Is Facebook's approach to Search similar to Yahoo's original idea of classification based on human input? Didn't Google win this battle years ago?

A few thoughts: (a) I believe Zuck understands the power of search. Even more so, the power that comes from controlling it. I recognized this as a key neck-of-funnel strategic leverage point for FB over 3 years ago. Search, Engagement, Discovery and Recommendation are key anchor points in the FB experience. (b) 'Traditional' algorithmic-based search-retrieve-and-send constructs work with fragmented content when there is a scalable weighting approach. PageRank is one of several. Don't make the mistake of thinking that FB is a sprawling fragmented content farm. Not true. That said, FB SEO is an undisputed market that is lurking. (c) I think its clear that a combo of human-powered and algo search techniques will likely be the most successful next-gen search paradigm. This is also what powers discovery, the flip side of search, and i daresay the more important half. (d) Implementing (c) at scale is not trivial. My analysis suggests that as a percentage of 'search traffic referrals,' Facebook probably would constitute ~39%+ of Google's searches. Today. (e) The current structure where FB is working with Bing+Yahoo Search sets up a perfect pincer strike under the cover of darkness. (f) I think Cuil has the best indexing tech avail in the market today and i strongly recommend (again) that Facebook acquires the tech (and absolutely golden-handcuff Tom and Anna - tier-1 talent if i ever saw it). Folks who have worked with me know i have been saying this for over a year. Sit tight. FB-style algo-man powered search and discovery coming soon to a theater near you...

No, the idea is not to use all the data they have to simply  classify pages, resulting in a slowly updated mess of wrong or incomplete classifications. Instead, the opportunity is to use the signals contained in all this social data, such as who likes what, and who are their friends and connections, as input to a scalable algorithm capable of ranking billions of pages. Whoever reaches that goal first will have a powerful social search engine, on  a scale similar to Google or beyond and leveraging social signals Google does not have access to.

Yahoo's approach didn't scale, so no, I wouldn't say it's a similar approach other than it happens to use user input, and even other search engines base algorithms on human input such as linking to things.  Pretty much all opportunity on the internet was attached to Yahoo, and that's why it was perceived to be really valuable, but as it turns out there are a lot of competitors and they came out with better solutions.  For Facebook, the friend graph (and other graph objects) changes all of that.  Challenge has been scaling human input to reduce spam.  It's not a solved problem, but it's vastly better than it was clearly, based on the success of Facebook and it's potential.