3 ms·

It's premature to take for granted that a paper brain emulator can exist in principle. Take something else less...emotionally laden. A black hole. Could we mo

It's premature to take for granted that a paper brain emulator can exist in principle. Take something else less...emotionally laden. A black hole. Could we model a black hole on paper? If so, we ought to just be able to drop a computational particle in and see exactly what happens at the singularity. Except we can't -- the whole reason we call it a singularity is because we get divide by reality errors if we try to compute it. But I would argue there's a different problem. A black hole is the most computationally efficient method of calculating itself. Any method of accurately emulating the black hole on the Planck level is going to take more energy (and mass) than the actual black hole. As a result, those pieces of paper would literally collapse into an actual black hole long before they could properly emulate the black hole they were modeling. Which isn't to say that the brain is anything like that but the point is, it's not a foregone conclusion that we can construct a completely accurate emulator of a physical object. Finally there's something of a category error in making the claim that pieces of paper can be sad. A paper emulation of a bar magnet would not itself be magnetic. In the world of the emulation it would produce an output that would model magnetism, yes, but that's the extent of it. Our paper brain emulator, if it were constructible, would produce output that would be a model of sadness in the emulation, but it would not actually be sad.

SAI_Peregrinus 6y ago
> As a result, those pieces of paper would literally collapse into an actual black hole long before they could properly emulate the black hole they were modeling. That is wrong. Any method of emulating the black hole within a volume less-than or equal to the Schwarzchild radius of the black hole would collapse, but you could do the calculation over a larger volume and avoid that. The most massive stars are far more massive than the least massive black holes. The smallest black hole known that I could find from a quick search is XTE J1650-500, at about 3.8 solar masses with a "15 mile"[1] diameter. There are a lot of stars more massive than that.[2] [1] https://www.nasa.gov/centers/goddard/news/topstory/2008/smallest_blackhole.html https://www.nasa.gov/centers/goddard/news/topstory/2008/smal... [2] https://en.wikipedia.org/wiki/List_of_most_massive_stars https://en.wikipedia.org/wiki/List_of_most_massive_stars
- Enginerrrd 6y ago
  >Any method of emulating the black hole within a volume less-than or equal to the Schwarzchild radius of the black hole would collapse, but you could do the calculation over a larger volume and avoid that. ...This was my intuition as well, but I'm far from certain about it. It's not uncommon for the universe to find sneaky ways to prevent us from "cheating" so to speak. These often result from deep symmetries/conservation laws and fundamental limits on information. I'm actually a big opponent of the simulation argument for that reason. I don't think you can accurately simulate the universe without having a universe to simulate it in. Otherwise, you could just 'turtles-all-the-way-down' the simulation. The information density would have to be unbounded, and this seems to be in disagreement with fundamental laws of the universe. We don't have a theory of quantum gravity... it's definitely possible that to simulate the blackhole in a more spread out manner would be impossible. I could envision a fundamental tradeoff where to replicate the information exchanges between the microscopic constituents requires that either you satisfy the hoop conjecture, or you keep increasing the size of your model without bound, whereby for any finite size you still satisfy the hoop conjecture. (Particularly since the required mass/energy density goes down significantly as the radius increases.) I know that's wild speculation, but I feel like it's not quite crazy enough to take for granted.
Enginerrrd 6y ago
I think I'm almost entirely with you except for these two statements: >It's premature to take for granted that a paper brain emulator can exist in principle. I think it is premature to say we could simulate all of quantum mechanics with sheets of paper, so on that technicality, I totally agree. However, I think it's quite unlikely to be the case that we couldn't in principle simulate the requisite components required for a faithful replication of consciousness. But you're correct that it's probably not a given. >Finally there's something of a category error in making the claim that pieces of paper can be sad. A paper emulation of a bar magnet would not itself be magnetic. In the world of the emulation it would produce an output that would model magnetism, yes, but that's the extent of it. Our paper brain emulator, if it were constructible, would produce output that would be a model of sadness in the emulation, but it would not actually be sad. I think I disagree with almost all of this. The bar magnet is a bit of a bait and switch because the system, the input, and the output weren't well defined. In consciousness, you need to define the system, system input and system output properly. If the system has persistence of thought/computation, and the inputs and outputs can be defined in a way identically to those of your own consciousness, then it would actually be sad, in every way that you are. In particular, if you replace bits of paper with "neurons", your argument should absolutely still hold. So the logical extension of this argument is that either YOU also aren't capable of being sad, or your consciousness does not originate in your brain. Both of those I think are pretty close to sufficiently absurd that we can accept them as givens.

