Sometimes, blog comments really do get to the heart of the matter.
At John Battelle's Searchblog Erik has commented on Eric Schmidt:
Schmidt's comment is pretty scary because it shows he does not understand the dynamics of click fraud at all, much less economic theory. (I am a Google advertiser and an old econ major but neither of these things is particularly relevant). His argument is completely wrong for several reasons, only one of which I will highlight. One of the primary forms of click fraud is competitors clicking on links in order to drive up others' costs. Under this scenario, if I am the victim, I am being subjected to the fraud and having my costs driven up while my competitor (perpetrator) is not being defrauded and is now able to bid less for his traffic as I eventually drop out. This is not a self-correcting system. In the extreme case, if the perpetrator can do this to all his competitors, he eventually must only pay the minimum bid. Not only is this bad for the perpetrators' competitors and good for the perpetrator, but Google loses lots of money, too.
It would take too long to go into it, but under the current Google system, there is not one form of click fraud that would ever be self-correcting from an economic theory standpoint. In fact, the only scenario where click fraud would "work itself out" and be efficient, as Schmidt implies, would be if all advertisers are subject to the exact same amount of click fraud and the click fraud happens in an equal proportion on all publisher sites.