Even ignoring affiliations, if I really believe I can make better voting choices than the average vote of minority X, then optimizing purely for voting outcomes means not giving the vote to minority X. And there are in fact minorities where almost all of the majority believes this, such as, indeed, children. (I do not believe this with respect to children, but I believe that most other adults do.)
Ah, but everyone thinks they know better … or something … I dunno :p
That’s just like saying “never act on your beliefs because you might be wrong”.
To be fair, that’s truer in politics than, say, physics.
Well, you want larger margins of error when setting up a near-singleton than while using it, because if you set it up correctly then it’ll hopefully catch your errors when attempting to use it. Case in point: FAI.
EDIT: If someone is downvoting this whole discussion, could they comment with the issue? Because I really have no idea why so I can’t adjust my behaviour.