Battlefield 3 Unlocks the Secrets of the Universe

[GMMichael]

I got hit by a whopper the other day. Hopefully some learned folk in here can help me suss this out:

Many first-person shooter games do a lot of CPU outsourcing. There's always a "host" system, but in order to make the game run faster and smoother, the "client" systems make a lot of the decisions for the host. One of these decisions is called "client-side hit detection":

Client-side hit detection:[sblock]
In client-side hit detection, each system gets intermittent reports from the host about where the other players in the game are. When you shoot, or "hit," an opponent, it's your system that decides that the hit actually occurred, not the host. Where this gets fuzzy: since each client gets its updates about player position at different times, you can see an opponent in one position while your opponent sees himself somewhere else (probably nearby). When you shoot your opponent, the hitting takes place where you see him, generally not behind cover, even if he sees himself as behind cover.[/sblock]

Here's where it gets interesting: in a first-person shooter (say, Battlefield 3, which I was playing at the time) there is no single, ultimate game going on in the present. Each client has a slightly different game going on from every other client. The host does some processing and records what actually happened, but this recording becomes the past. The future is based on a set of rules upon which all systems, client and host, agree.

Does this resemble general relativity?

I'm drawn to the example that two observers are witnessing an event occur. One, of course, is moving at "a high rate of speed." So when the event occurs, each observer sees the event occur at a different time. What's usually implied is that event though the event occurs at different times to each observer, they are still supposed to be in the same universe.

They are both in the same game.

Two observers can see the event occur at different times because they are each on a different client-system.

Does Schrödinger’s cat live or die? It depends on which system hit-detects his box opening first.

If everything is relative, how do we agree that only one past, or history or game, occurred? Because the host-system records the history upon which all clients eventually agree.

Why do quantum particles pop in and out of existence in empty space? That's your client-system waiting to see what the host declares is history.

Why can't we say exactly where a sub-atomic particle is, without changing its position by measuring it? Because our client-systems have an idea of where things are, where players are, but this is fuzzy due to the fact that each client-system, and the host, can change where a particle is at just about any time.

Einstein should have been a gamer.

 

[tomBitonti]

That's looking at both special relativity and quantum mechanics. (General relativity unifies special relativity with gravity.)

But that's a quibble: The ties between quantum mechanics and features of computing systems are fascinating. The fit can seem eerily close. (I'm drawn to quantum entanglement compared with lazy evaluation, modelling an entangled pair of particles a as system having a single, distributed, state which is realized only when one of the particles has an interaction which requires the realization.)

Thx!

TomB

 

[freyar]

The first point is that we should really separate relativity (both Einstein's theories, special and general) from quantum mechanics. Then there's the combination of the two, or quantum gravity. We should probably think about each type of physics separately. I'll start just by saying: fun question!

First off, special and general relativity: I'd say thinking about host and client systems as you describe isn't very reflective of relativity as Einstein presented it (or the math, really). Every observer agrees on what physical events happen. So, both observers see one player get hit, and they see it happen at the same "point of space-time," meaning if I get hit at the corner of the building when the clock at that spot shows 12 noon (or whatever), everyone agrees on that. But if you (at rest according to me) get hit two blocks away when your wristwatch reads 12 noon, different observers will disagree on which of us gets hit first and how many meters apart we are. Certain events in a relativistic space-time are unordered, and our notions of distance and time can get bent a bit.'

Quantum mechanics: This is a bit more interesting, though your proposal might be hard to implement with the mathematics of quantum mechanics (which just work). I might add that this sounds to me more what you're asking about. Anyway, what happens in orthodox quantum mechanics is that some portion of the universe is in an undetermined state (Schrodinger's cat is neither and both dead and alive at once, in likely a time-dependent proportion) until something makes a "measurement." Then the universe "decides" once and for all what the state is --- the cat "becomes" either dead or alive. Here you might be able to work in some client (measurement)/host (universe) dichotomy with that. However, the issue with this approach to quantum mechanics is that it treats the part of the universe that does the "measuring" differently than the part that gets measured. Many physicists prefer some variation of what's called the "many worlds" approach (somewhat tongue-in-cheek). In this version, measurement makes the whole state of the universe split --- so before, the cat is dead & alive; after, we think the cat is dead PLUS we think the cat is alive. It's just that the two different "pieces" of the universe can't talk to each other. This is actually a lot more self-consistent in terms of logic. So the host computer in this sense is doing parallel computing and different clients sort of have different games, in a hand-wavy sense.

