When Google luanched Open Social, many hailed the announcement as the end of walled-garden social networks, and the arrival of cross-network connections. While that enthusiasm may be too early, it is certainly going to be an important step forward.

What open social really addresses, though, is application portability not social graph portability. This will be more valuable to application developers than to users, who will benefit only indirectly.

Open Social is a basically a standard – it describes an API which applications can use. Any "container" (a social network platform, like Google's Orkut, or Ning, or Plaxo, or Friendster) which implements the Open Social Container API, can host any application (like those increasingly popular on Facebook) which implements the Open Social Application API. 

It's like a plug-in infrastructure that works across different applications. Open Social compliant applications or widgets will run in any Open Social compliant container. They get access, in that context, to informaiton about the user and his/her social network – essentially connections to friends.  

What the infrastructure doesn't provide, however, is any mechanism for reconciling different accounts in different networks, or allowing me to take my set of connections on one network and simply apply them to each new network I join. For that, we'll need much better cross network business collaboration, identity standards, and permissions based exchange formats. 

In short, open social will make it easier to create widgets that run in multiple networks, but won't make it any easier for users to connect across networks.