Interesting. Network analysis in the study of markets for sex-for-sale.
I was puzzled by this passage, though:
Another discovery is that a high rating for a particular sex worker is a good predictor of high ratings in the future. That's the kind of rich get richer effect that is seen in many internet phenomena (also known as the Matthew effect). However, average or poor ratings don't seem to affect future ratings either way.
Naturally, buyers tend to use more highly rated sex-workers more often. And over short timescales this can be seen in the data. However, look at longer timescales and the effect drops away. That's probably because sex-workers do not stay in their work for long periods of time, say Rocha and co.
Well, I would have thought that the differences would disappear as prices adjusted to offset quality differences. In fact, I would have predicted that price differentials would would have increased until the marginal consumer of the service is nearly indifferent (holding income effects constant) among different providers. Sure, high quality service is likely a luxury good, but I would still expect that price differences (in effect, rents for differences in attractiveness, effort, and talent) would have explained why top-ratings don't necessarily translate into greater quantities of transactions.
On the other side of the argument, of course, there is the old claim attributed to Napoleon: "In war, as in prostitution, an enthusiastic amateur may outstrip a professional."
(Nod to Angry Alex, who never has to pay for it)