Quantum gravity is a lot more fuzzy since it's hard to even know what we mean by space, time, or events in some ways. It has to incorporate both of the relativistic and quantum points of view. But it's hard to see how to make a sensible theory that doesn't have different observers agree on the actually happenings.

Of course the main thing I haven't stated is that, however we want to think about it, we have to put it into math to make a real physical theory out of it. The descriptions above are, at the core, just (somewhat clumsy) translations of the math into English.

 

[Umbran]

Does this resemble general relativity?

As others have already noted... not really.

I'm drawn to the example that two observers are witnessing an event occur. One, of course, is moving at "a high rate of speed." So when the event occurs, each observer sees the event occur at a different time. What's usually implied is that event though the event occurs at different times to each observer, they are still supposed to be in the same universe.

They are both in the same game.

Two observers can see the event occur at different times because they are each on a different client-system.

In General relativity, we never disagree about what happened. We never have the case where I think there was a tree between you and the target, and you do not think there was a tree, and one of us determines which is correct. We always both agree that you had a clear shot. We may disagree on when the event happened, in comparison to other events, but there is never a case of, "actually, you were wrong, and this event actually did occur, please update."

Does Schrödinger’s cat live or die? It depends on which system hit-detects his box opening first.

Again, not really. Because we both already agree that there's a 50% chance the cat lives, and a 50% chance it dies. Unlike the video game analogy, where we each think the universe is in a different state, and a decision is made about which of us is correct, in QM, we generally agree on the state of the universe, and it doesn't matter at all which system detects it first. We both agree that, until one of us looks at it, *nether* of us knows the actual state in the box.

If everything is relative, how do we agree that only one past, or history or game, occurred? Because the host-system records the history upon which all clients eventually agree.

The client/host game system is about information lag. Each client has its own little universe, and it may differ from the server's. Every once in a while, you get an update to match your client universe to the server, and when you update, some of your history gets rubbed out as incorrect. In the real world, there is no ultimate "host" system. There are only clients.

Why can't we say exactly where a sub-atomic particle is, without changing its position by measuring it? Because our client-systems have an idea of where things are, where players are, but this is fuzzy due to the fact that each client-system, and the host, can change where a particle is at just about any time.

Except, in the client/host view of things, each one of us thinks we know *exactly* where the particle is. But, for some reason one of us is decided to be correct, the other incorrect, and there is a discontinuity when the system updates, and some of us see things teleport about to match the chosen reality, and others don't. The host/client universe has inconsistent histories, while we do not.

Einstein should have been a gamer.

The guy who said that God does not play dice with the world, probably not so much a gamer

 

[Janx]

I'm guessing that for 2 gamers on 2 xboxes that they don't know what's actually happening inside the black box of the game system, it might seem like relativity/quantum mechanics/uncertainty.

But as the physics dudes say, it's not actually the same.

 

[SkidAce]

In the real world, there is no ultimate "host" system. There are only clients.

Or is there?

 

[GMMichael]

Well gang, I'll grant that 1) the laws of quantum mechanics and general relativity work really well, and that 2) you have much better understandings of these laws than I do. But I believe that the great unifiers are still battling it out, to see how the two laws (theories?) come together. And since I haven't heard explanations lately for where all the universe's dark matter is hiding, or why the universe's expansion is accelerating, I hope that we can agree that there are still some flaws in the maths.

I'll ask about relativity first:
[MENTION=40227]freyar[/MENTION]: how can every observer agree on what physical events happen, but not when they happen? The "when" seems pretty important to me, unless "when" is only a number on a clock. If you were in the open at 12:01 but behind cover at 12:02, then I hope I shot you at 12:01.
[MENTION=177]Umbran[/MENTION]: see the above case. If we can disagree on when getting shot occurred in relation to other events, and if another event is the arrival of a tank between you and me, then that tank might mean the difference between life and death.

