Comment by da_chicken | Hacker News Reader

Comment by da_chicken | Hacker News Reader

Parent Full thread

Well, it's a little more complicated than that. It's only not falsifiable because nobody has devised a test. It's a hypothesis without an experiment. That doesn't mean that people aren't trying [0]. If it is a simulation, you're not going to clip through the wall by running against a corner and jumping for an hour.

The thing is, we don't know where this line of reasoning would even lead. I mean, people used to think that light had to travel in little packets through luminiferous aether [1] and look where that investigation led: Einstein, special relativity, atomic weapons and power, nuclear medicine, etc.

Let's say the universe is fundamentally indistinguishable from a computer simulation. What consequences does that have? What does that say about the types of math that can be used to describe the universe? Or, perhaps better, the types of math that shouldn't be used to describe the universe? What does that say about our ability to model a portion of the universe and therefore predict outcomes? Does it have implications for stochastic processes? Indeed, where is the stochasticism coming from? Is everything predictable or is there some fundamental randomness?

On the other hand, what if we can show that the universe is distinguishable from a computer simulation. What does that mean for our understanding of the universe as it exists? Does it reveal a fundamental aspect of reality that we have missed, or a limitation of our mathematics, or a fault in our computer models or design, etc.? We're dogs learning calculus here, and the universe isn't obligated to make sense.

Even if we can never find a useful test for this hypothesis, the line of thinking itself potentially leads us into new areas of discovery.

[0]: https://gizmodo.com/5950832/how-to-tell-if-the-universe-is-a...

[1]: https://en.wikipedia.org/wiki/Luminiferous_aether

The problem is...you can even define what it means for the universe to be a simulation. You might as well argue like George Berkely that all sensations are signs and symbols directly from God - it is not even wrong. It is fundamentally unknowable - unless God or The Simulator choose to reveal themselves directly to us.

The luminiferous ether was a falsifiable theory - indeed it was a breakthrough for physics because it was falsified (experiments failed to detect it) so new theories had to be developed. So this is completely different from unfalsifiable metaphysics.

It is falsifiable if you consider that a simulation would have limited resources allocated to it. The null hypothesis would be the universe runs in a way where a simulation couldn't cheat by performing a rough calculation at the hard parts, thereby spending a tiny fraction of what would be needed and produce the same observable result.

The idea that quantum fields exist in multiple states simultaneously until observed become evidence for a simulation, since a simulator could be deferring calculations on things not observed.

Also, the very program this article is referring to could be considered as evidence for it, since if a rough simulation can be made where the observable result is indistinguishable from a result where every atomic element in the entire universe was simulated, then, again, a simulator with limited resources would make do.

The law of conservation of energy is built the same way. There is no way we can test ways that energy is transferred that we don't yet know about, but this law is still considered true, because we have so many examples of it working.

Well I guess you can disprove particular forms of simulation - e.g. since a computer with 16GB RAM can be adequately simulated, we can reject the notion that the current universe is simulated on a computer with 16GB RAM or less.

> The idea that quantum fields exist in multiple states simultaneously until observed become evidence for a simulation,

How is that evidence? Current physics can model that without relying on a simulation hypothesis. Adding a simulation hypothesis does not simplify the model, and it is unfalsifiable.

Why did you quote half my sentence, leaving out the part where I explain why, and then ask why?

> Adding a simulation hypothesis does not simplify the model, and it is unfalsifiable.

I explain this. It's falsifiable if it can be shown there is some extremely large and complicated process that be observed to be different than any possible rough calculation that could be accomplished by a simulator with very constrained resources (constrained in comparison to all the supposed atomic particles in the universe)

The notion that all forms of energy must be conserved, even undiscovered ones, has been used as the basis for the theory of Hawking radiation, isn't it? Yet claiming that undiscovered forms of energy are conserved, just as known ones, is unfalsifiable, too.

The law of conservation of energy, including undiscovered ones, is in the same boat as the theory that the universe is a simulation.

