But, I've got a few ideas about how to do some initial data mining to build a useful-yet-niched database. I'll target a specific industry and, I'm guessing, enough people will value their skills that they'll be happy to encourage their friends and colleges to rate them. The hope is that this can be done quickly enough to generate sufficiently many data points for accurate analysis without the initial dataset.
Once that happens, I'm banking on the utility of the system. If it's as good as I hope, it'll be adopted by other groups. If it's not, then it should fail.
As far as the grossness factor and potential for abuse goes; I'm protecting user information as much as is humanly possible -- at least, to the limits of my own ability. The system will be open sourced and as transparent as possible. Unfortunately, I can only guarantee so much. I can't think of a way for users to verify what code is actually running on the server -- but, they can ensure that the data they are sending to the server is exactly what they expect. But, there is no certainty that the server isn't recording their IP and associating it with the input for future data analysis. Hm.