If we lived in a computer program...

[Timeroot]

There have been many questions - and even several movies - about just living in a computer program. I was wondering what would happen if you made a computer to simulate life, and then hooked it up to an extremely motor and mass, and then it set it vibrating in small circles with a radial speed of 99% the speed of light while in orbit around the earth... due to relativity, it time would slow down for the computer, theoretically allowing it fully simulate life faster than life actually unfolds. We'd let it orbit for a year or two (it would be in space to reduce the air drag; maybe a vacuum chamber with magnetic levitation could work as well), and then check whether the virtual people - in whatever form they evolved, assuming we can interpret them - had discovered that they lived in a computer. Then we apply they're arguments, assuming they have, to whether the same applies for us.

But then the real question would be having a computer simulate another computer simulating cellular automata which simulate life.... and see how many "levels" of simulation the people uncover. :-P

Just a silly thought I thought I would share with you. Not very fractal related, but it's computer related and it's about cellular automata...

Thoughts?

[Tglad]

That's a funny thought experiment
I would say the little virtual people would never discover they exist in a computer... because in a sense they don't, they would be the same virtual people in every conceivable way if they existed inside a purple flying jelly, if that jelly was performing the same information processing. Any discovery they make about what they live in, they'll make the same discovery when inside the flying jelly. Since there is no measurable difference between virtual Bob that lives in the black-hole computer and virtual Bob that lives in the jelly, you could say that virtual Bob lives inside both, and in fact the most Bob could ever say about what he lives inside is that it is some sort of processor that supports his existence.

I suppose a better option if Bob was very smart would be to consider all possible situations that create the virtual world he is in then to assume he is inside the most likely. He can't know for sure though. Kind of reminds me of quantum theory and uncertainty.   

[Timeroot]

I suppose. What if the computer used some algorithm which computed more quickly, but for which normal physics were superstable (or maybe just stable) attractor... some kind of optimization which assumed that equations it modeled approached the actual physics. The maybe Bob, if he devised an idea of how he was potentially being simulated, perhaps he could manipulate particles in just such a way that he could trigger instability, potentially crashing the computer. I'm thinking along the lines of Planck length/Planck time = Round off errors, Quantum weirdness = instabilities, Black hole = numbers too large, etc....

Of course, he would want a way to control the instabilities, otherwise the computer might crash. This isn't really comparable to any other "errors" like the ones I just listed, because instead of manifesting itself some way, the entire universe would throw and error and Bob - or we - would crease to exist.

Would Bob be willing to risk his life to test this theory of "crashing the universe"? What if we played God and made him part of a sacrificial, yet scientific race? Am I actually making sense when I say this, or am I just tired?

[Tglad]

perhaps he could manipulate particles in just such a way that he could trigger instability, potentially crashing the computer

hmm... LHC ?   

[hobold]

If the simulation is complete, i.e. if it can cover the parameter space of the simulated physics completely, then there will not be any "magical" events where the simulation behaves strangely, contradicting its own laws.

If the simulation is self contained, i.e. if it doesn't use information from external sources to influence the simulated world, then failures due to incompleteness could not possibly reveal "outside" information to the simulated world. Simulated beings could learn of certain implementation details (or bugs!) that way, but those implementation details would effectively become a lower layer of natural laws.


If the simulation can crash, then the simulated world simply stops. But simulated time would stop as well, so the simulated beings would never know. The simulation might have been restarted time and time again from some checkpoint state, but the simulated beings would have no means of knowing.


So if the simulation was well done, the simulated world would be the only reality that the simulated beings could ever know. And it would be exactly as real to them as would the simulator's world to the simulator.

[makc]

The simulation might have been restarted time and time again from some checkpoint state, but the simulated beings would have no means of knowing.

Not necessary; that's kind of the same q-n as if debugged code can know that it is being debugged. There were many tricks developed, from simple timer checks to exploiting CPU bugs. Of course, simulation authors could restart over and over closing the holes, but in likely case that some bugs go unnoticed, creatures could find a way to notice such a restart. Of course, again, they would need at least to start looking for signs before they could see them.

[Timeroot]

The simulation might have been restarted time and time again from some checkpoint state, but the simulated beings would have no means of knowing.

Not necessary; that's kind of the same q-n as if debugged code can know that it is being debugged. There were many tricks developed, from simple timer checks to exploiting CPU bugs. Of course, simulation authors could restart over and over closing the holes, but in likely case that some bugs go unnoticed, creatures could find a way to notice such a restart. Of course, again, they would need at least to start looking for signs before they could see them.

If the simulation is complete, i.e. if it can cover the parameter space of the simulated physics completely, then there will not be any "magical" events where the simulation behaves strangely, contradicting its own laws.

Yes, that why I'm curious about the computer using some sped-up algorithm with the true physics just being the limit. Simulate the big bang, and by the time you've reached stars, you're physics are spot on. But if the simulation intentionally tried to avoid the stable state... could it? This could also be phrased as, "Is the Lorenz system (or any chaotic system) really chaotic... or does it refuse to be periodic?" It's an unusual question, but it does make sense. Could a system be locally attracted towards the one state that causes it to differ from the "true" calculation, without the optimization?

[hobold]

I suspect one of the formal definitions for "chaotic" is equivalent to your phrase "refuses to be periodic". It is a bit difficult to reason about these things, when the terminology itself is not clear.