In my paper PageRank Explained I claim that “Chris Ridings of www.searchenginesystems.net” has made a mistake in a paper on the same subject (see addendum below).
I didn’t want to labour the point in the main paper as I have nothing in particular against Chris but because several people have asked for a clarification, and in the interest of open academic critique (Chris does invite that in the first paragraph of his paper), I’ll describe the problem here.
Chris’ paper has roughly the following structure:
- Title: “PageRank Explained”, sub-title: “Everything you’ve always wanted to know about PageRank”
- Introductory discussion of Google and PageRank.
- A quote from the Brin and Page Google paper, including the original algorithm.
- Introduces, for illustrative purposes, a ranking calculation called MiniRank to quote: “which is very similar to PageRank. This should help us understand it.” [PageRank]. The terms MiniRank and PageRank are freely mixed in several sections thereafter though.
- Gives advice to webmasters about how to structure links in real web sites based on observations of MiniRank behaviour.
This would be fine, and some of the more general comments in the paper are useful, but a fundamental flaw in the implementation of MiniRank means that it models an equation very different from Google’s PageRank! Most of the conclusion in the paper, therefore, are not applicable to real websites.