Until we find a form of energy that isn't conserved, it is simpler to assume all types of energy are conserved.

Similarly, until we find a process that is of large enough scale that it couldn't be simulated on a computer with resources much more limited than the size of the universe, and be observable identical to a process taking much more resources, it is simpler to assume it all is being simulated in such a way.

> Why did you quote half my sentence, leaving out the part where I explain why, and then ask why?

Sorry I just didn't want to quote the whole comment, didn't mean to imply I was only responding to the first part of the sentence.

I understand you can hypothetically prove that a simulation of the universe would require a simulation engine of at least a certain complexity. But since the simulation engine does not exist in this universe, we have no idea about what limitations or laws of nature (if any) it is subject to. You will never be able to prove or disprove the existence if this hypothetical simulator.

Well, a Turing complete computer is about as abstract as you can get, and so is the concept of memory, regardless of the natural laws of some outer universe, it'd probably be a good starting point to consider a turing computer except with limited memory.

The point is not to prove one way or another whether the universe is a simulation, just like there is no point trying to prove the law of conservation of energy as universal for all forms of phenomena, discovered and undiscovered.

It's just if we can show that all known phenomena can be the result of a very limited simulation, then when analyzing new phenomena, we can also expect it to be runnable on a very limited simulation, too.

It's the same as assuming that the law of conservation of energy would apply to any new phenomena we discover. There is no proof, but it's a good starting point.

You cannot through experiments prove that the law of conservation of energy holds in all cases. Indeed you cannot through observation prove that any law of nature holds in all cases. But you only need a single counterexample to show that a law doesn't hold in all cases. This is exactly the rationale behind the principle of falsification. You cannot prove through observation that all swans are white. But you can prove that not all swans are white by a single observation of a black swan.

The bottom line is that it is possible to devise experiments or observations which would give different outcome if the theory holds and if the theory doesn't hold.

But you cannot devise an experiment which would give different outcomes if the universe is a simulation versus if it isn't. More specifically, you cannot imagine an observation which would prove that the universe isn't a simulation. Therefore it is not a legitimate scientific theory. At best it is a fun thought experiment.

> But you cannot devise an experiment which would give different outcomes if the universe is a simulation versus if it isn't.

I've already mentioned 5 times in this thread that the theory I'm talking about is of a simulation with very limited resources, but you seem to be ignoring this point.

Show that a particular phenomenon can't be simulated using extremely limited hardware (compared to the size of the universe) and you can falsify this theory.

I didn't ignore that, I mentionend the "limited resources" scenario with the 16GB example a few posts back.

> But you cannot devise an experiment which would give different outcomes if the universe is a simulation versus if it isn't.

Does this or does this not apply to a simulation with very limited resources (compared to what it would take to model all the atoms of the universe) or not, then?

> The luminiferous ether was a falsifiable theory

But it effectively wasn't for the first 150 years of it's existence. The theory originated in 1720 and it wasn't disproven until 1887. It took that long for someone to come up with a conclusive experiment. Nevertheless, for those 150 years the luminiferous ether theory was widely held, shaped the field of physics, and the research had to go into finding an experiment to test the theory.

Similarly, the idea that atoms might do things when high energy particles hit them was all theory until Wilson's cloud chamber became the first particle detector and suddenly physicists could see what these particles were doing when they hit things.

> The problem is...you can even define what it means for the universe to be a simulation.

You're confusing something that hasn't been done with something that can't be done. Indeed, what they're trying to do is exactly trying to define criteria to test that the universe is a simulation. It's the first day of school and you're complaining that Newton doesn't have his Principia and Gilbert doesn't have his De Magnete.

Theories often predate the knowledge of how to test them. This isn't weird.

In order for a theory to be falsifiable it first have to be a theory! The simulation hypothesis is not even a theory (i.e. it makes no verifiable predictions), so it is not even falsifiable in principle.

Reply on news.ycombinator.com