# Improbability of the Many-Worlds Interpretation?

#### PeterDonis

Mentor
to determine what happens in the lab we only need the boundary conditions defined on the lab.
And if the MWI is true, those boundary conditions are already going to include a huge number of terms in a vast superposition, based on what happened in the entire past light cone up to that point and all of the possible quantum branch points that occurred. So there doesn't need to be any "world splitting" in the lab itself to have multiple worlds according to the MWI; multiple worlds are already included in the lab's boundary conditions. For example, if you are going to claim that there is a world in which @DrChinese is the US president, that doesn't mean we had to do something in the lab to switch timelines to that one.

But the question is, which multiple worlds are already included in the lab's boundary conditions? Is there a world included in those boundary conditions in which @DrChinese is the US president? Supposing that there is some world in which @DrChinese is the US president, how many other things are different in that world as compared to this one? Is there even a lab in the same place at the same time? And if there isn't, it makes no sense to restrict attention to "the boundary conditions defined on the lab" but also talk about the possibility of @DrChinese being president, because those two things are mutually exclusive: if we are in a world in which a lab at the same place at the same time exists, then @DrChinese is not US president in that world, because none of the worlds in which @DrChinese is president have a lab at that time and place.

And in fact, even that doesn't fully unpack the issue, because to even have a world in which @DrChinese is the US president, it has to be that there is a US with a president, and a person who is properly identified as @DrChinese, and that person has to have gone through all the process of getting to the point where he is elected US president. And you can't tell whether such a world is even possible by just looking at the boundary conditions of your lab. And we can go further: there has to be a planet Earth that had humans evolve on it, with a history the same as our Earth; which means there has to be a solar system the same as ours, of the same age, in the same galaxy, etc., etc.

The point I'm making is that all this blithe talk about "multiple worlds" fails to pay attention to specifically how such multiple worlds get created, if the MWI is true. They don't get created by magic. They don't get created just because we humans can imagine them. They get created by having genuine quantum mechanical uncertainty, "Schrodinger's Cat" type uncertainty where single quantum branch points have macroscopic consequences, involved in particular key events in particular timelines. And the only point in the entire universe's development where we are pretty sure that actually happened was at the end of inflation, when the quantum fluctuations in the inflaton field got transferred to the Standard Model fields and the particular pattern of small-scale inhomgeneities was formed that were to eventually evolve into our present galaxies and stars and planets. Everything after that is classical evolution and need only have led to a single macroscopic world. And if other universes exist in MWI "worlds" that correspond to other possible patterns of small-scale inhomogeneities created at the end of inflation, they won't even have a solar system like ours, or a planet Earth like ours, or humans like us, or a US with a president--let alone have all that stuff be the same but still have just enough difference to have @DrChinese as the US president in 2019.

Last edited:

They get created by having genuine quantum mechanical uncertainty, "Schrodinger's Cat" type uncertainty where single quantum branch points have macroscopic consequences, involved in particular key events in particular timelines. And the only point in the entire universe's development where we are pretty sure that actually happened was at the end of inflation, when the quantum fluctuations in the inflaton field got transferred to the Standard Model fields and the particular pattern of small-scale inhomgeneities was formed that were to eventually evolve into our present galaxies and stars and planets. Everything after that is classical evolution and need only have led to a single macroscopic world.
It's not clear to me that Michael Price is saying Dr. Chinese is president right now. For me that doesn't make sense, since he clearly isn't president based on the information contained in this lab. I'm not even sure you can compare times between different worlds as they are casually disconnected from the point of branching.

But I also think you can't just consider classical evolution since inflation. When people perform quantum measurements now, it forms new branches. This is most clearly true when people use quantum outcomes to make macroscopic decisions. It's not clear if you are suggesting an alternative view, where quantum outcomes are overwhelmed by classical evolution and the branch we are living in is actually deterministic.

