I've written an article on what I think is the current implementation of the PageRank algorithm. Basically instead of calculating PR on a page to page basis, its on site to site links. Let me know what you think. Article here: HostRank Article It also explains why the Coop works so well.
Interesting ... The coop would also work well in the PageRank scenerio because it forces you put the adlink on all pages including the Home (root) page
I liked your article because I always felt google multiplied pagerank by the on page score to get a serp score. If pagerank is zero (no votes from someone in the google index) then the serp score is zero. 1. I am not sure you made consistent use of the words PR and LocalRanks in the latter sentences. LocalRanks now replaces PR correct? 2. Why not skip middleman page X and link to the homepage. I suspect because you need at least one link to each of the non home page non page x pages due to the way pagerank is calucated correct?
Some interesting conjectures. Some serious misinformation, though, as well -- some examples (not intended as a comprehensive list):
That's a simple fact. The whole PR paradigm is this: you have a set of N documents (pages, or sites or whatever); each document adds a PR of 1 to the set; the interlinking distributes this PR among the docs to produce their PR; That's why the sum of all PR on the web is equal to the number of all docs or 1 if it is scaled down by being divided by N (which is the same). Again. Very simple. A links to B which links again to A (common with internal pages). So, page A gets some of its PR from B but B gets some of its PR from A. When you put outgoing links on a page A, you decrease the PR it passes to the other internal links, which reciprocally return less PR to page A. Make a simple program that processes graphs and test it. It is simple. Under HostRank, this leakage is greatly diminished, because you get most HostRank from sites that don't exclusively rely on the Rank you feed them, as is in internal pages. Also, with HostRank a site adds only HostPR of 1 to the web no matter how many internal pages it has. That limits the amount of generated PR.
I'm not going to debate this with you, nohaber... it's been done to death and the only conclusion that ever results is that even if there is a theoretical basis for "PR leak" (which itself is moot) there is no practical issue. Your conclusion that "Webmasters understanding that, avoided putting links to other sites" would be better stated as "Certain webmasters worried about this myth avoided putting links to other sites". As for the remainder of the article, as I said there is some interesting conjecture and speculation there but I would caution readers against accepting any of it as anything more than that.