On OKC star ratings are not exposed to users - you get to rate someone, and they get to rate you, and those are (to some extent) visible, but there's no such thing as "average user rating" or anything remotely close to it.

> " My understanding that any woman who gets an extreme number of messages will have a red dot"

Except... every guy has a green dot and nearly every girl as a red dot. This is not a salient indicator of anything - for all you know she's just very particular, but doesn't get a lot of messages (I've certainly seen profiles like this). Or, she might be getting literally hundreds of messages a day. Or hell, maybe she's getting spammed with one-liners but never gets a real message. In anycase, whatever the reason, when 90% of your population has a red dot, it ceases to become a filter for anything.

The "replies selectively" messaging on the site is borderline useless, and certainly doesn't accomplish the competition signaling that would actually work to drive some traffic from one group to another.

Except people are unlikely to give someone a one or two star rating, and are likely to give four or five star ratings.

At least for women rating men on OkCupid in 2009, that's wrong. http://alvanista.wordpress.com/2009/11/27/another-okcupid-st... "Women on OkCupid have rated 80% of the men as unattractive." "The median man averaged 1.5 stars out of 5"

Also, all 1, 2, and 3 ratings are seen only by the person who made the rating so there is no penalty for rating someone low.

But it doesn't matter if the ratings are distributed evenly; there will still be some variance. Users who pay to have "A-List memberships" can use those ratings to search for matches of at least {3, 4, 5} stars to restrict their matches to the more heavily sought.