4 more ideas how to test if you are in simulation:
Overloading attack. We create so complex computations or a process which will require much more detailed simulation than usual simulation. Presumably, the owners will be unhappy and will turn off such simulation. Bitcoin could be such computation. The computation input should be based on some random numbers so it can’t be reused for different simulations. (The idea from Lem’s Solaris)
Superintelligence attack. We create a super-duper-intelligence, with solo task to search for glitches and inconsistencies, and try to break out of simulation. It is also an overloading attack. Presumably, the simulation owners will be not happy. EY wrote about such attack.
Indexical uncertainty. attack. Create our own simulation which will simulate the possible simulators of a typical simulation. They will be not happy and will terminate us.
Cruelty attack. Create our own simulation in which despicable things will be performed. Simulators will be unhappy. (I am not advocating actually doing it.)
If we continue to exist after all these and no glitches are found, we are less likely to be in simulation.
This assumes that we’re in a type of simulation that’s variable-load.
(As an example of a type of simulation that isn’t variable-load, consider a ‘naive[1]’ Game of Life simulation on a bounded grid. Computational load is (quasi)-constant regardless of how complex the pattern is.)
Bitcoin could be such computation. The computation input should be based on some random numbers so it can’t be reused for different simulations
Not “can’t”. Just more difficult. (And that’s assuming that they can’t easily break SHA, which seems doubtful seeing as they can simulate quantum processes.)
For an obvious way to do so:
Any time you (as in the simulator) see a Bitcoin miner, replace it with a black box that doesn’t simulate the computation at all (with proper “am I being observed” checks of course). At the appropriate randomized time interval directly output a result memoized and reused across simulations.
Bitcoin results are not easy to reuse in different simulations, as people are making different transactions and it affect computations of the blockchain.
If the simulation is a simulation of every atom in the universe, it is computationally expensive, and there is no many such computations, so I am unlikely to be in it.
Bitcoin/blockchain force cohesion of a set of sims around the merkle tree, but you can still have many different sims that share variations of the same transaction history (they just differ then only in details).
But looking at it another way blockchains are also strong useful constraining evidence, so historical sims would always just recreate the same merkle trees. This leads to the interesting idea that you could attempt to preserve blockchain wealth post-sim … but of course you would just be sharing it with other versions of yourself, and potentially your simulators.
TL;DR? How can one tell experimentally, even in principle, if the world is a simulation? And if there is no way, then why bother talking about it?
4 more ideas how to test if you are in simulation:
Overloading attack. We create so complex computations or a process which will require much more detailed simulation than usual simulation. Presumably, the owners will be unhappy and will turn off such simulation. Bitcoin could be such computation. The computation input should be based on some random numbers so it can’t be reused for different simulations. (The idea from Lem’s Solaris)
Superintelligence attack. We create a super-duper-intelligence, with solo task to search for glitches and inconsistencies, and try to break out of simulation. It is also an overloading attack. Presumably, the simulation owners will be not happy. EY wrote about such attack.
Indexical uncertainty. attack. Create our own simulation which will simulate the possible simulators of a typical simulation. They will be not happy and will terminate us.
Cruelty attack. Create our own simulation in which despicable things will be performed. Simulators will be unhappy. (I am not advocating actually doing it.)
If we continue to exist after all these and no glitches are found, we are less likely to be in simulation.
Interesting ideas.
> Overloading attack.
This assumes that we’re in a type of simulation that’s variable-load.
(As an example of a type of simulation that isn’t variable-load, consider a ‘naive[1]’ Game of Life simulation on a bounded grid. Computational load is (quasi)-constant regardless of how complex the pattern is.)
Not “can’t”. Just more difficult. (And that’s assuming that they can’t easily break SHA, which seems doubtful seeing as they can simulate quantum processes.)
For an obvious way to do so:
Any time you (as in the simulator) see a Bitcoin miner, replace it with a black box that doesn’t simulate the computation at all (with proper “am I being observed” checks of course). At the appropriate randomized time interval directly output a result memoized and reused across simulations.
Not something like e.g. HashLife.
Bitcoin results are not easy to reuse in different simulations, as people are making different transactions and it affect computations of the blockchain.
If the simulation is a simulation of every atom in the universe, it is computationally expensive, and there is no many such computations, so I am unlikely to be in it.
Bitcoin/blockchain force cohesion of a set of sims around the merkle tree, but you can still have many different sims that share variations of the same transaction history (they just differ then only in details).
But looking at it another way blockchains are also strong useful constraining evidence, so historical sims would always just recreate the same merkle trees. This leads to the interesting idea that you could attempt to preserve blockchain wealth post-sim … but of course you would just be sharing it with other versions of yourself, and potentially your simulators.
There are 4 types of evidence of simulation:
The chances of miracles are higher in simulation.
Afterlife—it is more likely in simulation.
Your special place in time or role – you more likely to live interesting life in sim.
Simulation termination—the end of the world is more likely in it.
But all of these can’t be results of experiments which you initiated, as simulation owners can prevent all these.