3 ms·

> All our experience with computational sandboxes so far is that preventing information leaks or escapes is extremely hard. I believe this is due to the path-d

> All our experience with computational sandboxes so far is that preventing information leaks or escapes is extremely hard. I believe this is due to the path-dependent nature of our hardware architecture, not anything fundamental. x86 is not really designed with virtualization in mind. Theoretically, perfect airtight simulation is the primary ability of computers, and we're just too sloppy to exploit it properly. Another angle: imagine you build a computer, simulation-optimized or not, and then run a physics simulation program on it. Can entities inside the simulation get info about your CPU architecture? Not really, not unless you really messed up (let's assume you don't make any gross concessions to performance over correctness). If you have the ability to simulate anything close to our universe, I just don't think it's that hard to make it actually correct and therefore inescapable.

thwarted 3y ago
Can entities inside the simulation get info about your CPU architecture? (From HAKMEM): Item 154 (Bill Gosper): The myth that any given programming language is machine independent is easily exploded by computing the sum of powers of 2. If the result loops with period = 1 with sign +, you are on a sign-magnitude machine. If the result loops with period = 1 at -1, you are on a twos-complement machine. If the result loops with period greater than 1, including the beginning, you are on a ones-complement machine. If the result loops with period greater than 1, not including the beginning, your machine isn't binary -- the pattern should tell you the base. If you run out of memory, you are on a string or bignum system. If arithmetic overflow is a fatal error, some fascist pig with a read-only mind is trying to enforce machine independence. But the very ability to trap overflow is machine dependent. By this strategy, consider the universe, or, more precisely, algebra: Let X = the sum of many powers of 2 = ...111111 (base 2). Now add X to itself: X + X = ...111110. Thus, 2X = X - 1, so X = -1. Therefore algebra is run on a machine (the universe) that is two's-complement.
- andrewflnr 3y ago
  > path-dependent nature of our hardware architecture, not anything fundamental Anyway, there's no reason entities in a physics simulation would have such direct access to the primitives of the underlying computer, and trying to crash your host simulation does not seem like a reliable, healthy, or in any other way advisable approach to finding out if you're in one.
  - vidarh 3y ago
    There's no reason to assume we will know how to prevent such direct access in a way that isn't sufficiently more costly that simulations without sufficient protections are more likely than those who are. It would presume simulations are limited to being run more often by entities willing to be exceedingly conservative and careful than by entities making mistakes that creates openings. You don't need to try to crash your host simulation - trying to find a way to gain control to e.g. effectively become "gods" of the simulation, or find a way out, or just explore would all be other motivations. If I thought we were definitely in a simulation and could find a way to gain information about the outside, you can bet I'd want to be part of an effort to do just that - if we crash it, who cares? We'd be gone and wouldn't know we messed up. Severe corruption would be far worse than a crash.
    - andrewflnr 3y ago
      Or they'll figure out how to run good simulations once and copy paste it as desired? It wouldn't require them to be "exceedingly conservative", just a reasonable approach to their hardware and software. Again, that's assuming their reality looks anything like ours. This is the real problem with these discussions, we're speculating about the motives of entities we cannot even in principle have any information about. By the way, you might want to figure out how much you're actually worried about breaking the simulation, since you acknowledge that corruption would be a problem. Reasoning from human simulations, as seems to be your wont, do you think that risk of corruption is small enough to be so gung ho about messing with things?
vidarh 3y ago
> Can entities inside the simulation get info about your CPU architecture? Not really, not unless you really messed up (let's assume you don't make any gross concessions to performance over correctness). I would consider it magnitudes more likely to assume we would really mess it up than to assume there are no information leaks, and I think it's wildly unrealistic to assume we wouldn't make gross concessions to performance over correctness. > If you have the ability to simulate anything close to our universe, I just don't think it's that hard to make it actually correct and therefore inescapable. I think it's naive to assume that increased ability create complex systems correlate with increased ability to predict all possible ways of exploiting those systems. But even more so, I consider it naive to assume we would even if we could.
- andrewflnr 3y ago
  The answer to both points: the universe exhibits, as far as we can tell, a high degree of correctness. We've already tested it in pretty diverse and extreme circumstances and found it consistent. Clearly if our universe is a simulation, the simulators are both willing and able to give a shit about correctness, enough that there's no reason they would use the shoddy engineering that plagues modern attempts at sandboxing. The cut-rate simulations humanity is or will be capable of anytime soon are irrelevant.