I also think quantum branching and decoherence is ubiquitous, though it's more controversial. This is based on the only analysis I've seen on the topic, which I linked in the other linked thread, looking at how the average coin flip is an amplification of quantum uncertainty.

#### PeterDonis

Mentor
When people perform quantum measurements now, it forms new branches.
Yes, but how often does that happen? And how often does the outcome of such an experiment affect who gets elected President?

This is most clearly true when people use quantum outcomes to make macroscopic decisions.
And how often does that happen?

It's not clear if you are suggesting an alternative view, where quantum outcomes are overwhelmed by classical evolution
No, I'm not. I'm just saying that "Schrodinger's Cat" type situations, where the outcome of a single quantum measurement leads to macroscopically different states of a macroscopic object like a cat, are rare.

I also think quantum branching and decoherence is ubiquitous, though it's more controversial
Yes, more controversial indeed.

#### bhobba

Mentor
Yes, but how often does that happen?
At a deeper level what exactly, defined entirely in terms of QM because everything is QM, is a quantum measurement? A lot of progress has been made in solving that, but issues still remain, It may even be incomplete giving Einstein the last laugh. People often talk about issues with QM, but the above is not often mentioned.

Thanks
Bill

Yes, but how often does that happen? And how often does the outcome of such an experiment affect who gets elected President?
Thanks! Your position is clear and comes down to a technical point which hasn’t been deeply explored beyond that paper I mentioned. Indeed MW supporters like Sean Carroll casually assume branching happens all the time without really showing the details.

But I have done stuff like use an online service tied to a quantum random number generator to create a 1GB program and run it. So some other version of me should be capable of doing anything that a compressed and focused AI can possibly explain.

Maybe it doesn’t happen all the time, but even the occasional event is enough to make a big difference.

#### DarMM

Gold Member
So some other version of me should be capable of doing anything that a compressed and focused AI can possibly explain
Can you explain this a bit more?

Can you explain this a bit more?
Well if I manage to create a branch for every possible program of size X, in this example 1GB, some of those programs should contain AI that can do any arbitrary thing, like make Dr Chinese president or explain QM completely satisfactorily. My point is about how essentially MW does allow everything to happen, even *if* it needs to be invoked explicitly.

#### Buzz Bloom

Gold Member
This thread discussion has reminded me of a variation of the MWI a coworker explained to me about 15 years or so ago. I am wondering it anyone at PF knows of any publication that discusses this variation. I tried to find one searching the Internet, but I had no luck.

The MWI variation is that the multiple universes are not real, but instead are contingent. This mean that each of the various possible combinations of all of the possible future measurements that could possibly happen, according the the wave function of everything, define one possible contingent universe. When a measurement occurs, the single real universe becomes constrained to have it's wave function of everything reconfigured to have new initial conditions consistant with the new measurement. The collection of all the contingent universes are also reduced to just those which are consistant with the new measurement.

#### DarMM

Gold Member
Well if I manage to create a branch for every possible program of size X, in this example 1GB, some of those programs should contain AI that can do any arbitrary thing, like make Dr Chinese president or explain QM completely satisfactorily. My point is about how essentially MW does allow everything to happen, even *if* it needs to be invoked explicitly.
By AI here do you mean strong AI that genuinely figures out how to make @DrChinese president and tells you how to go about it? How do we know such a thing exists in the 1GB program space? Even assuming the human brain works like a "program" in any way, we know it exceeds a petabyte in storage?

One odd one is that technically there is some subset of the state space of the gunpowder where due to entropy decrease the gun doesn't fire for an incredibly rare entropic decrease. So if you get somebody to fire a machine gun straight at you there's a world where every bullet failed to fire.

This is beginning to sound like a Greg Egan novel, pretty cool thread. I think it breaks the record for @DrChinese mentions in one thread.

