… Um, the default assumption is that any given hypothesis is wrong, you can’t get your priors to converge otherwise. Omnipotent intelligent beings are sufficiently complex that I’d need a few megabits in their favor before I gave them parity with physics.
Um, the default assumption is that any given hypothesis is wrong, you can’t get your priors to converge otherwise. Omnipotent intelligent beings are sufficiently complex that I’d need a few megabits in their favor before I gave them parity with physics.
This given hypothesis is wrong. A megabit means a 2^1,000,000+ odds ratio.
If you can specify physics with n bits, and pick out a narrow class of humans or computers or books with ~k bits, and humans can produce an algorithmic description of a world with an omnigod, then you can specify the omnigod with ~n+k bits.
I don’t think it takes a megabit to specify a world like Permutation City for an intelligence to be essentially omnipotent in. Specifying intelligence along the lines of AIXI with unbounded computation requires few bits.
Point. I was thinking in terms of “this particular intelligence,” and whatever an AGI looks like I will happily offer 999:1 odds it will take more than a megabit of disc space. But if you just want “an intelligence,” yeah, not nearly as much.
… Still would need an awful lot to compete with physics, though.
To clarify: This is because any given hypothesis has a prior proportional to 2^-(complexity), which for any reasonable hypothesis means ludicrously low odds of being true.
Then you update on evidence, and just seeing things gives you megabits to work with, so that goes up significantly for particular hypotheses, but… the point is, even aside from evidence something like “God exists” is starting at a massive penalty compared to physics.
No, that is not the default way to handle a hypothesis. The default is to ask: why should I believe this? If they have reasons, you look into it, assuming you care. If the reasons are false, expound upon that. If they are not false, you cannot simply claim that, since the proof is insufficient, it is false.
The point of my argument was that there was very little evidence either way. I implied that truth of the hypothesis would have no certain effect upon the world. Thus it is untestable, and completely unrelated to science. Therefore, any statement that it is false needs either a logical proof (all possible worlds), or to go on faith.
The physics analogy was on the other subject, of where we set the thresholds. In this case, even if we set them very low, we can say nothing. Your response makes more sense to the question of whether it is a belief you should personally adopt, not whether it is true or not.
Side Note: A few megabits? Really? You think you are that close to infallible? I know I’m not, even on logical certainties.
If they are not false, you cannot simply claim that, since the proof is insufficient, it is false.
Um, yes you can, or you end up being stuck believing any random “insufficient disproof” hypothesis. I’ll name a few, if you want: there’s an invisible intangible unicorn in your room, your boss is being mind-controlled by space aliens through subspace, and your door is wired to explode but only when you would be in a position to be killed by it.
If you give any of these meaningful credence—even enough to be “agnostic” about it—then you shouldn’t ever use that door, might consider quitting your job, and will work under the assumption that you never have privacy
And no, I don’t think I’m that close to infallible, I think I’d have to be that close to infallible to believe something that ridiculous.
… Um, the default assumption is that any given hypothesis is wrong, you can’t get your priors to converge otherwise. Omnipotent intelligent beings are sufficiently complex that I’d need a few megabits in their favor before I gave them parity with physics.
This given hypothesis is wrong. A megabit means a 2^1,000,000+ odds ratio.
If you can specify physics with n bits, and pick out a narrow class of humans or computers or books with ~k bits, and humans can produce an algorithmic description of a world with an omnigod, then you can specify the omnigod with ~n+k bits.
I don’t think it takes a megabit to specify a world like Permutation City for an intelligence to be essentially omnipotent in. Specifying intelligence along the lines of AIXI with unbounded computation requires few bits.
Point. I was thinking in terms of “this particular intelligence,” and whatever an AGI looks like I will happily offer 999:1 odds it will take more than a megabit of disc space. But if you just want “an intelligence,” yeah, not nearly as much.
… Still would need an awful lot to compete with physics, though.
To clarify: This is because any given hypothesis has a prior proportional to 2^-(complexity), which for any reasonable hypothesis means ludicrously low odds of being true.
Then you update on evidence, and just seeing things gives you megabits to work with, so that goes up significantly for particular hypotheses, but… the point is, even aside from evidence something like “God exists” is starting at a massive penalty compared to physics.
No, that is not the default way to handle a hypothesis. The default is to ask: why should I believe this? If they have reasons, you look into it, assuming you care. If the reasons are false, expound upon that. If they are not false, you cannot simply claim that, since the proof is insufficient, it is false.
The point of my argument was that there was very little evidence either way. I implied that truth of the hypothesis would have no certain effect upon the world. Thus it is untestable, and completely unrelated to science. Therefore, any statement that it is false needs either a logical proof (all possible worlds), or to go on faith.
The physics analogy was on the other subject, of where we set the thresholds. In this case, even if we set them very low, we can say nothing. Your response makes more sense to the question of whether it is a belief you should personally adopt, not whether it is true or not.
Side Note: A few megabits? Really? You think you are that close to infallible? I know I’m not, even on logical certainties.
Um, yes you can, or you end up being stuck believing any random “insufficient disproof” hypothesis. I’ll name a few, if you want: there’s an invisible intangible unicorn in your room, your boss is being mind-controlled by space aliens through subspace, and your door is wired to explode but only when you would be in a position to be killed by it.
If you give any of these meaningful credence—even enough to be “agnostic” about it—then you shouldn’t ever use that door, might consider quitting your job, and will work under the assumption that you never have privacy
And no, I don’t think I’m that close to infallible, I think I’d have to be that close to infallible to believe something that ridiculous.