That's what Amazon and Netflix have done. They should solve the data problem first, and being yet another "interactive social media mashup web 2.0 social networking thingy" may not be the best way to get lots of data. The companies that have done this best have monetizeable data - Amazon can actually SELL you the book it recommends.
As for being a collaborative filtering platform: "1. If your idea starts with 'We’re building a platform to ...' and you don’t have a billion dollars in capital, find a new idea. Now." (http://diffle-history.blogspot.com/2008/06/postmortem.html)
In fact, they completely suck. And in reply to GP, netflix's recommendations suck very much too, to be frank.
Why don't companies care? Simply because it's hard to figure out how to make a lot of money even with very good recommendations.
It's a python shop running either on a custom DB or on Berkeley DB. Their core business asset is the collaborative filtering mechanism that drives the Redux.com site.
I got the impression that the site itself is very much a proof-of-concept application of the underlying technology. They have a BHAG to become the default collaborative filtering mechanism for the internet. If they are successful, I consider it very likely that they will do so by licensing this technology to other companies.
It should also be noted that, as a company, they did not start out in the direction that they're going now. They've gone through several products and switched directions multiple times, eventually settling upon what they're developing now as a result of that process.
If I remember correctly, they are 11 employees, 8 of which are engineers and the CEO is a definite techy.
Smells like win to me.