I'm not a big fan of the 'social networking' term, though I understand it as a phrase associated with so many different web concepts that it is impossible to avoid its use by way of explanation of a concept.

I enjoyed this article by Chris Anderson and agree with many of the points involved. The general gist is that 'social networking' is a feature, not a destination.

Quite right I think. For me, 'social networking' boils down to two things, simple enough to understand:

Association. Some mechanism for the association of one piece of information to another. This could be a structured hierarchy, or a 'folksonomy' tagging system.

Proximity. A mechanism for defining how 'close' one user is to another. This could be in an obvious sense - "people who live on your street" - or something more esoteric "people who read this book went to these movies".

Chris doesn't think that there will be one platform to rule them all, and describes the many, many sites that are out there for a given domain.

I think that domain knowledge is key to this. Presenting the associations in a way that is recognisable and useful to a given domain and its audience is what makes a site successful. So a wine recommendation site like Corkd understands the particular niche needs of its users, even though the mechanics of association are the same as anywhere else. Well, perhaps it does, I'm not reviewing the success of the site itself - just making an example from a niche.

I'd assert, then, that the reason the generic 'social networking' sites are so successful is because the domain is very well understood: everyone has friends right? So given enough information and association and then surfacing a number of ways of defining proximity ("I lived here", "I'm interested in this", "I'm a ninja") then the networks form naturally and are easily comprehensible.

But things get more complex when defining proximity in a different way to 'social network'. I love my friends, but would I choose, say, motor insurance based on their recommendation, or am I more likely to trust the recommendation of strangers with other similar attributes?

Perhaps the answer lies in the wisdom of the crowd of course, but the basis for models needs to be provided by an organisation seeking to exploit the power of 'social networking' and then refined based on success.

What's the next step? Well, if you're able to create an association and proximity framework that sells your product and/or service on the basis of proximity recommendations, then you can begin to identify those who are providing influence as well as those methods preferred for users to find what they want, so the customer becomes employee, and as such can be rewarded and marketed to. You can gain a much deeper understanding of a customer based on new and emerging 'demographics' and engage in individual 'Micro-CRM' activities.

Could there be an uber-platform for this?  So, the web has all of this stuff already in terms of providing associations, and the general navigation techniques to move between data, but it probably lacks a generic way (a language if you like) for describing proximity models applicable to a domain. Perhaps the Facebook platform could deliver on this, or perhaps that is work for a successor?