Actually, this issue also applies to probability theory. It does get more subtle, though, because theories might predict observations with differing probabiliti

Actually, this issue also applies to probability theory. It does get more subtle, though, because theories might predict observations with differing probabilities and what matters is the difference in probabilities. But this this case theory A and theory B both predict observation C with the same probability p. So you gain 0 decibels of evidence (w.r.t A-vs-B) whether or not you observe C. When I said "You're missing the 'not-PilotWave implies not-E' part!", I was making a simplified reference to the fact that the probabilities predicted by not-PilotWave (e.g. Copenhagen) need to be different from the probabilities predicted by PilotWave in order for evidence/inference to pry them apart. See [1] if that all sounded confusing. 1: http://www.yudkowsky.net/rational/technical/ http://www.yudkowsky.net/rational/technical/