Facebook has made a few very important announcements this week.  It is now clear that they see themselves as a part of the software backbone of the Internet, with a increasingly-dominant sign-on and authorization ability, the new, atomized "like" button, and promising ambitions in the location and payments areas.

And, as Albert Wenger points out, they have executed very well on this very ambitious path, continuing to innocate at scale.

However, there's something I find disturbing in the dominance of any one company, at the core of the internet, which itself, has so far avoided this type of dependence, maybe except in the case of google.  However, Google, in its most dominant search and advertising verticals, does not constitute a single breaking point for the internet.  Take away Google Search, and Yahoo or Microsoft have products that work almost as well.  Take away Google Ads, and… well, nothing maybe.  Probably a massive drop in traffic to many sites and a commercial problem, but the internet works just as well, perhaps better, in providing us with the utility we expect from it.

Conversely, today, nobody else has the social graph data that we now rely on Facebook to provide for us as a web service.  And now many new services are being built on this layer.

If you read this blog, you know that I find the social graph extremely valuable.  I suspect that the single company domination of the social graph is dangerous and may have a stifling effect on the development of value-added services on top of the identity layer.  The strength of the internet has so far been its openness.  Had it been a Microsoft net or an AT&T net, no matter how innovatively executed, the internet would have fallen way short of what it enables today.

What would the alternative be?  If Sun had made the move to own the social graph, it would have been more authentically open. So far, Facebook has been somewhat open, but there are commercial pressures around Facebook that will strongly motivate it to misbehave in its shepherding of the social graph.  I hope that does not happen.

PS. Albert also declares he does not believe in a single social graph.  I actually do. I think the problem he points out is one that can be solved with intelligent search/find and filtering.