Earlier today, I was playing around with Snap, a new search engine. (Actually, there was a search engine by that name run by NBC Interactive during the dot-com bubble of the late 1990s, but it folded.) This engine features autocompletion, as does Google Suggest. It also shows you how many times a query has been submitted, how many times a result has been clicked on, and the cost to the advertiser (per click, acquisition, etc.), among other things. There are some drawbacks, such as it being very slow when used with Lynx, and it does not support + or - operators. Still, it's interesting to see new features. I wouldn't be surprised to see all of the major search engines adopt these features someday.
I'm still looking for technical discussion of click fraud. There was a W3C group called Web Characterization Activity that studied various aspects of web usage, but it is no longer active. In general, I'm still finding it rather puzzling that there is little technical discussion, especially within standards bodies, of something that poses a significant threat to a (presumably) promising industry.
Speaking of the dot-com bubble, the news headlines that trumpet the growing amount of web advertising, the vast sums of money that has been made, and a flocking of advertisers from older media (radio, TV, print) to the web seems very bubble-like to me.