I’ve been a big proponent of teaching machines by doing. I don’t mean that the machine is ‘watching’ my move and guessing what is going on, but that in the process of using the machine, explicit links are create, much like a path is warn across a park by the folks who use it.
Link: BuzzMachine » Blog Archive » The internet is the social network:
Yes, but the problem is that this relies on explicit, semantic links we just don’t use. It wants us to include rel= links when we link to someone defining the relationship. I just don’t see that happening. Sometime ago, the semantic folks wanted us to put vote links in (marking them as positive or negative); it never took off.
Here’s Brad Fitzpatrick of Google explaining the API:
I believe the killer social graph app will be the one that sniffs and understands our relationships without our having to take explicit action or by exploiting the actions we take for different reasons.
