I figured I'd jump on the bandwagon and share some thoughts of my own. I'll be adding/subtracting stuff as I see fit. Feel free to comment, however.
It would appear that most users consider two factors when determining how they vote on RfRs; whether or not the candidate is a vandal or not and whether the candidate is active or not. The first is self-evidently important. The second, on the other hand, deserves some consideration. If rollback truly trivial, why should we consider level of activity? After all, a semi-active, non-vandal user is unlikely to misuse rollback, so there's really no harm in giving it to him, and there's always the chance that he'll be around when some new user decides to go on a vandal spree, so his having rollback would potentially improve counter-vandal response time. Based on the aforementioned criteria, however, most users would probably oppose such a candidate. So what is it about the criterion of whether or not a user is active? My assumption is that most users conflate "level of activity" with "likelihood that a user will use the rollback function." Don't get me wrong, there's something to be said for activity; the more active the user, the more they improve counter-vandal response time. However, what people seem to misunderstand is that activity can be equated with the likelihood that a user will use the rollback function if and only if that user actually spends time reverting vandalism. That said, either we should be giving every user with, say, 100+ edits and no (recent) blocks, rollback, and disregarding level of activity altogether -- and effectively saying that rollback really is utterly mundane and trivial, or we should be considering a third criterion, i.e. the actual number of reverts that the candidate has. There's really no happy medium between those two options, because considering activity without considering the number of reverts is worthless -- in fact, the inactive user who occasionally comes on and reverts vandalism will actually improve response time by more that the active user who never reverts vandalism. Given that the standard has been to include level of activity as a criterion, which indicates that people do care on some level whether a user will actually use rollback, I'm inclined to opt for the latter.
Oh, and while I'm talking about RfRs, it should be noted that there really are no other noteworthy criteria when it comes to RfRs if you broadly interpret the question of whether or not the candidate is a vandal to include whether or not they respect policy, etc., i.e. if you use it as an all-inclusive gauge for the likelihood that a user will or will not misuse rollback. Any other criteria, like whether or not the user is a good friend or really nice, are completely superfluous.
- The section on my userpage labeled "Advice for my Fellow SmashWikians."