Hacker News new | past | comments | ask | show | jobs | submit login
The Many-Worlds Interpretation Has Many Problems (quantamagazine.org)
78 points by ykm 8 months ago | hide | past | web | favorite | 70 comments

>What can you do with this power to generate worlds and selves? You could become a billionaire by playing quantum Russian roulette.

Quantum suicide isn't exploitable, for reasons I went over in a thread on reddit:

>One thing about these quantum immortality (or modal realism, dust theory, etc.) arguments is they imply one should expect to end up in those proportion of worlds that have the highest measure in which "you" exist. And the most probable actions you should expect to remember are the highest-measure actions that leads to the perpetuation of your conscious experience.

>In this way quantum suicide is not really exploitable, as the vast majority of "your" measure will be in worlds where you find quantum suicide to be unappealing. Even failed attempts will be of much smaller measure then those in which you never try. So stories like Permutation City are sort of the equivalent of stories in which magic is the result of quantum fluctuations. They don't violate any laws of physics but they focus on such a ridiculously myopic slice of possibility space that it is a little absurd, even when you condition on the continuation of conscious experience.

Even if it were exploitable, this is not a reason to doubt many worlds, any more than time dilation is a reason to doubt relativity.

This great comment on that piece by Sengachi is a must read: http://disq.us/p/1wnl71q

I think there is a still more straightforward and robust argument against quantum immortality: The same reasoning predicts you can never fall asleep.

If futures where you are not conscious are inaccessible by the anthropic principle, then dreamless sleep cannot be in your future.

Most sleep is dreamless. At 4:00am this morning I will almost certainly be in this state. But being unconscious is not an experience I can have, so by the reasoning of quantum immortality, I should expect with certainty to find myself awake at 4:00am if there are any Everett universes where this is the case. And of course the same argument applies to every other moment of my entire life, and yet somehow I haven't been trapped in a state of wakefulness since the day I was born...

(If sleep is breaking the rules, the substitute it in with "temporary death" if you prefer. Why I am I allowed to "fork into" the branch where I die as long as I wake up later? How does the information about my eventual revival make it to the branching point, earlier in time? Or does the revived version of me exist with measure zero? These seem plain absurdities to me).

Think carefully about why many-worlds doesn't prevent you from sleeping. I think this leads to the correct answer.

I think of it in terms of a phase space diagram. You are conscious in some parts of phase space. Those regions of phase space have certain measures. You should expect to find yourself in parts of phase space with large measures, and not in the parts with small measures. Like this: [1]

In the diagram, increasing purple is "increasing probability consciousness." You should expect to find yourself in the purple area, not 10,000 years off the top of the diagram, deep in a white region. Quantum immortality supposes the latter, and that should be a self-evidently bad hypothesis.

[1] https://i.imgur.com/ETOqKs9.png

I don't think that's accurate. The many worlds interpretation is fundamentally an interpretation of something unobservable; if there were a way for you to observe it (by noticing you haven't been able to sleep) and 'report back' to all/most universes, it wouldn't be an interpretation anymore. (If you did a quantum suicide experiment, you might be able to 'report back' to the subset of universes where you still exist, like in Permutation City. I'm not sure that would actually be a valid demonstration of many worlds – the classical equivalent is just "something unlikely happened", and you can justify that with the anthropic principle. But in any case, that's very different from reporting back to all universes.)

As for the diagram… Even in a classical universe (or just a non-many-worlds one), if you set a mechanism to have a 50% chance of killing you, you should expect to 'find yourself' alive afterwards – in the sense that if you 'find yourself' anything, that anything can only be alive. It's just that you have an equal chance of not having any further conscious experience. That doesn't apply to sleep or even temporary death, since the resulting universe would still be observable by you at some point in the future. Now, even in a classical universe, there is a possible set of values under which participating in a 'suicide lottery' could be rational! That would be, "I fundamentally don't care whether I exist or not, but I do prefer existing while being rich to existing while being poor". The many-worlds interpretation just lets you amend that to: "I don't care whether I exist or not in 'my' future, as long as someone else identical to me exists in an alternate universe." There is no reason you must or should care about someone benefitting in an inaccessible universe just because they're (very) similar to you; but there is also no reason you must not. To whose benefit you strive is an arbitrary choice, after all.

Quantum immortality says something concretely different than what you're describing; it says that you always have a subjectively high probability of continued existence. By quantum immortality's reasoning, that is not the case under classical mechanics or non-Everettian quantum mechanics, where there are many situations where you have a conscious future with subjective probability of effectively zero.

