[eli_gottlieb]

>I think the lesson here has been evident for some time: There is no one best classifier, only classifiers that perform better on particular problems.

Isn't that called the No Free Lunch Theorem?