1) There is no framework so secure that no one is dumb enough to foul it up.
2) By having to use a crazy prior explicitly, this brings the failure point forward in one's attention.
I agree, but noticing 2 requires looking into how they've done the calculations, so simply knowing its bayesian isn't enough.
http://xkcd.com/1132/
Is this a fair representation of frequentists versus bayesians? I feel like every time the topic comes up, 'Bayesian statistics' is an applause light for me, and I'm not sure why I'm supposed to be applauding.