Of course I disagree with quantum immortality though I agree with Everett, and I think you are kind of restating my point: Finding outselves in the purple region is not surprising and is what we should expect.

But quantum immortality as I believe it is usually stated does not require you to be ambivalent about non-existence like you say. It promises future existence, which I think is invalid.

If something is unobservable, and has no observable properties - in fact cannot have observable properties, by definition - it's metaphysics, not science.

One day when we're able to use quantum technology to clone ourselves, we'll be able to test the theory. You split your consciousness into 1000 clones of yourself and kill 999 of them, do "you" always seem to end up in the one that survived?


I don't see how this refutes anything.

You cannot experience being asleep, but what you experience is waking up, having been asleep. I find this consistent with the fact that approximately 100% of the time I experience is me being awake.

Having been unconscious is also an experience with p>0. It's only having been dead where (as far as we know) p=0. Not temporary dead: dead, dead; since it's defined as having been conscious (even if on and off) and then never being conscious again.

It's similar to being born: you don't experience anything before, but it's p=1 for you being born if you're experiencing it.

It's not refuting anything, only changing the axioms.

I agree. I think quantum immortality is real, but even if I was 100% confident, I don't see a clear way to take advantage of it beyond having a long-term outlook.

If you did want to build a suicide machine, it better be perfect. If you are trying to win the lottery with say 1:10^9 odds, your machine better have less than a 1:10^18 chance of seriously injuring you but not killing you. If you are capable of building such a machine, you can become rich anyway.

On top of that, it's quite selfish - if anyone out there loves you and knows you are doing this, they should hate you.

This so much!

The fact is, we're all already exploiting it all the time. By existing in the reality where you decide not to try quantum immortality, you open your future life up to tons of possibilities. The moment you pick up the gun to kill yourself, you already are on a path limited to "puts the gun down" "shoots self, injured but survives" and "dead". Once pulling the trigger, quantum immortality will 'save' you, with a few worlds spawned where you survive just fine, but the overwhelming chance is you'll wake up in a hospital bed with whatever's left of you.

I actually know of someone who committed suicide to explore quantum immortality. They aren't hated for it since really, you don't commit suicide because of an interesting theory, you commit suicide because you're depressed. An interesting theory may be given as the reason, but it was depression and people are really sad the person is gone. (As am I)

We are already exploiting it by existing.

I mean, the probability of a human with genes and experiences like me is astronomically small. Even the probability of existence of humanity (and nonexistence of aliens that would wipe us out) is quite small.

But the multiverse is large, and here we are, and here I am.

If people knew that suicide attempts would surely fail in typically horrible ways, at least from their own perspective, I’m pretty sure it would put a good dent in the epidemic.

We need a Uber for Guillotines !

Assuming Many Worlds to be true, wouldn't exploiting it be inevitable? Decisions and choice would work somewhat differently under that model. Any decision that could possibly be made would be made. Should you try using quantum suicide to win a game of Russian Roulette? This question loses its meaning, because you do make the attempt in some universes (both successfully and unsuccessfully), in others you don't try it.

I agree the potential for exploitation doesn't present a counter argument though.

> In this way quantum suicide is not really exploitable, as the vast majority of "your" measure will be in worlds where you find quantum suicide to be unappealing.

Unappealing is subjective, though. Sure, the vast majority of "your" measure will be in worlds where quantum suicide will kill you. But there is no reason you can't think of "I will be dead but some version of me will be alive in another universe" as an appealing scenario.

As a thought experiment, suppose that instead of involving quantum mechanics, you were an emulated mind capable of making an exact copy of yourself in this universe. And, for some reason, someone is willing to give you money in exchange for you copying yourself and then immediately deleting one of the copies at random. In a lot of ways this looks similar to the quantum lottery scenario: "you" will split into two copies, one will end up dead, and the other will end up richer.

One difference is that in the quantum scenario, you're leaving behind a whole universe without you in it, whereas in the emulation scenario, the universe will end up with one copy of you either way. But if you don't care about the universe, only yourself, then that wouldn't matter to you. Another difference is the alternative if you choose not to perform the experiment: two copies of yourself (or a self with twice as large a measure) in the quantum scenario, one in the emulation scenario. But then, as an emulation, you could just decide to copy yourself and not delete one of the copies. Normally, that would be pretty different, since the two copies of you would have to interact with each other, perhaps compete for the same resources, while two quantum copies exist fully independently of each other. But you could imagine sending one of the copies off at lightspeed to some distant place, with no intent to ever communicate with the copy again – heck, you could theoretically send the copy beyond the cosmic event horizon so that it's physically impossible to ever communicate again. There wouldn't be anything wrong with doing that – but if you decided "nah, I don't feel like it; one of me is enough for this universe", you can't say that would be irrational or suicidal. So, in the quantum scenario, why should you prefer having two of you (by not doing the experiment) over having one (by doing it)?

