How does quantum mechanics give rise to parallel universes? (Patreon)
Content
[This is a transcript.]
Physics is fascinating because, well, because of many reasons of course, not least that it works. But one reason is certainly that it gives a touch of plausibility to ideas that otherwise seem like pure fiction, strange objects that you think shouldn’t exist: Wormholes, invisibility cloaks, or birds.
But the physics idea that has captured imagination the most are probably parallel universes. The “Many Worlds” interpretation of quantum mechanics has it that anything which can happen, does happen, in another universe. Really? In this much-asked-for video, I want to explain how the Many Worlds Interpretation works. Is it science? Does it solve any problems? And are the parallel universes real? That’s what we’ll talk about today.
When you look around yourself, you see a lot of stuff, stuff that is somewhere, in some particular place. Or if you’re watching this in a pub, maybe stuff that is walking around on two legs. But legs or not, everything around you has very definite properties. It’s not in two places at the same time, or smeared out across the room. Or if it is, maybe you’ve had enough beer for tonight.
If you look at very small particles, however, things are very different. Take for example our good old friend, the electron. An electron is described by a wave-function, usually denoted with the Greek letter Psi. But this wave-function doesn’t tell you where the electron is or where it’s going. From the wave-function you just calculate the probability of what you’ll find when you measure where the electron is or where it’s going.
For example, the wave-function tells you the probability of finding the electron in a particular place. But before you measure it, it isn’t in any particular place. It could be in several places, or indeed be smeared out across the room.
According to standard quantum mechanics, often called the Copenhagen Interpretation, nothing has definite properties until you measure it, not just small particles but also large objects. It’s just that large objects, like that pint in front of you, are constantly being “measured” in some sense because light and air molecules ping off them. The Copenhagen Interpretation does not explain just why some interactions result in measurements and others not, this is the “measurement problem” of quantum mechanics.
What makes quantum mechanics so weird is this switch from a particle that is described by a wave-function without definite properties, to a particle that has a definite property which you know because you just measured it.
Say you take a quantum of light, a photon, and you send that through a beam splitter. No, we don’t take our friend the electron. Friends don’t let friends go through beam splitters.
Because if you send a photon through a beam splitter, then its wave-function says that the photon goes both ways. It’s in what’s called a superposition of both possibilities. Let’s say that each has a 50 percent probability. But once you measure the particle on the right side, you know it’s not on the left side. This means you have to update the wave-function. Since the wave-function only tells you a probability, the outcome of a measurement can’t be predicted with certainty. This is why quantum mechanics is not deterministic.
This update of the wave-function is also sometimes called the collapse or reduction of the wave-function and it’s a key element of quantum mechanics. If you don’t update the wave-function, you will get wrong probabilities. If you want to know for example, what’s the probability of measuring the particle on the left given that it was measured on the right, the answer should be zero. But this only comes out correctly if you update the wave-function.
The update of the wave-function is instantaneous. It happens at the same time everywhere and is the reason why quantum mechanics is non-local. This wave-function update is what Einstein called a “spooky action at a distance”.
I did an entire video previously about what this means, but here’s the brief summary. Quantum mechanics in its standard version is non-local as a matter of fact, that’s just a property of the theory. The key question is whether this non-locality is also a property of reality. Whether that’s the case or not depends on whether you think quantum mechanics is fundamentally correct, or just a description of an underlying reality. We still don’t have an answer to that.
If quantum mechanics is fundamentally correct, then the world is non-local, period. But if there’s an underlying reality in which the outcome of a measurement was determined, we just didn’t know of it, then this reality could well be local. This is called a hidden variables model.
Think back to the example with the beam splitter. If there was an underlying reality, the photon went either left or right, you just didn’t know what it did until you measured it. In this case, the wave-function would just describe your incomplete knowledge. And while in this case the update of the knowledge is still non-local, that’s not a problem, because the photon itself travelled entirely locally to the place where it’s being measured. For this to work, however, you need something else, more variables, that tell you where the photon “really” went.
Since it’s a rather persistent myth, let me add that hidden variables models have not been excluded by experiment. But it’s somewhat off-topic and I explained this already in a previous video. For today the important point is just to understand the reason why quantum mechanics is non-local.
The reason that quantum mechanics is non-local is a combination of (a) the observational fact that a measurement outcome in one place tells you something about another measurement outcome in another place. If you measure the particle here, you now know you won’t measure it there. Fact, not interpretation. And (b) the absence of other variables in the theory that could have carried the information locally. This is why quantum mechanics is non-local. And this is also why hidden variables can restore locality. But this is not a video about hidden variables…
The Many Worlds interpretation now is based on the idea that you can throw out the update of the wave-function by re-interpreting what happens in a measurement. According to this interpretation, all outcomes of a measurement happen, each in its own universe. But we can only ever see the result in one universe, so for us it *looks like the wave-function collapses.
Instead of the measurement update, in many worlds, we have what is called a “branching” or “splitting” of worlds. This branching makes it impossible for one observer to see more than one outcome of a measurement. The major challenge for many worlds is to explain why the thing we call an observer does not itself branch with those worlds therefore sees all the outcomes, but somehow randomly only experiences one of those worlds. I have never found a good explanation for that.
But in all fairness, from a purely instrumental point of view, if you only ask for the outcome, it isn’t doesn’t really matter WHY observers see only one outcome. You just assume that this is somehow the case, which is as unsatisfactory as the measurement in standard quantum mechanics, but then Many Worlds makes the same predictions and standard quantum mechanics.
It also leaves you with the mind blowing idea that each time a quantum particle bounces off another one, which happens gazillions of times a second, our entire universe splits, and anything that can happen does happen. Had salad for lunch today? Well in some other universe you had pizza, with Elon Musk, on Mars. Whatever you can think of, so long as it respects the laws of nature, it’s real, in some parallel universe.
