tedster - 9:51 pm on Jun 9, 2011 (gmt 0)

It seems to me that the fuzziness in Google's response comes from the fact that they don't build algorithms the way we webmasters think they do - and they haven't for a long time, although up until now, our mental model was "good enough".

Matt may understand what "data" he is talking about re-running - but we don't. I assume that it's the actual algorithm building data itself, but I'm not inside Google.

Here's the question I have. Does Panda assign a "document classification score" when "it runs"? If so, is this something that cannot be changed dynamically, even by deep changes to the website content and structure?

And what was that earlier comment about internal link structures in Polish websites supposed to communicate? In other words, how does internal link structure affect the way Panda scores a website?