The venerable Professor Scott Sumner has criticized anti-utilitarian hypotheticals very many time lately, writing, in but one instalment, that:
"If opponents of utilitarianism are forced to come up with implausible examples involving cognitive illusions to make their point, then that suggests to me that utilitarianism is a quite useful system."
I have written here before that the criterion for hypotheticals to be valid is that they not be impossible by the standards of science. So if something is at all possible, it may be safely used in thought experiments.
Harvesting organs from one person in order to give them to many other people, thereby saving many lives on net, is certainly possible on a small scale (it is likely impossible on a large scale due to the fear and concomitant disutility which widespread harvesting would induce). Thus, the utilitarian conclusion must be that a doctor should occasionally kill his patients (particularly if they are unhappy and have no friends). But rather than really tackling this issue, Sumner instead argues for organ markets, which of course would be a lot better than killings, but what can the individual doctor do to make that happen? Moreover, some patients in need of organs might not be able to come up with the necessary funds.
Ignorance of utility-reducing facts stands opposed to Truth, another valuable thing which one may not want to trade off for (just any small amount of) utility. For instance, a man whose wife is unfaithful has been raising a number of kids who are not his own. If the wife tells him, he might be unhappy for the rest of his life. Of course, a consistent policy of honesty and non-cheating may do more for utility, but once these things have happened, utilitarianism seems to sacrifice a great deal of Truth for utility. This leads into the discussion about Robert Nozick's Experience Machine, in which no lived ideas are really true.
In a comment to Professor Sumner's post, I wrote that "homosexuals are a very small fraction of the population and are also quite unpopular in certain parts. One does not have to imagine death camps to make points against utilitarianism - would additional taxes be justified so as to compensate the heterosexual majorities for "putting up with them" (or perhaps discourage the activity)? More lenient sentences for crimes against gays than for crimes against others?" This is another instance of a plainly-not-implausible scenario in which utilitarianism's failure to acknowledge the individual's dignity is exposed. So why slough over it rather than engage in the hypothetical?
Here is another short hypothetical: A person, call him A, finds a perishable good in a place where no-one will set foot in weeks or more (by which time the good has gone bad). Maybe he is hiking in the Himalayas or is in Chernobyl just before the 1986 disaster. Anyway, A could consume the perishable good or leave it for absolutely no-one (it rots or becomes poisoned). A would gain a little bit of utility by consuming the good, and if he abstains (global) utility will remain unchanged. If one considers this an "implausible" hypothetical, try many "minute-wise, hour-foolish" strategies instead and notice the plausible prevalence of sacrificing own utility.
According to utilitarianism (and Ayn Rand's philosophy known as Objectivism), A is behaving immorally if he does not consume the good. Non-consequentialist moral theories, on the other hand, would claim that A's decision lacks any moral importance. The same holds whenever a utility-increasing action is abstained from which would have no impact on anyone else. Now of course there would be a very strong tendency for individuals not to abstain or to be "hour-foolish", but why should that be considered immoral? Is it not their prerogative?
However, the best arguments against utilitarianism come, I believe, from introspection. If a fixed amount of utility in a lifetime truly can be distributed in any way, then perhaps nothing but utility matters, but from introspection I believe many individuals would not be indifferent; for example (1, 1, 1) may be thought better than (103, -50, -50). Perhaps these non-indifferent individuals are wrong. In that case, does it follow that public policy should force them to act differently?
Now it is clear that none of this refutes utilitarianism. But perhaps this blog post manages to show that hypotheticals arguing against utilitarianism need not be extreme or implausible and may anyway be worthy of more respect than they frequently get.