SAI_Peregrinus 6y ago

> As a result, those pieces of paper would literally collapse into an actual black hole long before they could properly emulate the black hole they were modeling. That is wrong. Any method of emulating the black hole within a volume less-than or equal to the Schwarzchild radius of the black hole would collapse, but you could do the calculation over a larger volume and avoid that. The most massive stars are far more massive than the least massive black holes. The smallest black hole known that I could find from a quick search is XTE J1650-500, at about 3.8 solar masses with a "15 mile"[1] diameter. There are a lot of stars more massive than that.[2] [1] https://www.nasa.gov/centers/goddard/news/topstory/2008/smallest_blackhole.html https://www.nasa.gov/centers/goddard/news/topstory/2008/smal... [2] https://en.wikipedia.org/wiki/List_of_most_massive_stars https://en.wikipedia.org/wiki/List_of_most_massive_stars

Enginerrrd 6y ago
>Any method of emulating the black hole within a volume less-than or equal to the Schwarzchild radius of the black hole would collapse, but you could do the calculation over a larger volume and avoid that. ...This was my intuition as well, but I'm far from certain about it. It's not uncommon for the universe to find sneaky ways to prevent us from "cheating" so to speak. These often result from deep symmetries/conservation laws and fundamental limits on information. I'm actually a big opponent of the simulation argument for that reason. I don't think you can accurately simulate the universe without having a universe to simulate it in. Otherwise, you could just 'turtles-all-the-way-down' the simulation. The information density would have to be unbounded, and this seems to be in disagreement with fundamental laws of the universe. We don't have a theory of quantum gravity... it's definitely possible that to simulate the blackhole in a more spread out manner would be impossible. I could envision a fundamental tradeoff where to replicate the information exchanges between the microscopic constituents requires that either you satisfy the hoop conjecture, or you keep increasing the size of your model without bound, whereby for any finite size you still satisfy the hoop conjecture. (Particularly since the required mass/energy density goes down significantly as the radius increases.) I know that's wild speculation, but I feel like it's not quite crazy enough to take for granted.

Enginerrrd 6y ago

I think I'm almost entirely with you except for these two statements: >It's premature to take for granted that a paper brain emulator can exist in principle. I think it is premature to say we could simulate all of quantum mechanics with sheets of paper, so on that technicality, I totally agree. However, I think it's quite unlikely to be the case that we couldn't in principle simulate the requisite components required for a faithful replication of consciousness. But you're correct that it's probably not a given. >Finally there's something of a category error in making the claim that pieces of paper can be sad. A paper emulation of a bar magnet would not itself be magnetic. In the world of the emulation it would produce an output that would model magnetism, yes, but that's the extent of it. Our paper brain emulator, if it were constructible, would produce output that would be a model of sadness in the emulation, but it would not actually be sad. I think I disagree with almost all of this. The bar magnet is a bit of a bait and switch because the system, the input, and the output weren't well defined. In consciousness, you need to define the system, system input and system output properly. If the system has persistence of thought/computation, and the inputs and outputs can be defined in a way identically to those of your own consciousness, then it would actually be sad, in every way that you are. In particular, if you replace bits of paper with "neurons", your argument should absolutely still hold. So the logical extension of this argument is that either YOU also aren't capable of being sad, or your consciousness does not originate in your brain. Both of those I think are pretty close to sufficiently absurd that we can accept them as givens.