You can (and should) always go for some more respected form of publishing afterwards, especially if you need points for academia. But if you were first, and you didn't document that properly, and you couldn't get it published in any "respectable" form, guess who is laughing now? If the arXiv is good enough for Perelman, it is good enough for me.
Of course, a certain amount of immunity to social values is helpful here.
One problem with this approach though is the "double-blind review" policy of some prestigious publications. If your stuff is on the arXiv, it's difficult to double-blind it. One of the reasons why I think double-blind is a bad idea.
