A brilliant analysis piece on Gonzales v. Google by Tim Wu in Slate:
Google and other search engines argue—with some justification—that preserving search records is important to making their product the best it can be. By looking at trillions of search-result pages, Google, for example, can do things like offer a good guess when you've spelled something wrong – "Did you mean: Condoleezza Rice?" And Google's "Zeitgeist" feature is able to tell you what the top searches are every week and year—a neat way of tracking other people's passing obsessions. But even though keeping such logs may make their product better, or more fun on the margin, the justifications for keeping so many secrets in such a vulnerable place are just too weak.
Imagine we were to find out one day that Starbucks had been recording everyone's conversations for the purpose of figuring out whether cappuccino is more popular than macchiato. Sure, the result, on the margin, might be a better coffee product. And, yes, we all know, or should, that our conversations at Starbucks aren't truly private. But we'd prefer a coffee shop that wasn't listening—and especially one that won't later be able to identify the macchiato lovers by name. We need to start to think about search engines the same way and demand the same freedoms.
It all goes back to this basic point: How free you are corresponds exactly to how free you think you are.
Search and privacy: Danny Sullivan, Declan, GoogleAnon
Xeni on NPR "Talk of The Nation": Search Engines and Privacy Rights
HOWTO anonymize your search history
DoJ search requests: Google said no; Yahoo, AOL, MSN yes
DoJ demands user search records from Google