tedster - 11:44 pm on Mar 22, 2012 (gmt 0)
I thought that was one of the best blog posts ever put out - extremely informative.
My take on it - these are roughly the criteria that were given to human users. Those humans then established the training sets that were then used to develop Panda, via machine-learning algorithms.
Or is it that Amit Singhal has such a complex set of algorithmn arrays that not even he is able to give specifics anymore?
That's the way it is when a year's worth of intensive machine-learning and self-adjusting algorithms come into the picture. Google engineers have been talking about the eigenvalues of infinite Hermitian matrices and the like for many years.
I'm sure Amit "could" get a good bit more specific if pressed, but it wouldn't be wise to show that many cards. And when it comes down to the real nitty gritty, he'd have to research the current code-base that's in use to answer anything in fine detail.
All this is why I said in another thread that "checklist" or "punch list" SEO is no longer an effective mental model for the Google black box. If you try to understand SERP fluctuations that way, you'll be lost in mystery and probably become very angry after a while.