Recently Charlene Li from Forrester Research gave a presentation on the future of social networks at Graphing Social Patterns in San Diego, which isn’t a million miles from some of what I’ve been mulling over…
What needs to happen is that identity is in the users control, their connections over a more complete social graph need to be known to a wide array of services – with permission to access it. The social networks of today should all but vanish into a social layer of the web. She gives interesting examples of how players like Yahoo! could utilise the underlying connections between people to show more relevant content based in part on the interests or recommendations of your personal network(s).
The one thing she briefly alludes to, which is still missing in many articles is the nature of what people are calling the social graph. We as people have layers of relationships, some on a single plane such as someone you converse with on a website to a colleague you hang out with outside work to a family member or a more distant relative. We can relate to each other in multiple ways and each layer of relationship can reveal a different kind of graph.
If you take a friend, in the sense that they are someone you know and stay in touch with fairly regularly; they could also work in a similar field and potentially feature on a work-related social graph or network (think LinkedIn). While this moves slightly more towards replicating the types on interaction the two of you have, this doesn’t provide us with a clear sense of the depth of that relationship.
APML is gaining popularity amongst developers and marketers to record people’s interests or preferences as they meander through the web but maybe something similar could be employed to enhance the basic graph? If your friend volunteers through sites what they do or don’t link in books or movies and these recommendations are similar to yours, maybe their suggestions carry more weight through other services? If you message each other through services that have access to your ‘graph’ maybe these interactions could be noted to give a sense of the fact that you communicate often.
All of this leaves us with a lot of technical questions as to how we can achieve this and more to the point why. This has to be transparent to the user and more than that, people have to trust that their information won’t be abused. Unlike bank details, what is the gain for spammers or scammers?
This also illustrates that there is a limit to how far we can actually attempt to model real relationships because they are innately complex things that we ourselves can barely understand – what is the gain for us personally in having web sites/services understand us?