Interesting experiment MEARS performed during last month's auction:
http://www.mearsonline.com/news/newsDetail.asp?id=758
While the point they are making about the cost of shill bidding is sound, the experiment didn't really meet it out - the only bidder who bid above the previous high was a new bidder who hadn't bid before. It seems as though they decided what the conclusions would be before they ran the experiment and when the experiment didn't exactly have the desired result, they ran the article anyway.