the term ’social networking’ makes little sense if we leave out the objects that mediate the ties between people. Think about the object as the reason why people affiliate with each specific other and not just anyone. For instance, if the object is a job, it will connect me to one set of people whereas a date will link me to a radically different group. This is common sense but unfortunately it’s not included in the image of the network diagram that most people imagine when they hear the term ’social network.’ The fallacy is to think that social networks are just made up of people. They’re not; social networks consist of people who are connected by a shared object. That’s why many sociologists, especially activity theorists, actor-network theorists and post-ANT people prefer to talk about ’socio-material networks’, or just ‘activities’ or ‘practices’ (like I do) instead of social networks.

It’s indeed very relevant to connect this to the activity theory since it underline the very notion of ‘objects’ or goals that people want to achive by performing an activity (in this context, registering, logging and linking).
He quotes examples of object-oriented social software that are obviously successful like:

Flickr, for example, has turned photos into objects of sociality. On del.icio.us the objects are the URLs. EVDB, Upcoming.org, and evnt focus on events as objects.

He also points the problem with FOAF:

Sometimes the ’social just means people’ fallacy gets built into technology, like in the case of FOAF, which is unworkable because it provides a format for representing people and links, but no way to represent the objects that connect people together.

And even though I agree with this from the activity theory point of view, I would argue that FOAF should not be seen as a service but instead as an architecture on which designers can rely to design services. It might be possible to design object-based social software using the FOAF XML grammar.
Other interesting things about this in his post!