In general, I agree that reasoning about approximations to SI is trickier than reasoning about SI itself. But I don’t see how we can confidently say things like “any method of approximating SI will have such-and-such problem”. Such statements seem to be caused by limitations of our imagination, not limitations of SI. I propose to wrap up this discussion and agree to disagree, at least until we come up with some hard math about the limitations of SI.
I were originally speaking of vanilla SI, not of some not yet defined methods.
And I don’t think it’s even a limitation of SI. It just seems to me that the code vs data distinction is, in the case of machines that can run interpreters * , entirely arbitrary, subjective, and a matter of personal taste.
and not only can run interpreters, but ultimately should, if we are to represent a world that contains computers.
In general, I agree that reasoning about approximations to SI is trickier than reasoning about SI itself. But I don’t see how we can confidently say things like “any method of approximating SI will have such-and-such problem”. Such statements seem to be caused by limitations of our imagination, not limitations of SI. I propose to wrap up this discussion and agree to disagree, at least until we come up with some hard math about the limitations of SI.
I were originally speaking of vanilla SI, not of some not yet defined methods.
And I don’t think it’s even a limitation of SI. It just seems to me that the code vs data distinction is, in the case of machines that can run interpreters * , entirely arbitrary, subjective, and a matter of personal taste.
and not only can run interpreters, but ultimately should, if we are to represent a world that contains computers.