Procedures can be fair, but not due to tending to produce results that are fair. If two reasonable people are not agreed on who should get the last turkey sandwich, it would be fair to flip a coin to decide who gets the sandwich. But this would not be because there is some “fair owner” of the sandwich that coin-flipping tends to get right. Similarly, it seems that democratic procedure is an inherently fair way to decide issues of state governance.
Jason Brennan [AGAINST DEMOCRACY, ch. 1, pp. 10-14, ch. 5] denies that democracy is inherently fair (and, taking inherent fairness to entail non-instrumental value, denies that democratic procedure is non-instrumentally valuable). I’m not sure what he thinks of coin-flipping. Democracy, he thinks, is fair (or not) only insofar as it tends to produce substantively fair or otherwise relevantly good results. He finds democracy incompetent in this task, thinks incompetent government is unfair (because incompetent coercion is unfair) and thinks it likely enough that some version of “epistocracy” or rule by the competent (probably involving voting, but not universal suffrage and each vote counting the same) will work out better.
I think Brennan is wrong. It is almost as intuitive that democracy is inherently fair as it is that coin-flipping is inherently fair. He seems to think that there is no inherent fairness in any collective decision procedure if there is no non-instrumental value in any collective decision-making procedure. Though I agree that inherent fairness in procedure is always non-instrumentally valuable (and or perhaps because collectively obligatory or individually obligatory for relevantly-placed agents), I don’t think that the relevant sort of fairness in procedure is only instrumentally valuable. However, my main focus here is on inherent fairness in procedure (considered as a descriptive property), not the weighty, non-instrumental value (or obligation) of this. (In what follows, I’ll be concerned with value and normativity only insofar as it is relevant to the individuation of the descriptive property of a procedure being inherently fair.)
David Estlund [DEMOCRATIC AUTHORITY, ch. 4], in contrast to Brennan, argues that there are inherently fair collective decision-making procedures. But he does not think that democratic procedure is such a procedure. Inherent fairness in such procedures is properly defined in a narrow way, to include only “anonymity” or interchangeability/indifference between agents in the procedure. So appeal to inherent procedural fairness does not support the inherent fairness of democracy (however advisable democracy might be when all is said and done) – but does support the inherent fairness of making government decisions randomly via coin-flips or otherwise (however inadvisable making government decisions via coin-flip might be when all is said and done).
As I’m concerned with what inherent fairness in procedure is and because Estlund offers an account of this with considerable merit (whereas Brennan does not and often glides between inherent fairness and its non-instrumental value), I’ll sketch an account of my own largely through criticizing Estlund’s account. Then, with some results in hand, I’ll indicate how I might vindicate my above largely-intuitive criticism of Brennan.
Estlund seems to get to his conclusion that procedural fairness is “anonymity” via two premises. The first is that inherent procedural fairness is individuated in the following way: inherently fair procedure (in collective decision-making) functions to solve the problem of collectively coming to a decision acceptable to all (or to all reasonable participants) when access to generally-accepted standards for ranking the social options at stake is inadequate (so that no “outcomes-based” evaluation of decision and hence decision-procedure is possible). The second premise is a negative characterization of inherent procedural fairness to the effect that it is individuated (and made valuable) by reference to non-procedure-independent standards, not procedure-independent ones (as with fair-outcome-based evaluation of procedure). But the only kind of procedure that both does the work specified in the first premise and fits the negative characterization of the second is a procedural fairness that is constituted by anonymity (or indifference between or interchangeability of participants, including their particular preferences or interests). Think coin-flips or other randomization (or Rawls’ “veil of ignorance” way of generating norms of governance).
I’ll skip Estlund’s argument that anonymity (but not “aggregativity” such as characterizes democratic procedure or voting) fits both of these bills. My gripe is with his second premise. His negative characterization of procedural fairness is off-base (and ends up being too broad). The essential contrast to procedural fairness is procedures being fair due to tending to produce fair decisions or otherwise relevantly good decisions (or outcomes). The latter sort of standard for evaluating procedures is procedure-independent. Regardless of what procedure might produce decision, we have a ranking of options in hand, achieved via some standard. But I do not see any rationale for contrasting the fairness of fair procedure with anything other than the fairness of tending to produce better (or the best) decision (or outcome). This narrower contrast object fits nicely with the story of Estlund’s first premise. It is precisely when the standards that would make for outcome-based procedural fairness are inaccessible to reasonable agreement that we need something to do the work of inherently fair procedure.
And here’s another strike against Estlund’s non-procedure-independent characterization of the standards that determine fair procedure. It seems that procedural fairness is itself individuated by reference to procedure-independent standards. When the standards for coming to agreement regarding the evaluation of a given range of social options are inaccessible, the relevant (second-order) decision is that between types of decision-generating procedures (with respect to the first-order options). And, most plausibly, candidate decision-generating procedures are ranked relative to the costs of failing to any come to any generally-acceptable first-order decision. And that is a procedure-independent standard. So the procedure-independent (e.g., outcome-based) vs. non-procedure-independent distinction in standards for evaluating procedure does not succeed in marking out the distinction that we want it to. Fair procedure need not be individuated by reference to non-procedure-independent standards.
Perhaps there is a nice (and even required-for-some-cases) negative characterization of the relevant standards that is better. But inherently fair procedure thus characterized might not, when doing the decision-generation-mechanism-providing work of the first premise, have to be anonymous. And, more to the point, it does not seem that a procedure that both does the required work and meets the narrower, more intuitive negative characterization for evaluation (the contrast with outcome-based evaluation) will have to be anonymous. I suggest that we go with this characterization. But it is only Estlund’s broader negative characterization that pushes us in the direction of taking procedural fairness to by anonymity.
So we should feel free to take seriously intuitions that democratic procedure is inherently fair. More generally, we are free to suppose that taking into account specific agents and their preferences – say by aggregating their preferences in some proper way to produce rankings for social decision, as happens in majority-rule voting – might constitute procedural fairness in some or many cases. Of course, vindicating these intuitions or positions would require a fuller working out of an Estlund-like story about the work of procedural fairness without the sort of restriction he places on how procedural fairness could get its value.
The story here would not be as simple as Estlund’s. For instead of one feature constituting inherent fairness in procedure there are many. And they would seem to fit together like this: whichever of the two (anonymity, aggregativity) or whatever combination of the two is needed to do the relevant work in a type of context would be what constitute procedural fairness in that context.
Bringing the discussion back to Brennan: if the sort of account that I am suggesting is filled out, it might well support the intuitive idea that majority-rule democratic procedures are inherently procedurally fair. If we also grant, as Brennan does, that inherent fairness entails non-instrumental value, then his “instrumentalist” criticism of democracy loses much of its force. We might tolerate a good deal of incompetence in coming to the best decisions if democratic procedure is inherently fair and non-instrumentally valuable.
(The task of explaining why inherent procedural fairness is non-instrumentally valuable is another task. It seems plausible that, in large part for reasons grounded in the job for them that Estlund specifies, we do uniquely best when we value these procedures non-instrumentally. And being non-instrumentally valuable might come to something like this.)