Well, new information could suggest that you applied the wrong prior. If you’re trying to calculate the likelihood of a person’s guilt, and you find that the crime is more common than you originally thought, you’ll have to assign a higher probability of guilt, but you could reasonably say that it’s not evidence of guilt, because you’re revising P(A) rather than introducing a new (B).
This could be said to apply to the mention of the Chandler case, if you don’t have any information on how common such crimes are that would screen it off, but it clearly does not apply to the porn.
Well, new information could suggest that you applied the wrong prior. If you’re trying to calculate the likelihood of a person’s guilt, and you find that the crime is more common than you originally thought, you’ll have to assign a higher probability of guilt, but you could reasonably say that it’s not evidence of guilt, because you’re revising P(A) rather than introducing a new (B).
This could be said to apply to the mention of the Chandler case, if you don’t have any information on how common such crimes are that would screen it off, but it clearly does not apply to the porn.