It seems that the real test of whether or not it's useless pseudoscience is in the empirical evidence, not the narcissistic ramblings of a out-of-touch former stockbroker. Kosinski and Stillwell managed to make a number of predictions out of the models; it's a step from there to political ad targeting. There might be more effective methods of political ad targeting--maybe some steps could be skipped--but that seems to be an empirical question more than anything.<p>It's all bits of information in the end and how you use them. There's nothing "metaphysical" about them.<p>As for companies making overhyped empty promises? Not unique to political science.<p>I'm tired of this kind of imperious bigotry from overpriviledged silicon-valley types. It helps no one.<p>There's an important question about whether or not this stuff worked, but here we are, with Trump as the president when all the models predicted Clinton would win. That's not exactly an experiment I want to try again.