thwarted 3y ago

Can entities inside the simulation get info about your CPU architecture? (From HAKMEM): Item 154 (Bill Gosper): The myth that any given programming language is machine independent is easily exploded by computing the sum of powers of 2. If the result loops with period = 1 with sign +, you are on a sign-magnitude machine. If the result loops with period = 1 at -1, you are on a twos-complement machine. If the result loops with period greater than 1, including the beginning, you are on a ones-complement machine. If the result loops with period greater than 1, not including the beginning, your machine isn't binary -- the pattern should tell you the base. If you run out of memory, you are on a string or bignum system. If arithmetic overflow is a fatal error, some fascist pig with a read-only mind is trying to enforce machine independence. But the very ability to trap overflow is machine dependent. By this strategy, consider the universe, or, more precisely, algebra: Let X = the sum of many powers of 2 = ...111111 (base 2). Now add X to itself: X + X = ...111110. Thus, 2X = X - 1, so X = -1. Therefore algebra is run on a machine (the universe) that is two's-complement.

andrewflnr 3y ago
> path-dependent nature of our hardware architecture, not anything fundamental Anyway, there's no reason entities in a physics simulation would have such direct access to the primitives of the underlying computer, and trying to crash your host simulation does not seem like a reliable, healthy, or in any other way advisable approach to finding out if you're in one.
- vidarh 3y ago
  There's no reason to assume we will know how to prevent such direct access in a way that isn't sufficiently more costly that simulations without sufficient protections are more likely than those who are. It would presume simulations are limited to being run more often by entities willing to be exceedingly conservative and careful than by entities making mistakes that creates openings. You don't need to try to crash your host simulation - trying to find a way to gain control to e.g. effectively become "gods" of the simulation, or find a way out, or just explore would all be other motivations. If I thought we were definitely in a simulation and could find a way to gain information about the outside, you can bet I'd want to be part of an effort to do just that - if we crash it, who cares? We'd be gone and wouldn't know we messed up. Severe corruption would be far worse than a crash.
  - andrewflnr 3y ago
    Or they'll figure out how to run good simulations once and copy paste it as desired? It wouldn't require them to be "exceedingly conservative", just a reasonable approach to their hardware and software. Again, that's assuming their reality looks anything like ours. This is the real problem with these discussions, we're speculating about the motives of entities we cannot even in principle have any information about. By the way, you might want to figure out how much you're actually worried about breaking the simulation, since you acknowledge that corruption would be a problem. Reasoning from human simulations, as seems to be your wont, do you think that risk of corruption is small enough to be so gung ho about messing with things?

vidarh 3y ago

> Can entities inside the simulation get info about your CPU architecture? Not really, not unless you really messed up (let's assume you don't make any gross concessions to performance over correctness). I would consider it magnitudes more likely to assume we would really mess it up than to assume there are no information leaks, and I think it's wildly unrealistic to assume we wouldn't make gross concessions to performance over correctness. > If you have the ability to simulate anything close to our universe, I just don't think it's that hard to make it actually correct and therefore inescapable. I think it's naive to assume that increased ability create complex systems correlate with increased ability to predict all possible ways of exploiting those systems. But even more so, I consider it naive to assume we would even if we could.

andrewflnr 3y ago
The answer to both points: the universe exhibits, as far as we can tell, a high degree of correctness. We've already tested it in pretty diverse and extreme circumstances and found it consistent. Clearly if our universe is a simulation, the simulators are both willing and able to give a shit about correctness, enough that there's no reason they would use the shoddy engineering that plagues modern attempts at sandboxing. The cut-rate simulations humanity is or will be capable of anytime soon are irrelevant.