tedster - 6:01 am on Sep 6, 2012 (gmt 0)
I am not sure the algorithm has yet to tell the difference between keyword stuffing and a decent reuse of keywords.
I'm sure there are still false positives from time to time. But I think they're doing a pretty good job ever since they created their phrase-based spam detection process [appft1.uspto.gov].
As I read that patent, it means, for one, that keyword spam detection will end up DIFFERENT for specific keyword phrases (certain markets and website taxonomies.) So the stuffing threshold will be set by a statistical values that are seen across a large number of sites in any given niche. Articles will see a very different threshold than point-of-sale pages. It just the way that process works.