Just because you and I can't imagine a social sharing site that is to Facebook what Google is to Altavista doesn't mean it doesn't exist. We don't know until the alternatives have been tried, and "yet another social sharing site" are those attempts.
No, we do know that it doesn't exist. We'd see it.
And making "yet another social sharing site" are NOT those attempts.
Just because they share the same segment ("social sharing sites") doesn't mean of them are attempts at improvement. Just as making yet another car is not an attempt to build the next kind of car. E.g. Chevy Malibu 2013 edition is merely a re-hash, not an "attempt to build something new", like, say, Tesla is.
Google was an attempt to build something better than Altavista from the moment it was still an algorithm (not even a website yet). It wasn't just some me-too web search engine that evolved into better than Altavista -- it attempted to better it in it's very design intentions.
I don't see what's that hard to understand about it.
Or did you think that when I wrote: "Creating robots that can build cars: useful" I meant as some kind of axiomatic math robot that THAT AND ONLY THAT is useful?