Can someone explain how an “always return 0.5” bot is not a counterexample?
He’s talking about a probability distribution over the set of all bitstrings. You can’t assign p=0.5 (or any other constant real number) to each of them: that doesn’t sum to 1.0.
Can someone explain how an “always return 0.5” bot is not a counterexample?
He’s talking about a probability distribution over the set of all bitstrings. You can’t assign p=0.5 (or any other constant real number) to each of them: that doesn’t sum to 1.0.