A couple of days ago, I managed to combine these two interests: I've come up with a metaphor so distasteful that I think it all but precludes rational discussion.
One of the common arguments against affirmative action is stereotype regeneration: if some large fraction of minorities at a particular institution are there because of affirmative action, a modern generation of racists will develop that assumes that all minorities are there because of affirmative action, and that none are there because they would be "qualified" were they not minorities. This is the assumption that drives Kwesi Mfume and Clarence Thomas absolutely bugfuck; Mfume's response is to blame the new racists, while Thomas's response is to blame affirmative action itself.
In any case, this new form of stereotyping is mostly an exercise in botched conditional probabilities. Faced with a population whose variation on some metric is partially explained by a hidden variable, people just give up on the variable entirely and assume away the variation attributable to it. People don't deal well with r values other than 1, 0, and -1.
Having put the fallacy in this absurdly abstract setting, I then remembered where I'd seen something like it before. In some firing squads (possibly including Utah's), one of the shooters is (unknowingly) given a rifle loaded with blanks. This way, the executioners are allowed to believe that they are innocent of the killing.
Now, the parallelism isn't exact. The firing-squad issue has more to do with the difficulty of factoring causation and morality through probabilities. The squad is collectively responsible for a death; no hidden random fiddling with blanks and bullets changes this basic fact. Trying to parcel out portions of that responsibility is an open question in moral philosophy, but presumably one that shouldn't depend on the results of a die roll. And as for the legal treatment of statistical causation, it's not even coherent.
Anyway, these two scenarios do share a certain common structure. There's an identifiable group, some of whom are different from the others, in a way that's not easily observable, probably not even to them. And so the temptation arises, to attribute to the entire group the attributes of that hidden subgroup . . .
What makes this metaphor particularly egregious is that the offensiveness of the implied comparison cuts in the opposite direction from the point of the metaphor. It's pretty bad to compare minorities to executioners, but the thrust of the comparison is that the stereotyping created by affirmative action is stupid as a matter of profound irrationality.
After all, as a matter of principle, it's not so clear how much we should let important societal decisions be dictated by the known irrationality of particular groups of people. But making this argument isn't easy. I'm always on the lookout for catchy comparisons to help. If you come across any, let me know.