More abstractly, "costly signaling" theory requires that it should be hard for low-fitness individuals to send a fake signal. Clearly this isn't the case here - anyone can pretend to spout weird antisocial nonsense.

I disagree; I think this is an excellent use of the handicap principle (though it remains to be seen whether it's actually true). Any gazelle can go stotting, too, if by "can" you mean "it is physically possible for them to do so". But only a gazelle very confident in its speed actually will, because all the other gazelles are too worried about being eaten by lions to dare to try it.

Likewise, anyone can claim to like weird foreign films on MySpace, but only a person very confident in her popularity actually will, because all the other people are too worried about being shunned.

The number one objection I have to this idea isn't that it's not evolutionarily plausible, but that it doesn't fit observed data: it's not the high school quarterbacks and cheerleaders who are liking weird foreign films, it's kinda weird people who aren't popular anyway: anime is the most obvious example. Any further argument for this idea would have to explain why it goes so terribly wrong.