whereas the Solomonoff prior treats it like a random string and would converge to answering .5 for the probability of the next bit being 1.
Isn’t something like “If you have access to a halting oracle, ask it for Chaitin’s Ω” among the computable generators included in the Solomonoff prior? If so, the Solomonoff agent should converge to the correct sequence.
Isn’t something like “If you have access to a halting oracle, ask it for Chaitin’s Ω” among the computable generators included in the Solomonoff prior? If so, the Solomonoff agent should converge to the correct sequence.
No, at least not in the usual formulation of the Solomonoff prior that I’m familiar with.