The things described in the linked post (commonalities) are important parts of dating and attraction, but initial attraction is much more about the subtle things: Body language, subtle personality cues like clothing and hair style, all sorts of instant non-verbal communication, social proof etc. So either you have to remove these factors from the equation entirely (I can't really see how to do this) or you have to make an arena where they can be properly expressed.
This would mean, essentially, arranging real-life encounters where everybody has a decent chance at relaxing and being themselves. The thing is, nightclubs do a pretty good job at this already. The biggest solvable problem of nightclubs is differentiation; if you go to a nightclub you invariably meet the outgoing, confident, secure club crowd. It's always a high-pressure arena. But actually, relaxing and being yourself around strangers is pretty damn difficult as it is.
This is just the basic human part of the problem. But there are much deeper issues to think about. Sexuality and relationships are a very sore and difficult part of human nature, and it is never rational. There are chaotic, unmapped social issues here as well; for example a lot of women are interested in sex just for the physical thrill but have a huge social barrier against casual relationships. On the casual dating scene, this manifests itself as a series of games of playing hard-to-get, blaming alcohol for a successful one-night-stand, etc. I don't think anyone has the vaguest idea of all the other similar, irrational cultural norms that will bite you in the ass if you try to tame this problem.
Online dating is difficult because
+It is hard to convey your personality over an internet connection
+No one is able to tell you what you're doing wrong, since the rational and quantifiable part of dating is so small
+There are a lot of unmapped social and human behaviors that will fuck things up
Basically, I'd just try to tackle some difficult technical field instead. Human sexuality is too chaotic and hard to quantify.
However, I do like your idea of using an existing social network to bootstrap the thing. It is the best solution of the chicken/egg-problem I have heard.