Geezer, perhaps it depends upon what you consider a slice of baloney. Consider this. Using the best 100 m performance by an individual, the standard deviation for the top 10 guys is 0.0675 seconds. For 11-20 it's 0.0133 seconds. In fact to get the same variability as between the top 10 outliers, you have to take the 11-50 "sub-outliers". You'd expect similar results if you did performance experiments with proveably clean lab rats. The difference between N number of subjects at the tail end of the probability curve is greater than the difference between N number of subjects under the fatter parts.
I agree that we should be making every effort to "throw out the cheating bums", but disagreeing that one can identify cheats solely on the basis of performance. You can be suspicious and it certainly makes sense to focus more testing resources on the guys winning races, but you can't damn them for winning.