What if I'm facebook and I'm processing >500M "likes" a day?
Could I create a reasonable search engine only using URLs people have "liked" and weight search results based on number of likes and whether people in my friends list have liked it?
That'll work great if you're searching for the latest hamster-on-a-piano video. It might not work quite so well if you're searching for something with substance. Despite the hype, most people aren't especially good at uncovering original content.
I think that sharing that sort of thing will be in a more targeted way, via e.g. yourpane.com. I never share links that don't appeal to everyone on Twitter or Facebook, but I send things to specific friends all the time.
You could, but I doubt many would use it. I agree with the above, that people compartmentalize, and they see Facebook as a place to see pictures of their friends and play games. They see Google as place to look up information.
There already is a very good search on Facebook and the fact that it is powered by Bing is unknown and irrelevant to 95% of the people. However it hasn't made a dent in search, so you have to seriously ask why that is. If you believe in Facebook as a search engine (I don't), the only real answer is because it's not powered by Likes. But to believe that, you have to believe that people have a major affinity for Likes, enough to throw away their old search habits in favor of it.
I would see the fact that the walled Facebook encompasses more and more of previously searchable content as a bigger threat to Google search. Facebook that cannot be scraped for content competes for the pageviews share with the independent searchable sites, so there's less left for Google to search.
Could I create a reasonable search engine only using URLs people have "liked" and weight search results based on number of likes and whether people in my friends list have liked it?