Likewise, the search feature yields mysterious results; a search for “volokh” in all federal cases yields lots of cases that never mention “volokh,” nor anything that seems similar to volokh. And Google’s famous metadata glitches make date searching iffy, since some very new articles end up having old dates. Check out, for instance, the prescient 1761 source on computer file-sharing yielded by this query.

Google is more based on folksonomy than on raw indexing. That is, an article is important on the subject “volokh” because thinks interest in “volokh” also reference this article. There is also some evidence these rankings are not limited to the public web. Google relevance rankings are thought to include messages in Gmail as if they were web pages, although the messages themselves do not get appear through search result links. Some claim to have brought pages into Google rankings with no cross linkings solely with an email distribution campaign.

One thing that Google has always eschewed is formal metadata, at least any they would have to create. Cross-referencing a law document by material not formally in the document itself would seem contrary to their usual approach. The1761 article is an example of inferred rather than entered metadata.

I inquired recently because I started getting a warning screen everytime I tried to log into Westlaw.

“Warning! You have already used 90% of your account hours.”

I was told West limits student accounts to 180 hours of usage and 600,000 lines of printing per month. Thankfully, it was only a week until the end of the month, so I didn’t have to go elsewhere for my legal research fix.

I’m still leery of Google’s business model once they reach overlord status. How much is it going to cost in 10 years is the better question. Nothings free.

“I got my eye on you Google!”

BTW what’s with the apparent spam?

But again, I assume there are many more improvements to come in the coming years, and perhaps even one day support for Lexis– or Westlaw-like queries, with good proximity searching, segment searching, and more.

I doubt Google will ever offer support for Lexis/Westlaw-style boolean queries; that’s not what Google does. With Lexis and Westlaw, you have to construct your query extremely carefully or you’ll either find nothing or be buried in a mass of irrelevant results. Where Google has the potential to add value is to produce an algorithm that gets you the most relevant results at the top without needing a to construct a query with such precision, the way it does with web search. If Google Scholar can do that it will blow Lexis and Westlaw out of the water.