If you have time, I recommend watching Marissa Mayer's presentation at the Palo Alto Research Center (PARC) last Fall. She shows how Google is connecting social data (e.g., search queries) with more tangible ("meatspace") data to create predictive models that are ... well, predictive. Also significant (especially in the light of ...) are Google's efforts to expose data in ways that are more dynamic, such as Fusion Tables and Public Data Explorer (or even within basic search results - click through for more features).
Imagine where we'll be once scholars (of all persuasions) start making their data public and using tools designed for collaborative, cumulative research. Personally, I'm imagining something like a mash-up between an online Provalis suite, Public Data Explorer, and the steadily expanding universe of online data and commentary.