Hacker News Reader

Comment by bronzeage | Hacker News Reader

These kind of arguments forget that it would probably take more space and time to simulate a given space and time, simulations are less effective than the thing they simulate. So in order to simulate Earth we'd need much more space than the earth. Then when looking at the stars, those would either have to be copied from a real universe or badly simulated. It's going to be a hard task stimulating the signals seen from all the observable universe with time and space constrained within a planet. Something would be off.

View on HN

I don't think this is that decisive an argument. One, you don't actually know how fast the simulation might be running (or whether it regularly rewinds to deal with inconsistencies). Two, there appear to be many performance mechanisms built into the universe that one would at least consider in building a simulation of this kind: the speed of light allows for a large degree of isolation, and quantum uncertainty allows some degree of lazy processing.

I remain pretty strongly agnostic about all this but I don't think you can escape it easily. One interesting thing to me would be what it would mean for some _rich_ far off signal to be picked up. It seems relatively easy for the universe to maintain a low fidelity simulation about rocks and radiation in space at a distance (plus or minus things like dark matter!) But if we suddenly started receiving broadcasts from an alien civilisation, or even met them face to face, that would be two very rich and detailed timelines that would suddenly have to snap into sync. That's certainly not something I'd allow if I had to build and maintain this stuff (and one can only assume unreasonable deadlines and budgets are universal).

Fermi paradox explained

That argument doesn't hold much water for me. All 3D games simulate a space larger than a computer, while running on a computer.

And there's no reason why simulation requires everything to be simulated at the same time. Stars could be modeled as point masses, and only calculated in more detail when someone looks at them.

Take a look at the mandelbrot set https://math.hws.edu/eck/js/mandelbrot/MB.html infinite fractal complexity, shown using limited space + time.

How would we know that the stars are imperfectly simulated when we look at them? It's not like we'd have a real universe to compare it with. Base reality might be unimaginably more complex than our world and there's really no way we could tell.

An interesting counterpoint would be the Hashlife algorithm [1] for the game of life which is able to simulate the game of life at a scale and size well beyond any reasonable physical limits.

[1]: https://www.drdobbs.com/jvm/an-algorithm-for-compressing-spa...

There are no counterpoints because every turing complete simulation can't be made significantly more efficient. Any improvement would contradict the fact complexity classes classes strictly contain but aren't equal to each other. The most efficient way to run a universal turing machine on general input is to just run it.

True, but complexity doesn't necessarily have anything do to with time and space, that's all I really wanted to point out.

I don't see how complexity classes are relevant here. If we accept that simulated time can run more slowly than real time – if it can take 10 seconds of computation time to simulate 1 second of simulation time – then complexity classes have no real relevance. Time complexity classes are just about how long a computation takes, and given an arbitrary amount of time to run the simulation in, and an arbitrary simulation rate (number of real seconds per a simulated second), it can take however long it takes.

Even space complexity classes are not necessarily an issue. If the simulation runs out of storage, you can always suspend it, go provision more, and then resume it. The people in the simulation would never notice their suspension. It would only be a hard limit if some computation required more storage than could ever be provisioned, such as more bits than could fit in the observable universe. However, I doubt accurately simulating the contents of all contemporary human minds would need anywhere near that many bits.

What do you mean "well beyond any reasonable physical limits?"

Well there's no direct translation from the game of life to the real world, but for any reasonable value for the size of a cell or the timer per simulation step there are simulations that are impractically large or take an impractically long time were you to try to build a full physical simulation of them.

This is assume there isn't some sort of "lossy compression" going on with the simulation. I don't know how that would work, but none of us know how a simulation would happen anyway.

The compression algorithm is that reality only has to be simulated for one person: you, the reader of this comment.

This. In sense we are always thinking some cpu simulation. Where in fact can be more of edge computing case. As in you only need to simulate the whole personal experience.

In computer graphics terms. Culling. Objects are not simulated if u have no direct observer.

> Objects are not simulated if u have no direct observer

As Berkeley said, esse est percipi (to be is to be perceived)

That's one of the reasons why the simulation argument interests me. I think it may give one a reason to choose idealism over materialism. Idealism has a greater "simulation resistance" than materialism has. To a materialist, the question of whether we are living in a simulation may seem profound. An idealist can just shrug their shoulders.

A materialist has identified material reality as the most fundamental reality, and so the possibility that apparent material reality is not really the most fundamental reality does in some ways threaten the materialist worldview. For an idealist, for whom conscious minds and their experiences (qualia) are the most fundamental reality, the question of whether we are in a simulation is simply a question about whether their exist conscious minds who have the experience simulating us – and that may or may not be true, but either way it makes no difference to who or what we fundamentally are.

Or you could define most of the universe as a probability unless it’s being observed. You know...like quantum theory.

As we are part of the simulation, the code could include routines that make us perceive the simulation as accurate.

Reply on news.ycombinator.com