enigma1 - 9:38 am on Jan 20, 2012 (gmt 0)
I was pretty darned sure they were already doing things with page layout in algorithm
Yes that's also what I see for many years now. Related to the position of the content within the page. That is still very active. In this context the main difference between automatic page processing and manual rendering is the automatic party (crawler) sees only how the HTML is setup, while the human uses a browser to translate it and he's unaware of the content position he seeks inside the HTML code. He only cares how the page visually appears.
Therefore it is next to impossible for Google to reliably tell how the content appears using the bot only. They will have to render the page and then use OCR or equivalent methods to cross-reference the layout with for a given HTML page. Theoretically is quite possible to be done and automated, in practice I don't know if they have the capacity as they would have to render and process pretty much every page. So I think for the algorithmic change they mentioned is about processing HTML code at this point.