europeforvisitors - 12:58 am on Apr 13, 2006 (gmt 0)

So, using pattern identification based on visitor time spent on a page and movement doesn't necessarily penalize the good sites, and it could likely be used to red-flag the bad ones for special treatment.

Plus, it would be only one factor. Let's say that site A and site B both have visitors leaving after viewing only one page. Site A is a dictionary, and site B is a scraper site. Site A is likely to have quality inbound links and a low SEO "spam quotient," while site B is likely to have few (if any) quality inbound links while having site characteristics that make Google's virtual nose wrinkle in any sniff test. Furthermore, site B's users are likely to be leaving via ads or the back button, while site A's users may not have such obvious departure patterns. Given the preponderance of evidence, it wouldn't be unreasonable for Google to assume that site B's lack of stickiness suggests poor quality, while site A's lack of stickiness is neither good nor bad.