(Philosophers are generally not keen on this sort of resistance, seeing it as uncooperativeness or an attempt to dodge having to make a tough decision. That's certainly one element, but I think another important feature is that intuitive moral judgment is a learned skill, not an innate faculty. So it makes some sense to think that intuitions about unrealistic hypotheticals -- ones far removed from the types of situations people have practice in navigating -- don't tell us much. This is not to say, however, that intuitions about common situations are necessarily better. If I may reference Jane Addams again, intuitions become reliable when we reflectively engage in a diversity of morally significant situations.)
My second theory -- which I tried to post but was foiled by clogged internets tubes -- was somewhat confirmed by Sinhababu's later post. The classic trolley problem dilemma is that people are unwilling to push a fat man onto the track to stop a runaway train that would otherwise kill five innocent people, but are willing to throw a switch to send the train onto a side track where only one potential victim is sitting. I had a suspicion he might be trying to separate the "immediateness" explanation of the former intuition (you don't want to do it because you're so physically close to the fat man and his death) from the other possibilities (doctrine of double effect, etc).
Sinhababu takes his theory as confirmed -- people seemed more open to pushing the scaffold-collapsing button than are usually willing to push the fat man. What I found interesting was the reasoning offered by commenters on that post. As I read it, most people offered reasons that weren't particularly sensitive to the details of the case -- either that it's always OK to sacrifice one person to save five, because more lives is better, or it's never OK, because you're actively killing the one person. If we were to hold commenters to their stated logics here, we should theoretically get the same results if we present them with either the fat man version (where typically hardly anyone will sacrifice one to save five) or the side track version (which most people are willing to do).
It would be interesting to test -- perhaps someone has already done this -- the contrast between how sensitive people actually are to details of scenarios versus the breadth of the reasons they assert to explain and justify their choice. I know there has been a substantial body of research showing that most moral reasoning is post-hoc rationalization of intuitive processes rather than reporting the actual process by which the person came to their conclusions.
(To reveal my own biases, I have a generally consequentialist viewpoint, and in particular I have yet to come upon a very convincing justification for the act-omission distinction that anti-consequentialist responses to trolley problems and related one-versus-many dilemmas (involving lifeguards, vaccines, evil executioners, etc) seem to turn on. So I always say I would be willing to sacrifice the one for the many if I were to truly be in the scenario as described. But I am extremely uncomfortable with that choice in some scenarios (albeit less than many other people), and would probably not be able to go through with it in real life.)