Theoretical physicist and science communicator Sabine Hossenfelder has made several videos regarding the Many Worlds Interpretation (MWI) of Quantum Mechanics. And while these videos are interesting, I believe Hossenfelder makes significant mistakes regarding MWI (or science in general) at several points, and this post is a critique of her view.

Note that I won't give an introduction to either MWI, the alternative Copenhagen Interpretation (CI) or Quantum Mechanics in general, so some background knowledge is assumed.

Alright, so let's get straight to the points I wish to critique. In her video Does the Many Worlds Interpretation make sense? Hossenfelder makes the following claim:

The major challenge for many worlds is to explain why the thing we call an observer does not itself branch with those worlds, and therefore sees all the outcomes, but somehow randomly only experiences one of those worlds. I've never found a good explanation for that.

Not only is this not a major challenge for MWI, it's not even a challenge at all, as the observer does in fact branch with those worlds. Earlier in the video, Hossenfelder uses the example of a beam splitter, which causes a photon to take two different paths in a superposition of possibilities:

Both paths have a 50% probability associated with them. Of course, once an observer — call him Bob — measures a photon to be on one of these two paths, Bob knows it's there with 100% probability. In CI, this corresponds to the collapse of the wave function, and there's one Bob who sees one outcome. In MWI, both paths are taken in parallel universes— and crucially, each of these universes contains a Bob. For simplicity's sake, let's say there's two universes: one where the photon takes the path to the right, and one where the photon goes downwards. Then each of these two universes contains a Bob, and if these Bobs observe where the photon is, they observe the photon to be in different paths. So Bob does "branch with those worlds", and does see "all the outcomes": there's multiple Bobs who, taken together, see all the outcomes.

Regarding this experiment, Hossenfelder says:

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.

(No, it doesn't, but more on that later.)

This update of the wave function is also sometimes called the collapse or reduction of the wavefunction, and it's a key element of quantum mechanics. If you don't update the wavefunction, you'll get wrong probabilities.

(No, you don't, but more on that later.)

Hossenfelder later uses this point to criticize support of MWI:

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.

No! The probabilities are given by the Born rule. In MWI, they are the same before and after the observation happens, and describe the proportion of universes that have each particular outcome. There's absolutely no need for any collapse here. Uncertainty is only in the mind of any specific observer!

To see what I mean, consider the photon experiment again. From an MWI perspective, we can look at this as follows: from the start of the experiment — before any measurement/observation takes place — there's two parallel universes (again, this is a simplification). In one of them, after hitting the beam splitter, the photon travels to the right; in the other, the photon goes downwards. Both parallel universes contain Bob. Both Bobs measure where the photon is. Bob 1 observes the photon is on the right, Bob 2 observes the photon travelled downwards.

This illustrates that the probabilities don't change after the Bobs do their measurement: there were two parallel universes before, and there are still two afterwards. What changes is that, before the measurement(s), Bob 1 and Bob 2 were the same person: Bob. Bob didn't know where the photon was, and Bob 1 and Bob 2 do know. Bob believed there was a 50% probability that the photon would take the right path, whereas Bob 1 believes this probability is 100%. That is a change in probability, but only in Bob and Bob 1's knowledge of the situation: it changed because Bob 1 has learned in which parallel universe he lives. The actual probabilities — that is, the proportions of parallel universes with particular outcomes — remain the same.

So MWI really doesn't need any collapse postulate. CI does, and that's what makes CI more complex (in terms of Kolmogorov Complexity) than MWI. Superposition happens when the Schrödinger Equation has multiple solutions — in which case it simply describes multiple (parallel) universes. CI tries to predict only one universe by adding the collapse postulate, but unless this is backed up by evidence, it's an unnecessary complication. Occam's Razor favors MWI!

Then there's Hossenfelder's video The Multiverse: Science, Religion or Pseudoscience? In this one, she makes the following claim (note that she talks about several different multiverses, including the one of MWI):

The issue with all those different multiverses is that they postulate the existence of something you can't observe, which is those other universes. Not only can you not see them, you can't interact with them in any way. They are entirely disconnected from ours. There is no possible observation that you could make to infer their presence, not even in principle. For this reason, postulating that the other universes exist is unnecessary to explain what we do observe, and therefore something that a scientist shouldn't do. Making an unnecessary assumption is logically equivalent to postulating to existence of an unobservable god, or a flying spaghetti monster, or an omniscient dwarf who lives in your wardrobe.

So Hossenfelder is implying (and later explicitly stating) that multiverses aren't scientific. But the above quote contains a major flaw: MWI doesn't postulate the existence of parallel universes; it's the Schrödinger Equation that predicts them, and the Schrödinger Equation most certainly makes other predictions that are testable (and have been tested). As we saw, MWI throws away a postulate (the collapse postulate). And — in my view — MWI does explain what we do observe better than CI, as MWI is deterministic and doesn't imply non-locality.

They believe these other universes exist because they show up in their mathematics. You see, they have mathematics, and some of that describes what we observe. And then they claim therefore everything else that their mathematics describes must also exist. They are confusing mathematics with reality.

It's not entirely clear to me whether this quote is also about the Schrödinger Equation predicting parallel universes, but I'll address it anyway. Science builds theories, and if the best theory we have to explain some observed phenomenon also predicts something we haven't observed, we should take that prediction seriously. It doesn't mean that thing we haven't observed — such as a parallel universe — must exist, but it does mean it's made more likely. I realize that's a bit of a vague statement, but Hossenfelder claims science says nothing about the existence of parallel universes, when it literally does say something about them if a theory predicts them.

Hossenfelder later comes back to her earlier point of not being able to observe the other universes:

Second objection I hear is that we can only observe a patch of our own universe because light needs time to travel, and it's got only so far since the Big Bang. But certainly no one would say that therefore the universe stops existing outside of the part we can observe. No, of course not. No one says if you can't observe it, it doesn't exist. The point is: if you can't observe it, science says nothing about whether it exists or not.

So… Science says nothing about whether there's more than the observable universe?! That's what this quote directly implies, though I'm having a hard time believing that's what Hossenfelder actually believes. Anyway, the bigger point is that this quote misrepresents science in general. Obviously, observation plays a key role in science, but it serves to potentially falsify existing theories, not as a direct arbiter for whether it's scientific to say something exists or not. That is — to repeat my earlier point — something isn't unscientific just because we can't in principle observe it.

Alright, that concludes this post. Thanks for reading!