Google is missing the thing that Facebook does well here, "identifying real people". [0] When I looked at the "Friends" google identified it was really just connections I'd made with Twitter, flickr and various other services. Very few I'd met or consider a friend. Parsing data to find friends isn't as good as being sucked into the human "social graph".
Identifying real people is a core concept at Facebook. and something they have got right (identity) and google will find difficult to replicate.
The title of this post probably mislead you, but identifying friends is just a subset of what Google is trying to do here.
If I'm searching for a good online backup service I care more about the opinions of people I follow on twitter than those of my non-technical friends and family, etc.
Good point I hadn't considered the real purpose of the service. But one thing I can't understand is why google - the kings of simple - want to include such messy data? When I see "do more" I smell failure.
Google doesn't mind messy data at all. Their goal is to get more data, even if it's not perfect. If they can get massive amounts of data, their statistical accuracy will improve.
Identifying real people is a core concept at Facebook. and something they have got right (identity) and google will find difficult to replicate.
[0] I wouldn't have realised this point as quickly had I not read "The Facebook Effect" ~ http://www.amazon.com/Facebook-Effect-Inside-Company-Connect... which despite its flaws gives some useful insights. Excerpt: http://techcrunch.com/2010/05/06/facebook-david-kirkpatrick/