By AI here do you mean strong AI that genuinely figures out how to make @DrChinese president and tells you how to go about it? How do we know such a thing exists in the 1GB program space? Even assuming the human brain works like a "program" in any way, we know it exceeds a petabyte in storage?
I don't mean a fully general AI, just the most perfect program to accomplish one task. It doesn't even really need to figure it out - it just knows and explains the steps to take. 1GB is arbitrary but it's about the size of the human genome. Surely it's enough for all the content needed: the perfect ads, bots, private bitcoin keys to bribe and passwords / encryption keys to hack voting machines.

#### PeterDonis

Mentor
One odd one is that technically there is some subset of the state space of the gunpowder where due to entropy decrease the gun doesn't fire for an incredibly rare entropic decrease. So if you get somebody to fire a machine gun straight at you there's a world where every bullet failed to fire.
You're assuming that there is quantum uncertainty involved in gunpowder firing. That's an example of the kind of assumption I am objecting to.

#### DarMM

Gold Member
"Everything" and "rules" meaning what exactly?

I also wouldn't connect a clarification request from me about what exactly you were referring to to be related in any way to what physicists think about or don't think about. I just wasn't sure precisely what you meant.

#### DarMM

Gold Member
You're assuming that there is quantum uncertainty involved in gunpowder firing. That's an example of the kind of assumption I am objecting to.
Doesn't tunneling permit exploration of the chemical state space even if it's very small? My (very rough) intuition is that quantum effects in chemistry are suppressed when you reach arsenate molecules.

#### PeterDonis

Mentor
Doesn't tunneling permit exploration of the chemical state space even if it's very small?
The state space of what? A single gunpowder molecule? (Well, I know it isn't made up of molecules of a single chemical species, but you get the idea.) You can't fire a bullet with a single gunpowder molecule. You need something like an Avogadro's number of them. Yes, quite possibly some miniscule fraction of that Avodagro's number of molecules will be in a "can't fire" state due to quantum tunnelling, but so what? Enough will be in that state to fire the bullet. Or if the bullet doesn't fire it will be for classical reasons, like you left the ammunition unused too long and exceeded its shelf life, or the firing pin didn't work. It won't be because quantum uncertainty put the Avogadro's number of gunpowder molecules into a superposition of "can fire" and "can't fire", which is what would need to happen for quantum uncertainty to cause a splitting of worlds according to the MWI when the firing was attempted.

#### DarMM

Gold Member
Yes, quite possibly some miniscule fraction of that Avodagro's number of molecules will be in a "can't fire" state due to quantum tunnelling, but so what?
Because then isn't there a chance they'll all be in the "can't fire" state?

#### PeterDonis

Mentor
Because then isn't there a chance they'll all be in the "can't fire" state?
No. Or at least, if you're going to allow for that possibility, then you are allowing for possibilities that invalidate our belief in past data, and that version of the MWI undermines itself, as I argued in post #24. The chance that enough of the molecules will happen to all have tunneled into the "can't fire" state, and none of them will have tunneled back into the "can fire" state, at the exact instant that you try to fire the bullet, is comparable to the chance that @DrChinese actually was president of the US 5 minutes ago, but then a quantum fluctuation happened that changed everything on Earth, including all of our memories, to the state we all perceive and remember now. And if you allow for possibilities like that, all bets are off and there's no point in doing science at all.

#### DarMM

Gold Member
What prevents it though if you permit any arbitrary tunnling outcome to occur? What principle would stop me from allowing it in the space of outcomes?

#### PeterDonis

Mentor
What prevents it though if you permit any arbitrary tunnling outcome to occur? What principle would stop me from allowing it in the space of outcomes?
If you take the MWI entirely literally, nothing does. And I'm saying that doing that makes the MWI undermine itself, since it tells us we can't trust our memories and records of past data, because we could be in one of the worlds where a quantum fluctuation changed them. And what forces us to adopt QM in the first place is our memories and records of past data. If we can't trust those, what reason do we have to even believe QM, let alone the particular MWI interpretation of it?

