jmccormac - 7:00 pm on Mar 17, 2012 (gmt 0)
So it is an inversion of the infinite monkeys scenario where they may some day write the entire works of Shakespeare? Those training questions are effectively conditioning reviewers into thinking that every site should be like Wikipedia. Wikipedia In Wikipedia Out. The web is far more diverse than this simplistic approach would credit. If these reviewers and Google's machine learning go looking for sites that resemble Wikipedia then that's what they are going to promote in the SERPs.
I disagree. What Amit is listing in that blog post are the kinds of training questions they used at the beginning of Panda - the one's they gave to human reviewers to build the seed set of sites that they turned their machine learning loose on.