4 ms·
> One way to make it crash is making the simulation crawling to a stop by requiring always more computation to be done by preventing the simulator to take short
by jusssi 5y ago
> One way to make it crash is making the simulation crawling to a stop by requiring always more computation to be done by preventing the simulator to take shortcuts in the computing process.
Isn't this the reason for mass-induced time-dilation? When you add more mass (computational complexity) into a region it runs slower, to the point of being removed completely from the sim, hidden behind an event horizon and described by just a couple of parameters.
- GistNoesis 5y agoCould be, but it only works in a non-adversarial context. Biological numerical instabilities are a tougher beast to tame. When you have to simulate a big star you can do it quickly by approximating it by a sphere of a certain temperature. The bigger the star, the more the law of large numbers apply and the better your approximation get. When you have to simulate a computer that would be the size of a star, you can't take any shortcut. And the amount of computation that a computer the size of a star would allow you to do according to the limits of physics is huge, and would allow to simulate the full universe many times (see https://en.wikipedia.org/wiki/Limits_of_computation https://en.wikipedia.org/wiki/Limits_of_computation). One way to avoid a simulation crash would be to introduce time-dilation effect to throttle the quantity of computation locally, but fundamentally it's a very hard problem equivalent to the halting problem.