I'm not sure why you're being downvoted so heavily. You're contributing to the discussion.

It's also possible that every version plays out physically but your spooky consciousness only picks one of them. Quantum suicide involves a little philosophy, it's not a direct implication of MWI.

Quantum suicide is a fascinating thought experiment. If consciousness is a property of a certain type of brain (or calculation?), then the concept of ‘’’me” is just a variation of the anthropic principle: I believe my subjective experience to be something unique, but there’s nothing unique or special about a human brain and the conscious experiences it produces.

So even if there are billions of worlds, there aren’t billions of “me”s out there. There are simply billions of consciousnesses which share common memories and thought patterns. Each variation would believe itself to be the true “me”, when in fact they are all wrong: there is no “me”. Just the consciousnesses that all those brains create.

So in quantum suicide, there is no transfer of any sort. One consciousness is simply extinguished forever. The other continues.

This aligns somewhat with Buddhist teachings about the illusion of “self”. I don’t believe in the supernatural, but one could consider the idea of reincarnation as merely the fact that there are not instances of consciousness, but instead manifestations of one phenomenon. Which could possibly mean that if you are a conscious being reading this comment at this moment, you can rest assured that your death will not stop “you” from continuing to experience subjective experiences as another person, in another place, at another time. You just won’t realize it, because you won’t be “you” anymore.

Or maybe it’s all nonsense.

I have a hunch there is a variation of quantum suicide that would be less lethal, but no less measurable.

>One consciousness is simply extinguished forever. The other continues.

If there is one instance of a widely-distributed program installed on a particular computer, and the instance of the program on that particular computer is deleted, but the program continues to exist in identical form on many other computers, then the program continues to exist in spite of the fact that one of its instances has been deleted.

If consciousness is thought of as a process inextricably tied to a particular physical substrate, then our own consciousnesses are constantly being extinguished, and we all die countless 'deaths' every day as the matter in our brains is recycled and replaced.

If, on the other hand, consciousness is viewed as an information process that can be realised on any suitable physical substrate, then extinguishing one instance of consciousness in one particular universe by means of quantum suicide doesn't necessarily mean that that particular consciousness is extinguished forever, as long as an identical instance, or instances, of it continue to exist somewhere in the wavefunction. The Many Worlds Interpretation suggests there may be an infinity of such instances.

There is a tendency to think of these instances as copies, and to think of each such copy of a consciousness as a separate doppelganger existing in some unreachable reality. But if the instances are truly identical in every respect, then Leibniz' principle of the identity of indiscernibles [1] suggests that it might not be right to think of them as being separate: in some deep sense, they're all actually the same single thing.

The philosopher Derek Parfit developed a now-famous thought experiment in the phiolosophy of identity [2] in which he encouraged people to think about a scenario in which a malfunctioning teleporter creates two identical copies of a person, one of which must then be destroyed by the machine. Obviously we tend to sympathise with the 'copy' faced with imminent destruction, and that might lead us to reject the idea that the continuation of a physically identical copy of our self is a sufficient condition for survival.

There is a problem with this thought experiment though, in my view, and quantum suicide is in a sense a type of 'fix' for it. If there is a situation in which two copies of a person exist, and one copy is about to be destroyed, then the two copies are necessarily no longer 'indiscernible' in Leibniz' sense. Parfit's example is actually a striking case of indiscernibility, because at the point at which one of the copies is about to be destroyed, the two copies actually have some very different memories and thoughts. This version of Parfit's thought experiment therefore really involves two similar, but not identical, people, one of whom is about to be killed.

The quantum suicide thought experiment, in contrast, necessarily ensures that no conscious entity ever experiences its own imminent death in the same manner as Parfit's version, because the 'suicide' is designed to take place faster than the brain can process a thought. The other major difference is of course that the 'copies' that survive the suicide are not created by a teleporter, but are a necessary feature of the wavefunction.

An alternative to saying that one consciousness is extinguished forever in quantum suicide would be to say that the single consciousness existing across the 'copies' in the wavefunction no longer has conscious access to events in the universes in which the suicide occurs, but survives in, and continues to have conscious access to, the universes where the suicide does not take place.

It only makes sense to think of one consciousness as having been 'extinguished' if that consciousness is thought of as being inextricably linked to one particular physical substrate. If consciousness is thought of as an information processing pattern that can be instantiated on any suitable substrate, then all that matters is that such a substrate exists somewhere. If the substrate is physically possible, then the Many Worlds Interpretation requires it to exist.

