The biggest problem in the Many Worlds theory of quantum mechanics

[jasonbraun]

One thing that has always bothered me about many worlds and this explanation of probability as well is the following:
According to many worlds, every possible event does in fact happen but only the probabilities of experiencing certain branches differ, right?
If you look at a person's life then and map out a complete tree diagram of every single time they split into several different branches does that not imply that there is a version of that person that actually exists that has experienced all of the most unlikely events in their life?
If we consider this world for a moment, would their empirical research and therefore some of their understanding of the world not be completely different from ours?
Also, I'm not a physicist so I apologize if I misunderstood something but is entropy not a result of probability? Does that mean there is a world where energy reorganizes itself or at least works very differently?

[KenMathis]

Once you start assigning probabilities to being in one of two worlds, the many worlds theory loses its elegance. What determines that probability, and what does that probability even mean since another version of you experienced the other outcome? If there is only one world for each outcome with one of you in each, then it would be impossible to for the probabilities to differ. Every time the experiment is run, all outcomes would happen exactly once, thus making them all equally likely.

The only way to maintain the many worlds simplicity is to say a large number of worlds are created for each event, and the proportion of the worlds created for each outcome equals the likelihood of that outcome occurring. So if two outcomes could happen, with the probability 'A' being 1/3 and the probability 'B' being 2/3s, you'd have at least three worlds created. One world with event 'A' and two worlds with event 'B'. But that too gets extremely messy when the probability of an event happening is an irrational number that can't be represented like this. An infinite number of worlds would need to be created to get the proportions right.

[TheViolaBuddy]

I feel like this answer was kind of unsatisfying, probably because it doesn't really answer what "our experience" means. We sit here with a 2/3 blue, 1/3 red particle in front of us, and we know that if we do the measurement, we will become two separate people in two non-interacting worlds. But, like, "our experience" travels into one of the worlds, with a 2/3 probability in the blue way and 1/3 probability in the red way. And that's what you kind of tried to define with probability at the end, but that definition still didn't define what it means for "our experience" to travel down one path and not the other; it kind of just assumes that we know what "our experience" is. And intuitively we do, of course, but it feels very handwavy and non-rigorous.

I don't know if there is an answer to this, though. At some point it does go down the "what is consciousness?" issue that easily veers into the pseudoscientific philosophical talk that often surrounds quantum mechanics in pop culture. Maybe the next video about how to mathematically interpret "probability" will help, not sure.

[LookingGlassUniverse]

One thing I should clarify: the definition of probability I propose here certainly isn't rigorous! That's probably why philosopher's prefer defining the probability in more defensible ways using decision theory. But I was looking for an understanding of probability that felt like probability even if it has flaws

[lolroflmaoization]

Also there is actually a lot of problems with reducing quantum probabilities into self locating probabilities, you can look up the video on youtube "David Albert - "Worries About Accounts of Probability in Everettian Understandings of QM", once again this is a video by a philosopher of physics, and i still think the challenges he gives there are really unasnwered to the point when combined with Tim Maudlin ontological worries, really makes the Many Worlds interpretation as of now, not really a valid theory until these issues are answered.

[erikb]

With branch counting, what stops there from being an uncountable number of worlds in each branch? In other words, the fraction of branched worlds with each outcome is equal to the probability.

From another perspective, there are an uncountable number of events happening in each instant across the universe. Discretely counting the branches doesn't make sense.

[HuragokSlayer]

For the second example, when we're looking for A or X and then C or B if X, shouldn't it be 1/3rd for A and 2/3rds for X, since X includes two options? Just because you're only testing for A or X initially doesn't give them equal weight in the probabilities because X is actually multiple things. Probabilities are so weird.

[boogerie]

Many worlds is the most flagrant violation of Ockham's razor imaginable!

[danny]

I guess I don't understand the premise that an interaction with 2 outcomes only creates two worlds. if A has probability 1/3 and B has probability 2/3 why can't B just have two branches connecting to it assuming each branch as an equally likely chance to be "picked". My point being that there could only be 2 outcomes, but many different routes to get to those outcomes. Feyman diagrams come to mind.

[mad_vegan]

Many comments here suggesting to split into 3 branches, where each has 1/3 probability. This doesn't work, because the words "world" and "branch" here actually refer to a precisely defined mathematical object -- a vector |v> in a Hilbert space. These objects by definition have to obey the linearity rule: a|v> + b|v> = (a+b)|v>. If we split into 3 branches where 2 of them represent the same measurement result with probability 2/3, then |a+b|² = 2/3. But |a+b|² > |a|² + |b|² for a, b ≠ 0. So the branches can't all have equal weights.

