Aria PN

I had an interesting meeting today.

We had results of several auction experiments that could not be explained by current auction theory. The experiments suggested that players did not always necessarily go for the highest payoffs. In fact they also cared for the payoffs of others.

REW believed these results can be explained using different utility theory, not always a different auction theory. All kind of payoffs should be considered, when estimating the utility function. Right now, especially in auctions, people only look at pecuniary payoffs. There is a new model that includes motivational payoffs that explains our results pretty well.

Then, we had an interesting conversation about utility theory. I was skeptic about the new model. If we to replace or extend something so basic like utility model, does that mean the new model should work universally as well? Or, is it enough to offer an alternative model that works only for some generic (yet not universal) instances?

REW explained how the vN-M utility theory has been accepted as given over the years, partly because it has all the nice properties and fits really well. That doesn’t mean it’s perfect. There has been some effort on refinement. But then, just like this new model, there hasn’t been any breakthroughs that change the old theory substantially.