A simulation running on a simulator will be affected by anything that affects the simulator. Electrical noise warping the speed at which the simulation happens, but flips from cosmic rays, even the hard latency due to the speed of light, all things that affect the simulation itself. It is unavoidable, some information about the real universe will leak into the simulation. So if we are a simulation, it is detectable, unless the bare metal universe allows for 0 entropy processes, in which case the simulation would likely be as unobservable from the universe as the universe is from within the simulation, which makes it not really a simulation at all.
The speed at which the simulation happens is completely irrelevant to the inside of the simulation and does not have to be detectable from the inside. Latency doesn't matter, and even in our world we already have protections against things like bit flips. You're jumping to conclusions way too fast.
The simulator may misplace the rock[0], but you have no guarantee that it ever will during this particular simulation. And even if it does, but a state that fails integrity checks does not continue to be simulated, you'll never have the opportunity to actually use it for anything inside the simulation.
It's not enough to affect the simulator in some way to affect the simulation. The fact that your breath provides some thermal energy to your computer does not make it detectable from a ZX Spectrum emulator running on it, even if you could detect it from your computer itself. If you were to detect whether a ZX Spectrum program runs on a simulator or on the real thing, you would have to look for differences in how the real ZX Spectrum behaves versus the emulator. Without that knowledge, you're completely in the dark, and even bit flips won't help you (the real thing could have been bit-flipping in the exact same way after all).
You're focusing on the details of my examples too much. In our universe, you cannot create a system of any kind unaffected by the world around it, this is a fundamental law. No system you build can be entirely isolated from the world around it. You can isolate it in this way or that way, but you cannot entirely prevent interaction with the outside world. A computer is not some magical foam that insulates software running on it from the universe if the programmer so desires. If you run a program where some entity inside is even capable of investigating that it is just inside a program, there will be indicators, period. That is in our universe. It doesn't matter if it is a computer or some other method of running the simulation, from inside, there will be detectable effects to any entity that tries to investigate enough.
In a universe where this is not true, that is, systems can emerge that are entirely isolated from the rest of it, even so far as not having to obey the physical laws of said universe, that would be the stable state of emergent systems within that universe. Systems generally would not interact with one another. So even if an entity in that universe could construct a simulation of some kind, and the universe around it had no impact on it whatsoever, it would not be able to observe the simulation and so it would be either pointless or not a simulation at all which both amount to the same thing.
I should note, I'm not saying that we aren't in one, I don't think we are but if we are, it is fundamentally falsifiable and if the more we understand our universe we see no indication of it, that means we aren't in one.
> You can isolate it in this way or that way, but you cannot entirely prevent interaction with the outside world.
The mere fact of some interaction happening does not automatically let you answer the question whether you're in a simulation or not. The whole point of simulations where you can inspect them from the outside is some kind of interaction, but it doesn't mean that an entity inside the simulation is able to determine what's going on outside. Just that the simulation is observable makes it not isolated in the sense that you're using, so yes, you're right, but at the same time it's not very relevant to what I was talking about. A simulated Game of Life field isn't isolated either.