Also, shooter games do a certain degree of smoothing, so that updates to your game world don't look (too) awkward. If the universe runs on something similar to the Frostbite game engine, then there wouldn't be any obvious "please update" moments. Just tiny inconsistencies, like a single photon causing an interference pattern with itself.

 

[Umbran]

And since I haven't heard explanations lately for where all the universe's dark matter is hiding, or why the universe's expansion is accelerating, I hope that we can agree that there are still some flaws in the maths.

That would depend on what you call a "flaw", I suppose. Current theories are not perfect, if that is what you mean. However, I think our positions are based less in the math, and more in what is observed in the universe, and that happens to be consistent with the math.

[MENTION=40227]freyar[/MENTION]: how can every observer agree on what physical events happen, but not when they happen? The "when" seems pretty important to me, unless "when" is only a number on a clock. If you were in the open at 12:01 but behind cover at 12:02, then I hope I shot you at 12:01.

The time and distance differences are not arbitrary, and are linked to the relative speeds involved - we can't end up with them being at just any old time, and the math works out such that we will not disagree on what actually happened.

The only things that we will disagree about the time-order are things that are "non-causal" - we will only differ on whether event A came before event B when A could not have affected B, due to the lightspeed barrier. If not even light could have passed from one event to the other, then they could not impact each other, and their order is largely arbitrary.

[MENTION=177]Umbran[/MENTION]: see the above case. If we can disagree on when getting shot occurred in relation to other events, and if another event is the arrival of a tank between you and me, then that tank might mean the difference between life and death.

See the above answer. If it is a matter of life and death, then the order of events will be preserved, whoever is looking at it. General Relativity preserves causality.

Also, shooter games do a certain degree of smoothing, so that updates to your game world don't look (too) awkward. If the universe runs on something similar to the Frostbite game engine, then there wouldn't be any obvious "please update" moments. Just tiny inconsistencies, like a single photon causing an interference pattern with itself.

Two things:

1) "tiny inconsistencies" sounds like it isn't a big deal, but the word "tiny" is being used in a human subjective sense, not in a technical sense. The corner cases those supposedly tiny inconsistencies can generate are going to be issues.

2) You are leaving out a major point - in the client/host arrangements, for any given fact, one client will need updating, and another will not. The view of two observers will not be consistent - thus we call it an inconsistency. In the real world, as we've pointed out, for both GR and QM, we don't disagree about events that can impact each other. Our observed universes are consistent! If we had cases where one observer saw an electron, and another saw an interference pattern, then you'd have a point. But we *BOTH* see an interference pattern - so it isn't an inconsistency at all.

 

[tomBitonti]

Here's a thing: The problem that you describe, of latency between clients, is exactly what is spookily solved by entanglement. Two particles are emitted, with balancing properties. One heads to the left, the other to the right. You might expect that the selection of which particle went in which direction to be set at the point of emission, but, it works out that decision isn't made until one of the particles is observed.

The strangeness is that because the particles have opposite properties, this fixes both particles. This happens no matter how far apart the particles are, and with no propagation delay. Physics avoids a problem which seems to be the same problem you describe between clients.

TomB

 

[Umbran]

Here's a thing: The problem that you describe, of latency between clients, is exactly what is spookily solved by entanglement. Two particles are emitted, with balancing properties. One heads to the left, the other to the right. You might expect that the selection of which particle went in which direction to be set at the point of emission, but, it works out that decision isn't made until one of the particles is observed.

I'm sorry, but that's not quite correct.

With a latency issue, there are two people *who disagree* about the state of the universe. What I observe and what you observe is different, and then gets overwritten with an update later. In latency, the universe gets rewritten every once in a while to make our views consistent.

That doesn't happen with entangled particles. In that case, at one point we *agree* that the particles are unobserved, or we *agree* what the states of them are. There is never a time when we disagree on what is going on in the universe. This is an important distinction - we cannot compare the latency situation in which we have both seen events, and they don't match, with the QM situation where neither of us have yet seen events, and the state of the system is not known to either of us. They are not equivalent situations.