One cool way you could spin this idea is by letting people play match maker. Imagine a Facebook application where a person could set-up two of their friends on a blind date. I think something like this could spread fairly quickly; not to mention people are more willing to accept a friends suggestion rather than a complete strangers.
Regarding looks: Sure, looks are superficial ways to recommend matches, but that's just to get the conversation started (as it is on every dating website, pretty much ever). There's also a key here that you don't get while searching on a traditional site: (in theory) people will recommend matches for you that are actually somewhat attainable rather than everyone drowning the hottest people in contact requests. This should get people closer to more likely matches than they'd get otherwise.
Looks absolutely matter-- and while it is superficial, it's clear that this is what people base first impressions on. We're not looking to be highbrow-- we're looking to help people meet those they'd look good with. I personally agree that I'd hesitate a little, but we tried to find a middleground without going overboard. We'll keep an eye on that, though-- thanks for the feedback there.
As for matchmaking: The idea is for friends to play matchmaker as much/more than random people-- after you signup, you're given a personal widget that matches you up with potential people. And you would send it to friends, embed in your MySpace profile, etc.
For example: http://www.yumbunny.com/public/3FDdUkHKSrSzDy2njcF28u/
Thanks again for the great feedback, keep it coming.
Looks like a good site idea to me, as long as they design to keep down their customer support costs.
Look at their stats: http://siteanalytics.compete.com/hotornot.com/?metric=uv
People used HotorNot for fun, not for finding real dates. I suspect the case will be the same here.
http://www.startup-review.com/blog/hotornotcom-case-study-mi...