A few days ago, I proposed an analogy between blog entries and law review articles. I noted that searching the blogosphere is similar to searching the law review databases in Lexis or Westlaw in the sense that search results of both types are largely content-driven rather than reputation-based. Only relevant content, not past popularity, can generate search hits.
There is, however, a distinction not discussed in my earlier post that is worth mentioning. As Dave Hoffman points out, searching the blogosphere is not entirely a reputation-neutral process because Google PageRank makes reputation a factor in ranking search hits. Of course, pages will not come up in a search if they don’t contain information matching the search query. But where (as is usually the case) a search locates many pages containing search-relevant content, PageRank determines each page’s relative importance on the basis of reputational factors such as past traffic and links to other pages. A page’s past can thus affect its future and in this sense blogosphere searching is to some extent path dependent in a way that law review searching is not.
Nonetheless, I remain convinced that blogosphere searching is largely content-driven, because PageRank apparently ranks individual blog posts rather than the blogs on which the posts appear. That is, PageRank is page-based, not site-based. A “new blog� posting that gets linked to and visited many times will receive a higher rank from PageRank than a seldom visited posting that appears on what is overall a better established and historically more popular blog. So at the end of the day, content is king, a reality that PageRank reinforces rather than undermines.
An interesting question that emerges from all of this is why Lexis (which currently ranks law review search hits alphabetically), Westlaw (which currently ranks law review search hits chronologically), and other electronically searchable databases don’t either license Google’s PageRank technology or homebrew a similar ranking mechanism if Google refuses to license it or demands too high a price. Perhaps efforts to do this are already underway, but certainly they should be. In any case, I predict (as Tom Smith implicitly suggests) that in the not too distant future, relevance ranking will be a universally available feature of electronic database searching.
RSS - Posts
Keith: The Google PageRank value on this blog is currently 0. Any ideas how we can boost it?
Great post. Here are my questions:
PageRank presents a particular method of aggregation, and PageRank values are like prices, conveying some information about the market’s collective valuations. Chronological or alphabetical results are not aggregated in any fashion that itself conveys information; they are neutral. So are there reasons to be skeptical of the former? Is Google’s particular method of aggregation unsuitable for certain purposes (like, say, legal research)? Are there circumstances where a neutral presentation is, in fact, more valuable? (Obviously, Lexis and Westlaw could make results available in either fashion, which would overcome some of the problem, if it is one). And even if the particular method of aggregation is imperfect, can individual searchers compensate effectively, such that imperfectly-aggregated results are superior to neutral ones?
See, generally, my dad’s posts at Ideoblog on the theory of price formation (http://busmovie.typepad.com/ideoblog/2006/01/more_from_henry.html and http://busmovie.typepad.com/ideoblog/2005/12/toward_a_theory.html).
Keith — When I use the Google toolbar, there’s a page rank for my blog but no page rank for each individual blog post. On the other hand, in doing particular searches in Google, specific pages pop up more often than the blog’s main page itself. So perhaps Google is one step ahead and has a ranking for the main blog site plus a different ranking (perhaps based in part, rather than in toto, on the main site’s pagerank) for sub-pages such as each blog entry. In other words, perhaps it is both site-based and page-based.
And there’s a flip-side to your interesting post — perhaps the kind of searches possible on Westlaw / Lexis might become available on the Web.
Bill: My info comes from http://www.google.com/technology/, Google’s page explaining PageRank, which states: “PageRank relies on the uniquely democratic nature of the web by using its vast link structure as an indicator of an individual page’s value. In essence, Google interprets a link from page A to page B as a vote, by page A, for page B.” Links aren’t the exclusive factor; visits matter too, and in all likelihood are probably more important than links (though Googe does not reveal the precise weightings that its search algorithms assign to the various factors relevant to search rankings). My point is simply that for purposes of ranking search results, it’s the quality of the individual blog entry that matters, not the quality of the blog site on which it appears.
Keith: What’s the basis for your statement that “PageRank apparently ranks individual blog posts rather than the blogs on which the posts appear?” I haven’t heard this before, and my posts at business law prof seem to rank fairly high in a google search of words in the post title regardless of whether anyone linked to the posts.