|
|
|
|
|
|
by vidarh
4694 days ago
|
|
|
I got into an argument with Kurzweil on his blog once years ago regarding the details. The simulation does not necessarily need to be simpler, subject to a few constraints: It needs to be possible to represent the simulation in less space than its containing universe. It also needs to be possible to execute the simulation in the available space/energy, but if acceptable it may be sufficient to be able to execute it at reduced speed which can substantially reduce the amount of matter/energy required at any one point as you'd be able to e.g. more aggressively apply compression etc. to the data. The space requirement can be fairly small depending on the constraints on your simulation. E.g. are you simulating a whole universe? A planet? One person? Everything not directly observed by a simulation subject can be aggressively pruned as long as it can be reconstructed in a believable state (unless you are trying to accurately simulate the physics of everything), and even things that are observed can be simplified dramatically. E.g. in almost all cases nobody would be able to tell a rough approximation of the physics and appearance of cloud formations from "the real thing", so you don't need to simulate them particle by particle. On the other hand, one such optimization/approximation then could be to fake the result of any attempt to run your own simulation. One thing I find interesting is that Bostrom talks about "ancestor simulations" a lot, but there's no reason to assume that if we're in a simulation it has to be a simulation of the past of whomever runs the simulation - it could be a game, or someone simulating aliens (to them), or it could even be a civilization that decided to create themselves a nice simulation to live in, or a "theme park"/"zoo". |
|
|
Or it could have been created in their offices, to make the job of maintinain a guide to galactic hitch-hiking easier to maintain.