I said the programs have a single output, instead of a probability distribution. It matches the sequence so far or it doesn’t, maybe programs are 100% eliminated or 100% not eliminated. The probabilistic nature comes entirely from the summing-over-all-programs part.
If programs can output a probability distribution then clearly the program “return {0:1/6, 1:5/6}” will do very well and cause the results to converge appropriately.
I said the programs have a single output, instead of a probability distribution. It matches the sequence so far or it doesn’t, maybe programs are 100% eliminated or 100% not eliminated. The probabilistic nature comes entirely from the summing-over-all-programs part.
Yes, I’m using the same definition. The “deterministic programs” mentioned in my comment are programs that output a sequence of bits, not programs that output a probability distribution.
That definition is equivalent to a mixture of all possible computable distributions. I suppose that equivalence is an important and surprising fact about SI: how come it makes no difference whether you quantify over programs that output a sequence of bits, or programs that output a probability distribution? But yes, it is true.
That’s not the same definition I was using.
I said the programs have a single output, instead of a probability distribution. It matches the sequence so far or it doesn’t, maybe programs are 100% eliminated or 100% not eliminated. The probabilistic nature comes entirely from the summing-over-all-programs part.
If programs can output a probability distribution then clearly the program “return {0:1/6, 1:5/6}” will do very well and cause the results to converge appropriately.
Yes, I’m using the same definition. The “deterministic programs” mentioned in my comment are programs that output a sequence of bits, not programs that output a probability distribution.
That definition is equivalent to a mixture of all possible computable distributions. I suppose that equivalence is an important and surprising fact about SI: how come it makes no difference whether you quantify over programs that output a sequence of bits, or programs that output a probability distribution? But yes, it is true.