However, if I remember OKCupid's posts correctly, they explicitly pointed out that they weren't trying to generalize, and acknowledged that all of their conclusions applied only to the users of their site. I think it's everyone else who've tried to generalize their conclusions.

The author of the article from this thread is actually practicing skepticism. He is skeptical of the general conclusions drawn from the data set, and he proposes an alternate, simpler theory - selection bias - that would explain the behavior seen.