[georgerevell]

You've made a mistake sorry, you have the branch with X and A each with probablity 1/2, thats not correct, X has a probablity of 2/3 and A of 1/3, so this mistake is what lead to the calculations being incorrect. You say "our rule tells us they should both be a half" what? No, not all measurements must be 50, 50 outomes... when meausuring position for example, it takes on a continuum of values. I just don't understand why you assign equal probablity to both A and x just because we are not sure which it will be, thats not correct at all, you can't just assume that of two possible measurement results the odds are always 50/50, thats defintely not true.

[Acrt]

So you define the probabilities with the word "expected proportion". But isn't this self-referential? Can you define what "expected" means without using the concept of probability of the outcomes in the first place?

[TheWyrdSmythe]

What the MWI does to probability has, for me, always been a huge strike against it. The general notion that, contrary to our experience that rare events happen rarely, under the MWI _all_ events, no matter how rare, _always_ happen seems a strange and wasteful way to run a universe. I’ll have to see what your next video says about the Born rule, but here you’ve demonstrated nicely why branch counting doesn’t work. By branch counting, after three trials, your probability is 1/8 of seeing all three red when it should be (1/3)^3=1/27.

[lolroflmaoization]

There is another huge problem with the Many Worlds interpretation, i recommend you read the Chapter 4 "Can the World be Only Wavefunction? " in the book "Many Worlds? Everett, Quantum Theory, and Reality".

In that chapter Tim maudlina argues that the wave function alone is insufficient to account for the result of any measurement. To do so, says Tim Maudlin in the chapter "Can the World Be Only Wavefunction?", one must add particles, i.e., localized objects in low-dimensional spacetime, into the ontology. Maudlin's conclusion is that Everett's interpretation, and similarly collapse alternatives in which nothing but the wave function exists, are epistemically incoherent: they do not make the connection between theory and the results of experiments comprehensible, and yet these results are presumably what serve to confirm these theories to begin with.

The worry here seems to be that if, according to the Everettians, the wave function is all there is, and if, further, it 'lives' in an abstract, multidimensional space, then it is unclear how such an object can account for our experience which is, roughly put, the behavior of localized objects in the low-dimensional spacetime we inhabit. Bohmians can easily address this problem, says Maudlin, because they simply postulate such localized objects by adding them into the ontology. GRWf theory (collapse with flash ontology) has a similar solution. But Everettians (and first generation collapse theoreticians with them) face the serious challenge of coming up with a comprehensible link between the state of wave function (which is all there is) and what warrants our belief in the theory, namely, the behavior of localized objects in a low-dimensional spacetime, which is our experience. Decoherence, argues Maudlin convincingly, simply cannot meet this challenge.

These kind of worries are unfortunately not appreciated by most physicists because they don't engage with the philosophy of physics, i recommend you read/watch Tim Maudlins work, he brings a lot of clarity on what a quantum theory has to have in order to be a satisfactory account of what the world is really like.

One good lecture i recommend you listening to on youtube is " Tim Maudlin - The Metaphysics of Quantum Mechanics"

[brothermine]

In David Deutsch's variant of MWI there are an infinite number of worlds that have the same state prior to the measurement, and in this model the probability of each outcome is the fraction of the infinite worlds that have the outcome. In Deutsch's model, the disproof that begins at 2:36 doesn't hold. In particular, the probability of measurement X where the electron energy is greater than energy A is 2/3, not 1/2, and the total number of worlds is infinity, not 2.

[geekworthy]

You're A vs. X is incorrect. A is 1/3rd since X = B or C. For example if it is known that people live to 100 years old and every age group is equally represented, then choosing a 1 year vs. any other age group is not 50% for a year old and 50% for all ages over 1. That is just bad math.

[a.hardin]

This experiment she does in the final part of the video would end up disconfirming the born rule in most but not all branches. So if one believes in MWI, then one must believe that one has ended up in a very rare world where the born rule is confirmed by simple 50/50 splitting.

[lunafoxfire]

So one thing that bothers me with the many-worlds interpretation and this branch-counting idea is that branching is usually presented as binary decisions, which makes sense for something like spin. But what about for something like position? Surely when a particle's wave function collapses to a particular point in space from some interaction, there is a world for every possible value that its position could have, right? Which is presumably an uncountable infinity of branches created from a single collapse event (or a very large number if you think space is discrete).

I'd love to know if you have any thoughts on this, though.

[hermestris]

💡Idea :
In the Red / Blue ball case
Because the probability distribution is 1/3 + 2/3, wouldn't it make more sense to split into 3 different worlds? 🤔

I mean, branching into 2 worlds where the ball is blue and 1 where the ball is red?

[BIasphemer]

the day superposition was explained to me was the day I first thought "oh, I live in a computer simulation" and I haven't shook it since