What a stressfull day! I worked on some NoiseRiver's backend improvement and I'm strating to build with only javascript and JSON some new UI elements. But the hardest of it all, was the loss of the domain name and then the re-registring of it again... Story discussed on Friendfeed --of course :)
Anyway, today I wanted to talk about the one feature that, when implemented, will make all the friendfeed gals want to mary me! True! :)
Many people are complaining about the fact that almost all social websites recommand allways the same people as friends (understand A-listers: Scoble, Loic, Arrington...). They're right, being popular doesn't mean that we share intersts! Period.
So here's the idea: Imagine, that given Paul's Interests and neighborhood settings, we may compute and find the most resembling people, those who share the maximum of these with Paul. The more people share these things with Paul, the more they're likely to be/become friends. Isn't it?
So, mathematically spoken (don't shudder!) The biggest the intersection (red zone in the pic) of two persons sets is, the most likely they're to be real friends in a social context. This is what I call so pompousely: "Smart Social Connexions". And when implemented on NoiseRiver, this will bring a new, and fresh meaning to social interractions. Where our "intersts and neighborhood" define us. Watch out :)
