etc., or otherwise vote to acquit for some bizarre reason.

This seems to be pretty close to many real-life cases — a lot of obviously guilty people get arrested every day.

In that case, summary punishment strictly dominates a jury trial for all reasons *except* procedural ones. Would anyone say that we ought to deny that defendant a jury trial? I submit the answer is no, and it’s no because we place independent value on the procedure, for reasons like community participation in criminal punishment, bringing about a Millian “school for public spirit,” etc. (Call that “substantive proceduralism,” perhaps. Maybe this is addressed in the next chapter?)

If that’s right, substance can’t have priority over procedure.

But maybe (e.g. the previous comment) that’s really an appeal to a substantive standard? But then Jonathan’s comment seems really compelling. That’s a “substantive” value, sure, but it’s a substantive value that is only addressed to the method of making a decision, not the outcome of the decision. We care about procedural/substantive distinctions because we care about making decisions when there’s disagreement over the outcome of a decision, and so “substantive proceduralism” is procedural in the way we need?

Thanks for the great summary and questions. I have a question of my own, for either you or David. I confess to being puzzled by David’s claim that pure proceduralism requires anonymity. David claims that ‘a procedure is fully anonymous if and only if it is blind to personal features: its results would not be different if any features of the relevant people were changed’ (p.80). This entails that an anonymous procedure must be blind even to people’s preferences, since preferences are also personal features. This means, then, that majority-rule cannot be justified by pure proceduralism since majority-rule fails to be anonymous.

I think there are two distinct questions we can ask about a decision procedure. One is whether there is an independently correct answer to the question being decided by the procedure. If we use a decision procedure to answer the question – should abortion be legal? – one issue is whether there is an independently correct answer about the legality of abortion. Let’s call this the substantive question. The second question is, by what standards do we evaluate the decision procedure? For example, do we evaluate the decision procedure on its capacity to promote artistic beauty in the world? Let’s call this the procedural question.

David’s argument seems to be about the procedural question. He claims (I think) that if a decision procedure is evaluated using anything other than the principle of anonymity, then that evaluation fails to be purely procedural – it has invoked procedure-independent standards. But it seems to me that this doesn’t quite capture the debate. Suppose a proceduralist democrat, when asked the procedural question, responds as follows: ‘I think democratic procedures should aim to distribute political power according to the following principle: give each adult a maximal share of political power (measured in votes and opportunity to run for office), consistent with a similar share for each other person. This is the standard by which democratic procedures should be judged’. This principle clearly fails to be anonymous in David’s sense, but I don’t see why that should be troubling for the proceduralist. The reason I don’t think it’s troubling is because the proceduralist’s answer to the substantive question can still be no, that is, the proceduralist can say ‘there is no standard, independent of the procedure I’m advocating, by which political decisions can be judged right or wrong’. So long as the answer to the procedural question does not entail or assume a ‘yes’ answer to the substantive question, then I think we can have an argument which is distinct in the right sort of way from epistemic arguments for democracy. I think this point is somewhat similar to a point Ben makes in his first comment.

David’s argument, very crudely, is that coin toss and majority-rule are both fair in some sense, and thus to justify majority-rule, you need to go beyond pure procedural fairness and invoke some procedure-independent standard. I think David is right about this, but I don’t think that’s sufficient to show that the proceduralists are then no longer being purely proceduralist in the sense we care about. It doesn’t show that they are invoking procedure-independent standards in the same way that epistemic theories do. An argument which mixes egalitarianism with pareto to offer a principle for distributing political power remains procedural on the substantive question, and I think that’s all that matters. So long as the argument doesn’t imply that there are procedure-independent standards to that question, I think it remains distinctively procedural. What do you think?

I’ll let 1) go at that, adding only that if indeed there is ultimately nothing in fairness to choose between a lottery and majority rule, that might tend to undermine 1) if there are circumstances in which fairness favors a lottery over majority rule. And there do seem to be such circumstances–e.g., deciding whom to sacrifice in dire emergencies. Imagine a Trolley variation in which one has to choose a second-level method of deciding whether to throw the switch to save the 5 at the cost of one life. One’s options are: a) place a quick call to an ethical expert (a la “Who Wants to Be a Millionaire?”) and follow that advice; b) flip a coin (a la John Taurek) giving each an equal chance of survival; or c) put it to a quick vote of the seven in the situation. I think most of us would choose b), among these options.

Why then would we think it wrong and undemocratic to choose a representative by lottery? Suppose there are 20 eligible to choose one representative, and this can be done either by majority vote or by a lottery using numbered ballots and numbered balls in a basket or a wheel of fortune. Why the strong intuition that in this case the only right and democratic thing to do is reject the lottery and to let the majority rule? Maybe David is right that it has to be some non-fairness reason, because on reflection all of the objections to a lottery on fairness grounds fail, I suspect.

But is the reason to prefer majority rule really epistemic, as 3) states? Elster claims that honesty favors the use of lotteries in cases where reason, owing to “uncertainty, indifference, indeterminacy, and incommensurability,” is too feeble to decide. That might turn out to be most of the time, especially in politics. But rather than dither and debate endlessly, we call the question. Elster quotes Otto Neurath as saying “Somebody may indeed approve the majority principle only because it enhances the ability to act; it is a beloved substitute for the unloved drawing of lots.” Could it be that what makes it lovable isn’t any epistemic superiority–nor its (non-existent) superiority in the scales of fairness–but that it brings closure in what we now conventionally regard (emergencies aside) as the more decorous way?