Could you please provide more details on this?
As for the delayed effects of spamming I noticed the same.
About the time of the update in November when G demoted mugshots (a good thing for the affected individuals), I put up six sites for a customer that basically had no information about the person under the name. I also linked the name to several other areas through spamming blogs so that there were more than professional pages available to work with. Most of these sites took several weeks to appear. In that same time, I was monitoring several other customers and URL's. I began to notice that the serps behind page two on a search began to look like serps I had seen before. Since I keep historical records of the first 100 links in a search for anything I track, and these records go back two years, I examined it further. After about the first three days of looking, I began to see a pattern in which the serps are being tracked backwards in time, and in which the results are seemingly randomly taken from G's historical records as G works the pages back in time. Pages 1,2...n. In turn, this appears that G is looking at historical link velocity in page placement while not affecting the first page very much. There is still some movement on the first page but this movement is not very rapid.
Then in a test case, I sent about 400 spam links from high obl, low PR (>500 PR, < 0 PR) sites that spammers use to a link that was in position 11 and that had no links, but that had also been in the serps for two years. The link began the dance within three days, jumping between position 11 and position 25. In the meantime, on the sites I created for the above customer, I sent around 2000 low OBL, high PR links to each of the sites for that customer. For this customer, there has been no serp movement.
Then I began examining the time frame for the searches and noticed that the serps are "adjusted" about six times per day. I liken this to prison head counts. I also noticed that links are now more local, based on IP.
There are many conclusions that can be drawn, but each will have to draw their own at this time.
If you are asking how I am tracking these changes, I am using my own scraper intended for ORM, which the scraper alone will be released as a free tool when it is ready. Right now I am rewriting the socket classes to overcome some deficiencies of the .net framework. Call this a preview.
