In fact it's pretty well established that typical levels of risk aversion cannot be explained by any halfway-credible utility function.

Yep, I've been mentioning that on LW over and over again, but people seem reluctant to accept that.