The irreversibility of time

bogosort

Joined Sep 24, 2011
884
Two important points:
  • The universe can't be simulated
  • Time is not reversible
I haven't read the original paper, but it does not seem like the two points you mention are necessary consequences. For one, obviously we can simulate the universe; we do it all the time! Now, if by 'simulation' one means a perfectly accurate copy, i.e., something indistinguishable in principle from the actual universe, then we've known for a long time that such a thing is impossible. Quantum mechanics -- in the form of uncertainty relations, indeterminate states, the no-cloning theorem, etc. -- forbids it. Full stop.

Even if one takes 'simulation' to mean a program that produces a reasonable facsimile of the observable universe, we've long known that this is impossible from within the universe itself. There are many technical reasons why this would be true, chief among them energy constraints. The observable universe has unbounded degrees of freedom, while any computing device would be subject to strict information density limits (the Bekenstein bound). In short, any computer that's big enough for the job would collapse into a black hole long before it'd be able to do it.

There's also the crucial fact that, in order to accurately simulate the universe, one needs to know all of the laws of physics. As we don't (yet?) know all the laws of physics, we can never write an accurate simulation.

But I think all of this is secondary to the elephant in the room: chaos dynamics. Forget simulating three self-orbiting black holes, try to simulate a marble balancing on the head of a pin. As an unstable system, the marble will always fall. But it will have an innumerable number of possible ways to fall, with an innumerable amount of resulting trajectories, dependent on an innumerable numerable of initial conditions, many of them based on quantum states. Playing the simulation in reverse will almost certainly lead to different trajectories.

That's the defining characteristic of chaotic systems. I don't understand why anyone should find it surprising that the N-body problem is chaotic, even with Planck-scale resolution. More importantly, the fact that we cannot in principle predict where the marble will fall does not in any way prevent us from writing simulations of falling marbles, nor does it stop us from understanding the physics. Time-reversibility is a mathematical notion, not a physical one.

To put it another way, I can write a deterministic program that produces sequences of random numbers using a PRNG that's seeded with the entropy of some random physical process, like the noise from a Zener diode. Everything about the program is perfectly known, yet no matter how often we run it, we'll never be able to predict the initial conditions from the output sequence. The program is not time-reversible, but so what? We understand it perfectly; we can even "simulate" it with another program. That the original program and the simulation will not produce the same output sequence is inconsequential to the understanding of it.

So, if we're talking about simulating the universe, and we're doing so in terms of understanding it, then the fact that we can't replay the simulation backwards to get to the initial conditions is totally expected and inconsequential. On the other hand, if we're talking about simulating the universe as an attempt to disprove that the universe is a simulation, then any arguments about simulating the universe within the universe are clearly moot.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
6,873
I haven't read the original paper, but it does not seem like the two points you mention are necessary consequences. For one, obviously we can simulate the universe; we do it all the time! Now, if by 'simulation' one means a perfectly accurate copy, i.e., something indistinguishable in principle from the actual universe, then we've known for a long time that such a thing is impossible. Quantum mechanics -- in the form of uncertainty relations, indeterminate states, the no-cloning theorem, etc. -- forbids it. Full stop.

Even if one takes 'simulation' to mean a program that produces a reasonable facsimile of the observable universe, we've long known that this is impossible from within the universe itself. There are many technical reasons why this would be true, chief among them energy constraints. The observable universe has unbounded degrees of freedom, while any computing device would be subject to strict information density limits (the Bekenstein bound). In short, any computer that's big enough for the job would collapse into a black hole long before it'd be able to do it.

There's also the crucial fact that, in order to accurately simulate the universe, one needs to know all of the laws of physics. As we don't (yet?) know all the laws of physics, we can never write an accurate simulation.

But I think all of this is secondary to the elephant in the room: chaos dynamics. Forget simulating three self-orbiting black holes, try to simulate a marble balancing on the head of a pin. As an unstable system, the marble will always fall. But it will have an innumerable number of possible ways to fall, with an innumerable amount of resulting trajectories, dependent on an innumerable numerable of initial conditions, many of them based on quantum states. Playing the simulation in reverse will almost certainly lead to different trajectories.

That's the defining characteristic of chaotic systems. I don't understand why anyone should find it surprising that the N-body problem is chaotic, even with Planck-scale resolution. More importantly, the fact that we cannot in principle predict where the marble will fall does not in any way prevent us from writing simulations of falling marbles, nor does it stop us from understanding the physics. Time-reversibility is a mathematical notion, not a physical one.

