4 ms·

Okay, but you’re not running the simulation time 1:1, are you? What would be the point? So very rapidly your simulation develops people, who develop algorithms

Okay, but you’re not running the simulation time 1:1, are you? What would be the point? So very rapidly your simulation develops people, who develop algorithms and computers for which there are no physically possible (in the real universe) optimizations, so you’re computing for them 1:1. And they develop simulations of their own. And then their simulations develop simulations. And you’ve got to pay the energy cost for all that because some things can’t be faked; or you kill the simulation when it learns to develop its own simulation. And then... what would be the point? There are some problems with “Okay, but then you just make things simpler,” where this would allow simulations to determine they were simulations. I don’t recall the details off the top of my head. (Its late and I’m typing this right before bed.) So maybe there is infinite energy in the real world. Then why wouldn’t our universe look the same? There are certainly kinds of simulations that are possible, but the most commonly discussed one is the ancestor simulation. And it just doesn’t physically work while also sticking to the point of running an ancestor simulation.

kiwicopple 6y ago
> What would be the point? We do pared-down simulations of things all the time. I think the rest of the comment presumes that the only type of simulation we should conceive of is one that is an exact mapping of some external world, and that external world adheres the same laws that our "simulation" has. Don't get me wrong - i'm not for or against the simulation argument, I just don't understand the argument that a simulation of "a" universe would consume all the energy of some "other unknown" universe. Nor do I think that the simulation would necessarily have to map 1:1 for us to be in a simulation.
- cgriswald 6y ago
  You should read the papers associated with the simulation argument. There are, of course, loads of simulations you could run that don’t fit the confines of the argument. But if you don’t have a well-defined definition of simulation you can’t reasonably talk about the probably that you’re in one. Yes, you can simplify a great deal. You don’t need to simulate the moons of Jupiter before Galileo comes along. You can’t simplify an algorithm that is at its theoretically most efficient. That’s when you start running into the 1:1 problem. My argument is that the only way to avoid the 1:1 problem is to craft the simulation in such a way that it’s not an ancestor simulation as defined in the papers.

kiwicopple 6y ago

> What would be the point? We do pared-down simulations of things all the time. I think the rest of the comment presumes that the only type of simulation we should conceive of is one that is an exact mapping of some external world, and that external world adheres the same laws that our "simulation" has. Don't get me wrong - i'm not for or against the simulation argument, I just don't understand the argument that a simulation of "a" universe would consume all the energy of some "other unknown" universe. Nor do I think that the simulation would necessarily have to map 1:1 for us to be in a simulation.

cgriswald 6y ago
You should read the papers associated with the simulation argument. There are, of course, loads of simulations you could run that don’t fit the confines of the argument. But if you don’t have a well-defined definition of simulation you can’t reasonably talk about the probably that you’re in one. Yes, you can simplify a great deal. You don’t need to simulate the moons of Jupiter before Galileo comes along. You can’t simplify an algorithm that is at its theoretically most efficient. That’s when you start running into the 1:1 problem. My argument is that the only way to avoid the 1:1 problem is to craft the simulation in such a way that it’s not an ancestor simulation as defined in the papers.