> EE shows that EUT only gives correct results when U = log (Wealth) You seem to think that this invalidates EUT. On the contrary, it's a vindication of EUT.

> EE shows that EUT only gives correct results when U = log (Wealth) You seem to think that this invalidates EUT. On the contrary, it's a vindication of EUT. In that particular case, EUT works and the preferences of the agent would be correctly described by that particular utility function. Otherwise you wouldn't say that EE and U=log(w) give "correct results". It can also happen in other cases that EE cannot be used to explain the preferences of the agent while EUT is still applicable because a utility function (maybe logarithmic, maybe not) can be found which describes them adequately.