This analysis has two major problems. The first is that housing ads are not a good proxy for total population using Craigslist; different cities tend to have different "velocities of housing" (how many times people move per year). He should have divided housing ads by some other housing supply metric and multiplied by population.
The second is that Craigslist is not the only channel for prostitution. There are high-end channels such as those used by Eliot Spitzer which may be more common in affluent cities, any "brothel"-type establishments that exist, and of course women on the street hollering at cars (the ability to do this may be severely hampered in colder climates).
Ok, to the first point, how do you determine the population of a city on craigslist? For example metro Atlanta has around 5 million people I think, while the actual city is ~600k. And some locations split out, such as LA and Orange County. I think the number of posts is both a good determination of population and usage rates. I'll agree though that it's not perfect, but I'm not sure there's an easy way to determine a "housing metric" for a craigslist city. It's hard to know the size of the surrounding area that uses the service.
As to the second point, the channels may differ, but a city that's affluent and embraces technology should see an proportionate increase on both categories, housing and erotic services. And LA has a great climate for "hollering at cars", while Boston should probably have the higher rate due to weather.
I'm certainly not going to argue that this analysis is anywhere near perfect, but I think you'd agree that Boston at 4% and LA at 24% is probably not just a matter of people moving more frequently.
Also, it assumes that usage of each category on craigslist increases at the same rate. Since housing and erotic services have strict separations between them on the site, it seems as if the network effects would be relatively (but definitely not completely) isolated to individual categories. Secondly, network effects increase the utility of the site logistically, so the ratios between different categories at a given point in time aren't meaningfully comparable.
This is absolutely true, however I did preface it with "The best I could come up with was the Housing category"
I'm completely open to a better metric. But as I said in the reply above, it's not easy to determine just how many people in an area use craigslist.
That being said, I'm not sure the two categories have such a hard seperation between the users. People on craigslist who use it for finding prostitutes probably also use it to find housing and vice versa.
Boy am I naive, did not know they ran prostitution thru Craigslists. I do know if you catch a businessman with a hooker he will go to great lengths to destroy your credibility and your life to protect his reputation and his hooker love. Believe me I accidently saw it and was stalked and harassed for 5 years.!!!
Wow did not know they ran prostitutes thru Craigslist but that makes so much sense now as to what happens there is so much organized crime hiding right out in the open, I learned the best place for criminals is out in the open...
The second is that Craigslist is not the only channel for prostitution. There are high-end channels such as those used by Eliot Spitzer which may be more common in affluent cities, any "brothel"-type establishments that exist, and of course women on the street hollering at cars (the ability to do this may be severely hampered in colder climates).