While generally this approach seems like a bad idea, the kind of thing our kind of people do in their teenage years with no success and look back on with varying degrees of embarrassment like others in the thread have reminisced on, the survey to calculate priors seems like a useful and good idea. Like maybe you can't get the signal-perceiving ability of neurotypicals from books and have to get that stuff drilled in by experience, but having a prior to work with is one point where this approach might actually help you out.

But, you need to be careful about what prior you're trying to get and why, what assumptions you're making, and make the relevant parts of that clear in survey questions. The fact that people can get into relationships with strangers seems to be completely ignored, for which there could be a whole lot of possible reasons, some good, some terrible, and you need to think about exactly what your reasons are there.