Since I get this question frequently, energy is conserved just fine in the many worlds universe. It’s not like you duplicate all energies each time worlds branch. If you calculate the entire energy of all those worlds, then you must give weight to each world by the probability with which it came into existence. And that works out the same way as it does in normal quantum mechanics. So there’s no problem with energy conservation.
No, the biggest problem with many worlds is that its supporters believe their interpretation is somehow better than the standard interpretation with the collapse when it’s really just as mediocre.
Many worlds supporters often claim that their interpretation is simpler because it just does away with the collapse postulate. But as we saw earlier, you need the collapse postulate to calculate probabilities. You can’t just throw it out, that doesn’t work. And indeed, this is not how the many worlds interpretation works. It’s how Many Worlds supporters *say* that it works, but it’s not true.
At this point things get a bit murky because there isn’t just one many worlds interpretation. There are two original ones, going back to Hugh Everett and Bryce DeWitt, but meanwhile there are dozens of slightly different versions. They differ in how they deal with the branching of the worlds, but they all have to make new assumptions about how a measurement works and under which circumstances it happens. I guess it’d better be called the many many worlds interpretation. Hey, there’s untapped potential here. How about many many many worlds?
It should not be surprising that many worlds interpretations need new assumptions. Let’s just leave aside all the talk about interpretations and imagine we’d have to write an algorithm for a computer. If we take standard quantum mechanics and remove the collapse postulate, the resulting algorithm simply does not give predictions that agree with observations. The collapse postulate is there for a reason.
I don’t mean to say that the collapse postulate is the *only way you can do it. In QuBism for example, which is another interpretation of quantum mechanics, the update of the wave-function is interpreted as a Bayesian update of knowledge. And this is just as well in terms of predictions as the collapse postulate.
And likewise, to make the many worlds interpretation work, one needs to add other axioms. You can interpret those as something to do with observers in branching worlds, but if you look at it algorithmically, in the end they just do exactly the same thing as the collapse postulate. So it’s not correct that the many worlds interpretation is simpler. They take out one axiom, but have to replace it with others.
Those who believe in many words, excuse me, many worlds, also often claim that their theory is local. I am guessing they believe this because they have thrown out the collapse postulate which is the non-local element of the standard interpretation. But as we saw earlier, you can’t just throw out the collapse postulate, it needs to be replaced with something else. That’s of course also the case in the Many Worlds interpretation and as a result, it’s exactly as non-local as the standard interpretation.
To see why, remember that the reason quantum mechanics is non-local is (a) that a measurement in one place does, as a matter of fact, tell us something about what happens in another place. And (b) the theory has no variables that could transport this information locally. These two reasons are still fulfilled in many worlds. Consequently, it’s exactly as non-local as quantum mechanics with the collapse postulate.
I find it surprising how many physicists are confused by this. Lots of papers have been written about how many worlds can be made local. But of course, the only way to make it local would be to introduce some kind of hidden variable that transports information locally.
This was exactly the point of the famous paper by Albert Einstein, Boris Podolsky and Nathan Rosen, now just known as the EPR paper. They said, if you want reality to be local, you need an element of reality that underlies quantum mechanics, therefore quantum mechanics is incomplete. The paper’s now almost 90 years old, but physicists still don’t get it, do they.
More worrying still, I’ve recently noticed a curious development which is what actually triggered this video. It’s that Many World adherents have managed to convince themselves that they avoid Bell’s theorem. They are claiming that Bell’s theorem has a hidden assumption which is that a measurement has only one outcome. They call it the “one world” assumption.
Bell’s theorem, as a reminder, says that all theories which are local and fulfil measurement independence predict correlations that are in conflict with observations. These observations are what the 2022 Nobel Prize in physics was awarded for. And since the Many Worlds interpretation fulfils measurement independence, it must be non-local.
But many worlds supporters don’t want to admit that their theory is exactly as mediocre as the standard interpretation. They want to believe it’s local. So they need an explanation for how they avoid conflict with Bell’s theorem. And to get this done, they have invented this supposed extra assumption, that a measurement has only one outcome.
Well first of all this assumption isn’t exactly hidden. But more importantly, Bell’s theorem is about correlations that we observe, and it’s an observational fact that we only ever observe one outcome of a measurement. If you throw out this assumption from Bell’s theorem, you have a theorem that isn’t about what we observe for a theory that doesn’t explain what we observe.
As Einstein, Podolsky, and Rosen said, if you want to have a local theory, you need something to transport the information locally. The wave-function doesn’t do it, so you need something else. The many worlds interpretation doesn’t introduce anything new to get the job done, so of course it’s still non-local.
That said, I don’t mean to say that the Many Worlds Interpretation is wrong. If you replace the collapse postulate with suitable other assumptions about branching worlds, it gives the same results and the standard interpretation, so you can as well believe in it. Since we cannot observe the other worlds, not even in principle, we cannot ever prove that they exist. But for the same reason we also can’t prove that they don’t exist.
One final thing, because I’ve encountered this misunderstanding before. The Many Worlds interpretation has nothing to do with the path integral. Yes, you could think of the paths in the path integral as being in different worlds, but the point of the Many Worlds interpretation is that the different branches exist *after the measurement. In the path integral approach, they only exist in the period leading up to the measurement. So that the path integral approach works is not a justification for Many Worlds, they’re two different things.
In summary, the Many Worlds interpretation is neither wrong nor unscientific, but it’s exactly as problematic as standard quantum mechanics. Whether you believe that all those parallel universes exist is up to you. We can neither confirm them nor rule them out.