Wednesday, 4 October 2017

A virtual universe?

I think many people are getting the implications of the Simulation hypothesis wrong.  In the above article Cory Doctorow, to sum up, presents it as:

1.  There are likely many advanced civilizations in the universe.
2. Over time an advanced civilization would be able to produce a simulated reality as rich as our own.
3. They would likely produce millions it not billions or trillions of such universes
4. Therefore a person in a universe is most likely in a simulated universe, as there will be a vast number of these for every real universe.

He takes this simple idea apart, but this form of statement of the simulation hypothesis is really just a straw man.  Certainly this is one way of thinking of potential simulated reality; but it is not the only or best way, and it really misses the deeper point.

I would point out that philosophers from Plato of Descartes presented a form of the simulation hypothesis in their works.  Plato, as far as I can tell, seems to believe it in a form while Descartes thinks he proved it is not true. 

This idea which has had such a massive impact on western thinking is not as silly as the one present by Doctorow nor as easily disproven.

A more accurate summing up of the simulation hypothesis would simply be:

This universe is really a simulation run in an information machine. 

And the idea is a flavour of a much deeper and more important idea:

Our universe is really just information.

The second wording is probably widely accepted by Quantum theorists and Cosmologist today as far as I can understand.

What is interesting is that from the point of view of an observer in the Universe it might be hard to tell the later from the former definition,.  If my reading of modern Quantum theory is correct  it may be said that the universe might as well be a simulation.

I will focus in on one of Doctorow's main points against simulation: that to simulation all the particles in our universe would take a device so extremely massive as to be massively larger than our universe. And I quote

It turns out that figuring out these particles' locations scales at n^2, meaning the amount of computing power needed doubles with each additional particle, which means that "storing information about a couple of hundred electrons would require a computer memory that would physically require more atoms than exist in the universe."


A simulation does not need to model each particle of it's universe, it only needs to fool a observer for each and every observation they might make. 

Think about if you were going to make a simulation of a universe to fool a conscious being into thinking it was in a real universe.  You could write a program for each and ever particle in the universe meaning you could only build a small simulation relative to the universe in which you started.

Or you could be more clever.  You could keep more high level descriptive data of the universe, and then when your conscious observer asked about the existence of a particle you could just come with the answer for that one question on the fly. 

You could answer these questions with a set of probabilistic models that govern the details of a coarser grained universe, and when someone asks a question the answer would be governed by a model on the fly.

Sound familiar.  This is what we get with quantum mechanics.

And it gets more interesting as you bush this further.

Ask yourself ways that someone could figure out if the universe was being modelled via bits.  You could ask if the universe was a simulation lying to you with far less information then it seems to have.

I could random select a particle, which I am utterly free to select at will, and the universe would then have to give me it location or speed.  I will get to this in more further.  Then I could follow that same particle randomly for as long as I want.  By doing this the computer would have to keep data about the history of all particles I might want so that the universe could keep its 'story straight'.

I could then ask where a particular particle was over a vast amount of time making sure that the universe was keeping everything straight, right?

This is the standard way we find out if people are lying, we ask about the same thing over and over, going over the story until something does not fit and we conclude we are being lied to.

One problem is that this is not how the universe works. 

I assume my reader knows something about Quantum Uncertainty.  If not click the link.

When I look at a particle on the quantum level, I then can never ask another question about that particle.   I can then only get a limited set of one time data points about a particle in an initial observer, then I can never check it again.

It seems like the universe is saying, when really pressed, that it won't take any more questions.

Its almost as if some programmer is trying to make it so that I can be fooled by a system that did not need to keep track of where ever particle is?

I am not saying that Quantum theory means that the universe is a simulation, but I am saying that the simulation hypothesis is a better metaphysics for the quantum universe than our ordinary assumptions about how reality works from day to day life.  We assume things have speed and location, and that things remain things moving along set paths over time.  On the quantum scale we find these assumptions break down.

But, and this is where things great really exciting, quantum physics makes perfect sense if we assume the universe is a computer model and the programmers could not model every particle in the universe.  Rather we can ask about a single particle once, and our answer is determined by probability theory.  We could fool an observer this way by using a probability formula to answer questions about a single particle, and then not allow then not allow any follow up.

This does not mean we are living in a simulation, it means we might be living in a simulation.  The thing that is certain is we are not living in a universe that acts the way we assume it does before we know Quantum Uncertainty.  Its our everyday universe in our own mind which is a false simulation.

Step back and think about how you would fool an observer about a universe, making them think the universe was vastly bigger than it was.  That is how would you create the experience of a massive universe using fairly limited computing resources.  This is a very common issue with simulations, you keep the model running a set of variables that sum up the interactions of invisible lower particles.

In the 1990s I created a economic simulation with this very goal in place, I wanted to fool people in to believing that it was a proper market place.  I created a model that simulated millions of consumer deciding what to buy with a few variables. 

It worked pretty well.  And it worked because I hid the information about actual consumer, I didn't allow you to drill down to the person in the model.  And I also added levels of probability in determining the sales in the market based on the players turns in the simulation.

I was able to make the kind of computing available to be in the mid 1990s on the LambdaMoo MUD object oriented programming language.  We even had to deal with the fact LambdaMoo didn't support real numbers. 

Dozens of users felt the game was realistic, and the outcome over time confirm to economic theory, a few players with early advantages were able to form a monopoly over certain markets just as real retail markets tend to behave.

Now lets say someone is playing a simulation, they are able to make precise observations about the universe, but I am assuming the universe is not really keeping detailed information about the particles when not observed.  How can I get away with it.

For example even given quantum uncertainty I might be able to look at say a cup or water and follow it conduct to determine, given a long enough time of observation, what the life spans of all the particles would have to be to check that the computer is keeping track of all the particles, that the program is keeping its story straight.  By looking at the system long enough I could collect enough information to conclude what the trillions and trillions of atoms in the glass of water are doing.

Now imagine I wanted to pull this off with a fairly limited computer.  How would I keep an observer from making billions of observations of a glass of water in order to drill down to a model of every molecule?  How about I make it so that as the user observes the system the amount of data they can capture is decreasing over time. 

I could do it by making the case that each 'turn' of the simulation I get to hide more and more information to compensate for what the observer is collecting, so that their total knowledge of my universe never excides my computer model.

I know this is a bit esoteric but I would argue is that is precisely how entropy works, it is like as we collective knowledge about system more and more underlying knowledge becomes hidden, meaning the systems could be simulated with a probability computer, really just generating a limited number of random numbers each time I make an observation. 

Again this is NOT to say we are living in such a simulation, that our universe is actually a much smaller computer trying to full us that it is a bigger reality, but the universe we find ourselves and understand can be fully explained in this idea, while the ideal of a all knowing God who knows were everything is at all times seems to be impossible.

So the simulation hypothesis turns out to be more than just a silly argument about alien races and computer systems, it is one of the greatest thought experiments into trying to define ourselves and more deeply understand how our physics emerges as a kind of dance between our minds, which run an inaccurate simulation, and a universe that looks more and more like an information system.

No comments:

Official Linden Blog