Goffman's 1952 Matching Hypothesis states that people are more likely to form lasting relationships with people of a similar level of physical attractiveness.

While I'm still not 100% convinced that this hypothesis holds, it has been experimentally confirmed several times, so it's at least worth thinking about. This seems to rely on the idea that physical attractiveness is objective, which I'm not sure about, but if it's not, then an average of subjective opinions may be close enough provided there's no confirmation bias in those opinions.

How does that apply to the matching on this site? Well, even if the average of ratings people have been given for looks is used in determining matching (and I suspect it's not), there is interference from so-called 'internet disease'. Definition from urbandictionary (117 up, 3 down): " People on the internet have pictures of themselves in which they look far more attractive than they really are. Quite often from odd angles, unique lighting, or a combination of both."

Clearly this means that people with a better aesthetic sense appear to be more attractive. I'm certain if it was studied, there would be a correlation which shows that art-school graduates appear to be more attractive than engineers, regardless of their level of attractiveness "in real life". But there's really no way around this short of enforcing passport-like photo restrictions, where everyone looks ugly. And that would be a shame.

It has been suggested (Huston, 1976) that the fact that this hypothesis has been confirmed suggests not that people are innately driven to seek partners of the same level of physical attractiveness, but simply refrain from choosing more attractive partners for fear of rejection. That sounds a little more plausible.

On the other hand, my last relationship lasted six years, and we got together before I even knew what she looked like because of this crazy interweb thing. Anecdotal evidence may not be admissible, but it's something to think about.