Yeah, of course, if the name is clearly absurd you’d use the insights from that and update your posterior, and if you have more information about the person you

Yeah, of course, if the name is clearly absurd you’d use the insights from that and update your posterior, and if you have more information about the person you can include it in the prior. The point was that 1/n over all possible names is far from a reasonable prior, and that the Bayesian update barely depends on the name itself (within reason), but on other information gained during the introduction.