To put it another way, I can write a deterministic program that produces sequences of random numbers using a PRNG that's seeded with the entropy of some random physical process, like the noise from a Zener diode. Everything about the program is perfectly known, yet no matter how often we run it, we'll never be able to predict the initial conditions from the output sequence. The program is not time-reversible, but so what? We understand it perfectly; we can even "simulate" it with another program. That the original program and the simulation will not produce the same output sequence is inconsequential to the understanding of it.

So, if we're talking about simulating the universe, and we're doing so in terms of understanding it, then the fact that we can't replay the simulation backwards to get to the initial conditions is totally expected and inconsequential. On the other hand, if we're talking about simulating the universe as an attempt to disprove that the universe is a simulation, then any arguments about simulating the universe within the universe are clearly moot.
My argument (and I'm pretty sure the article's argument too) is simply that the universe cannot be simulated with 100% accuracy. That is, with perfection. At least, as you've just said, not within this universe.

And as for the marble on the head of a pin (or even on its point), I bet you a week's supply of beer that it can be done... :p
 

bogosort

Joined Sep 24, 2011
884
Life itself has no meaning unless those things are true. So I'm inclined to agree.
It seems very strange to me that the meaningfulness of life would depend on certain conditions of the universe being satisfied or not. What would change if you found out tomorrow that our universe is actually a simulation by some unimaginably advanced creature? Would your life suddenly lose meaning?
 

bogosort

Joined Sep 24, 2011
884
My argument (and I'm pretty sure the article's argument too) is simply that the universe cannot be simulated with 100% accuracy. That is, with perfection. At least, as you've just said, not within this universe.
Quantum mechanics has been telling us that for 100 years. There's no such thing as 100% accuracy; it's an impossible idea, like greater-than-unity energy mechanisms. So, what does that have to do with simulations?

And as for the marble on the head of a pin (or even on its point), I bet you a week's supply of beer that it can be done... :p
Lol, well, if we're going to bet, then we need constraints. Let's say you have 1 minute to balance an ordinary marble on a sharp pin.

In any case, I'm sure you recognize that even if you win the bet, the marble is in a highly meta-stable state. All it takes to devolve into chaos is a tiny local temperature gradient!
 

bogosort

Joined Sep 24, 2011
884
Personally, I don't believe in free will. But my daily existence wouldn't change one bit if I found out definitively that I do or don't have free will. Why should it? Life is only as meaningful as I make it. Whether my love for my daughter is borne of my own volition or genetically programmed into me, what does it matter? I'll still love her the same.
 

Berzerker

Joined Jul 29, 2018
589
cmartinez said:
Time is not reversible
I thought Einstein said space and time are interwoven and predicted you could go forward in time.
Of course he also stated that travel to the past wasn't possible but added you could leave at the speed of light and travel out far enough and long enough that when you came back more time would have passed on the earth than for you.
But later when a friend of his passed he wrote:
Einstein said:
"...for us physicists believe the separation between past, present, and future is only an illusion, although a convincing one."
So he contradicted himself later.
Edit: I was getting to that too, Wendy.
Brzrkr
 

xox

Joined Sep 8, 2017
452
  • Time is not reversible
Not to be overly pedantic, but that's not exactly true. Time is most definitely reversible, being one of the many possible states of a given system. But just as you've probably never heard of say one billion coin flips resulting in "all tails", the possibility of seeing a spilled drink jump back into its glass is likewise equally unlikely (if not more so)...
 

xox

Joined Sep 8, 2017
452
I thought Einstein said space and time are interwoven and predicted you could go forward in time.
Of course he also stated that travel to the past wasn't possible but added you could leave at the speed of light and travel out far enough and long enough that when you came back more time would have passed on the earth than for you.
But later when a friend of his passed he wrote:
[ ... ]
So he contradicted himself later.
Every single time you experience a change in accelleration, you are in fact time travelling! Weirder still, there is no actual speed limit. Get up to 99% the speed of light, turn off your engines, then do it all over again. Rinse and repeat and you're in, say, 9862020 A.D. within your own lifetime. That's not science speculation either - that's a science fact!

As far as the quote in the letter, I think what he's hinting to is this implication that light itself experiences no "time", that the past, present and future are but a single "moment" from the viewpoint of a photon after all.
 
Top