Thanks to Loren for the great summary and questions. Loren is largely sympathetic, so, as a reward, this can be more brief than most weeks. But he does raise two concerns.
First, Loren proposes a way of avoiding the disjunction problem. He says that we might suppose there is one correct way of enumerating the alternatives, leading to n alternatives, and then take individual competence to be above 1/n. “After all, we might reply that the assumption of slightly better than random voter competence presupposes correct specification of the choice problem.” I’m not sure what a correct specification of the problem would mean. Suppose we are deciding about building a bridge. We could build a cheap bridge, an expensive bridge, or no bridge. Suppose the best thing is to build a cheap bridge. If we ask the voters to choose between these three options, random competence would be 1/3. If we give them two choices: Build a bridge (cheap or expensive), or build no bridge, random competence is .5. I don’t know what it would mean to say one of these is the correct way to put the question. In any case, Loren says that even if there is a correct or privileged way to enumerate the alternatives it would remain unclear who should get to make that decision. I don’t know how to evaluate that point because I don’t get the idea of a privileged enumeration. For all I can tell, if there is a privileged enumeration it might be beyond reasonable objection. So I might benefit from some clarification of this suggestion.
Loren’s second concern is that there could be a persistent minority (call them the vegans for short), and in that case it might be hard to expect them to accept that the system performs better than random overall. Even if it performs better than random on (what I call) primary bads, members of this minority might well think that it is wrong on every single other thing, giving them reason to doubt that it is better than random overall. To evaluate this worry we could generalize it in this way: suppose I am a person who typically disagrees with the decisions of my community, even those bearing on the primary bads. I accept, of course, that they are very bad, but I disagree about how to address them. It might seem that I could not accept, at the same time, that the system tends to do better than random. Am I forced to decide between either deferring to the system’s conclusions or denying that the system is better than random?
This is closely related to my discussion in the section called “The Deference Objection.” I’m strongly disinclined (for somewhat mysterious reasons, as I say) to think that beliefs about some procedure’s epistemic value are sufficient to ground (epistemically permit, or require) changing one’s moral judgment to agree with the procedure. So I agree with Loren that the vegan sect needn’t change its views simply because they are in conflict with democratically produced decisions. How, then, can they still think the democratic procedure does better than random? It seems, to them, to get so many things wrong.
We could beef up the conception of primary bads and try to argue that they are of such overwhelming importance that no procedure that does significantly better than random on those could be worse than random overall. I haven’t tried that. I prefer to suggest a list of primary bads that supports the view that a procedure that is good on those is likely to be good on many other issues that will come up politically. This is where Loren supposes the vegans balk.
So here is the position the vegans are placed in on my view: They grant that the system is significantly better than random on the primary bads, and their own political views don’t lead them to any doubt about this. But they find themselves disagreeing with democratically produced decisions in almost all other cases, and so they aren’t sure whether they can grant that the epistemic value spills over from primary bads to other issues. If they grant this, then they are under some pressure to think that their own views on those other matters must be worse than random. I say only that they are under some pressure because they could possibly hold that the period of time in question is statistically anomalous and that the procedure will tend to agree with them more often over the truly long run. But this might be a weak suggestion if the vegan creed goes back many generations and the conflict with political decisions has always been about the same.
Another possibility remains. They can believe that they are indeed worse than random, but since they don’t know which specific instances have been right and which wrong there is no particular basis for deference on any issue. I’m not sure I see any problem for my view in the fact that it might put some voters in that position. If they had good reason to think they were indeed worse than random then it’s true that they should realize that could do better by simply choosing randomly. In practice, however, they might well teeter between thinking they’re in a statistically anomalous period and thinking they are worse than random, each of which would be a qualified view.
Keep in mind that the congruence of the collective decisions with their own judgment is not the only basis they have for judging the epistemic value of the collective procedure. There is also the argument, which we are assuming they accept, that the procedure does significantly better than random on primary bads, plus the argument that good performance on that list is likely to translate into good performance on other issues.
It’s doubtful that it should be counted as qualified to deny the procedure’s epistemic value solely on the basis of your own disagreement with its conclusions. The reason is that there is that other general case for its epistemic value, and something more than one’s own disagreement would be needed to address that.
Finally, people who have deep and persistent disagreement with the democratic procedure’s conclusions will be the most tempted to deny the epistemic claims my view requires. But if the system and circumstances are good, we could expect many people in this position to be so persistently in dissent partly because their views are seriously deficient, and potentially outside the bounds of the qualified.
Having said all this, there’s no intention of ruling out in advance the possibility of a point of view that must be counted as qualified, but which disagrees with the majority so persistently and over so many years that they would not be unreasonable to conclude that the procedure is worse than random. It’s always good to be clear about what, exactly, would have to be the case to prove a theory wrong. That would be one way.