#### DarMM

Gold Member
I'll need to chew on that, very interesting.

What do you make of the fact that in "single world" QM there is still a small chance that such an event could occur. Is it basically that such an event essentially has never occurred so we can ignore it, but that in MWI you know that there are definitely "science nullifying" worlds.

I suppose there would also be worlds where the Born rule was never validated due to unfortunate fluctuations causing it to be rejected.

#### PeterDonis

Mentor
What do you make of the fact that in "single world" QM there is still a small chance that such an event could occur.
That depends on whether you think sufficiently small probabilities--something like 1 expected event per $10^{70}$ years, say, where our universe has only existed for $10^{10}$ years--are actually meaningful. I personally think probabilities that small are just a sign that the theory is incomplete, just as the presence of singularities in certain GR solutions is a sign that GR is incomplete.

#### PeterDonis

Mentor
I suppose there would also be worlds where the Born rule was never validated due to unfortunate fluctuations causing it to be rejected.
According to the MWI, yes, there should be worlds (lots of them) in which the Born rule does not hold. I have seen papers claiming to show that the set of such worlds is a "set of measure zero", or something similar, but I'm personally not convinced by such arguments.

#### DarMM

Gold Member
That depends on whether you think sufficiently small probabilities--something like 1 expected event per $10^{70}$ years, say, where our universe has only existed for $10^{10}$ years--are actually meaningful. I personally think probabilities that small are just a sign that the theory is incomplete, just as the presence of singularities in certain GR solutions is a sign that GR is incomplete.
You might be interested that Roland Omnès has similar views to be seen on pages 82-84 and 235-236 of his "Understanding Quantum Mechanics". Ultimately these are related to earlier views due to Émile Borel in:
"Le jeur, la chance et les théories scientifiques modernes" (1941)

#### Klystron

Gold Member
Information theory expands the latter question to:
"How information describes everything?"
Otherwise, are we not limited to rule-based descriptions?

Do physicists (scientists) think everything can be known?

Self-referential form:
Does Information inform everything knowable?

#### Hans de Vries

Gold Member
And if the MWI is true, those boundary conditions are already going to include a huge number of terms in a vast superposition, based on what happened in the entire past light cone up to that point and all of the possible quantum branch points that occurred.
I should try not to get involved with quantum philosophical discussions, but okay, having done this sort of "hypothetical MWI" calculations in the past:

- There are ~$10^{80}$ elementary particles in the universe.
- Interactions / State transitions take place at the femto second level in stars.
- The Universe is circa $4*10^{32}$ femto seconds old.
- Each event may actually span much more than 2 branches, say $N$ branches.

So very, very roughly there may be $N^{10^{112}}$ different universes right now.
The total number of particles (and thus energy) of the multiverse increases by a factor of $N^{10^{95}}$ per second.

Each of the $10^{80}*N^{10^{112}}$ particles in the multiverse somehow knows exactly to which of the $N^{10^{112}}$ universes it belongs and only interacts with the particles labeled to be in the same universe. Each particle would need some universe label to distinguish between other universes. The number of bits you would need for one such label for one particle would be far more then the total information content of an entire universe.

The whole universe must split near instantaneously over the entire size of the universe according to some absolute reference frame to which time is synchronized, otherwise the order of splitting and the universe labeling would mess up completely. The average time between two splits is $10^{-95}$ seconds during which the split has to propagate circa $5*10^{10}$ light years.

So the minimum speed at which the splitting propagates must be something in the order of $10^{114}$ times faster as the speed of light.

Yeah really, MWI makes the task of understanding the physics of nature so much easier........

Last edited:

#### PeroK

Homework Helper
Gold Member
2018 Award
I should try not to get involved with quantum philosophical discussions, but okay....
Perhaps you should read this first:

It might save you from tilting at straw men.

"Improbability of the Many-Worlds Interpretation?"

### Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving