[yummyfajitas]

You did.

In a real life discrete time instant (e.g., an evening at the bar), most women will choose to go home alone rather than accept a second best partner. In the stable marriage algorithm, not a single woman will.

I.e., the stable marriage algorithm is a poor model for reality.

[roel_v]

But the case at hand wasn't talking about a time span that short, either. When you look at it in the long run, I'd say that both men and women have an equal likelihood of selecting a mate who is not the optimum for all qualities they look for. (i.e., to use less Greenspan-eque language, that I think that both men and woman will in the end "settle" for someone if they can't find someone who they'd consider "perfect" rather than remain alone).

However, come to think of it, I guess a prerequisite would be that there are an equal amount of men and women in the pool; which I think is so, I seem to remember from one of the okcupid analysis posts that the majority of profiles are male. Then again, on the whole, there are roughly the same amount of both, so I'm not so sure about your statement that "the stable marriage algorithm is a poor model for reality"; maybe for dating sites, but on the whole?

[klipt]

> I seem to remember from one of the okcupid analysis posts that the majority of profiles are male.

This is a significant factor. If you run a stable marriage algorithm with 400 men and 200 women:

Men less attractive than the median will remain ummatched.

A 50th percentile man will be matched to a 0th percentile woman.

A 60th percentile man will be matched to a 20th percentile woman.

A 70th percentile man will be matched to a 40th percentile woman.

An 80th percentile man will be matched to a 60th percentile woman.

A 90th percentile man will be matched to an 80th percentile woman.

In other words, all men below the 100th percentile will end up dating way below their "league".