> Any decent treatment of the MV states that it is the median *actual* > voter who matters, not the median *potential* voter. It's the Median > VOTER theorem, not the Median CITIZEN theorem, or the Median SENTIENT > BEING theorem.
I still think this is true but still misleading. Consider how American politicians succeed - first, they must fund raise and win the favor of party big wigs; then they must must survive a round of primaries; then they must survive the general election. We have at least three successive rounds of MVT. This suggests that policies are probably tailored to one of these three audiences. Thus, I find that arguments of the form "survey X says people hate policy Y" really miss the point. For there to be a real puzzle, you have to show how policy Y is not preferred by party activists, primary voters and general voters. Ie, you have to understand how institutions partition voters into specific groups. > The real puzzle (which I take as further evidence of voter > irrationality) is that voters use optimal punishments for superficial, > trivial transgressions like saying one wrong word, but forgive > politicians for blatantly breaking campaign promises. Bush senior lost > after breaking his "no new taxes" pledge, but it was not a foregone > conclusion. > Prof. Bryan Caplan This is not a real puzzle to me. Political beliefs are much like other kinds of beliefs, you can have enormous disputes over details but you can't challenge fundamental beliefs in obvious ways. Ie, conservatives can't flaunt violation of low tax pledges, liberals can't flaunt abandonment of affirmative actions, etc. But you can argue endlessly over the small potatoes. Violation of symbols signals not being part of the group anymore. Fabio