[1] https://en.wikipedia.org/wiki/Identity_of_indiscernibles

[2] https://en.wikipedia.org/wiki/Teletransportation_paradox

For any of this to have meaning, you need to define what "consciousness" even is. I'm not aware of any definition that is not hopelessly metaphysical.

The continued experience that you exist. You literally have never known anything else, only the representations which consciousness allows.

The stuff itself is hopelessly metaphysical, the only good way I know to define is the above: since you have are consciousness, its easy enough to just say 'look around'. Its difficult to think of any material definition when there's not much of an understanding of how it relates to anything material. Obviously the images you see are dictated by physical rods and cones in the eyes, but what the hell are colors?

> The continued experience that you exist. You literally have never known anything else, only the representations which consciousness allows.

Right, but that's what makes it so meaningless. We generally define things by specifying how they're different from other things. But you can't do it with consciousness, because you can't really know what the absence of consciousness is like. And if you can't do that, the definition of consciousness becomes null - and nothing meaningful can be derived from it.

With colors, at least, you can define different ones in terms of different sensations - red is not-green etc. And because they're derived from observation of phenomena, you can compare notes with other people to make sure that what you consider red is also what they consider red etc - it's not a perfect system, but it's "good enough". I don't think we have anything even remotely resembling such a "good enough" definition of consciousness, though.

Great points.

I think what you and I are spending a lot of words working toward is a simple question: does subjective experience have one subject or many?

Philip Ball explores the problems that many world interpretation poses to the concepts of self and meaningfulness of life in the sense that he understands them.

Nothing in the article clearly points to problems in physics or even philosophy of physics in many world interpretation.

Makes me feel uncomfortable is not very deep argument.

Agreed - the idea that selfhood and duality are illusions of perception have been around at least as long as Buddhism, and philosophically well articulated. They are hardly grounds for dismissing a theory of the fundamental universe.

Larry Niven had a nice short story on that, called "All The Myriad Ways". The Wikipedia article on the short story collection of the same name [1] includes a link to the story itself in their collection of old Galaxy issues.


Here's the article in short.

1. "Roland Omnès says the idea that every little quantum “measurement” spawns a world “gives an undue importance to the little differences generated by quantum events, as if each of them were vital to the universe.” This, he says, is contrary to what we generally learn from physics: that most of the fine details make no difference at all to what happens at larger scales."

2. "What can it mean to say that splittings generate copies of me? In what sense are those other copies “me?”" / "Consciousness relies on experience, and experience is not an instantaneous property: It takes time, not least because the brain’s neurons themselves take a few milliseconds to fire. You can’t “locate” consciousness in a universe that is frantically splitting countless times every nanosecond"

3. "You could become a billionaire by playing quantum Russian roulette. Your quantum splitter is activated while you sleep, and if the dial says Up then you’re given a billion dollars when you wake. If it shows Down then you are put to death painlessly in your sleep." / "But a committed Everettian should have no hesitation about doing so using the quantum splitter. For you can be certain, in this view, that you’ll wake up to be presented with the cash. Of course, only one of “you” wakes up at all; the others have been killed. But those other yous knew nothing of their demise. Sure, you might worry about the grief afflicted on family and friends in those other worlds. But that aside, the rational choice is to play the game."

4. "Alice is put to sleep before the measurement is made, knowing she will be wheeled into one of two identical rooms depending on the outcome. Both rooms contain a chest. Inside one is twice her stake, while the other is empty." / "But what if Alice were to say, “The experience I will have is that I will wake up in a room containing a chest that has a 100-percent chance of being empty”? The Everettian must accept this statement as a true and rational belief too, for the initial “I” here must apply to both Alices in the future." / "In other words, Alice Before can’t use quantum mechanics to predict what will happen to her in a way that can be articulated — because there is no logical way to talk about “her” at any moment except the conscious present (which, in a frantically splitting universe, doesn’t exist)."

Sengachi's reply in the comments below basically summarises my reaction.

It occurs to me that a lot of these arguments revolve around the author's preferred definition of consciousness, which is essentially arbitrary. There's no physical phenomenon that is consciousness - it's a term that we came up with for the subjective feeling that we experience as we function. You don't really know if someone is conscious or not unless you ask, and you don't really know if the answer is true or not anyway (because they might be conscious but unable to communicate, or because it might be an unconscious automaton that's programmed to answer that it is conscious) - so there's no way to define it experimentally.

I can't help but think that "consciousness" is really just a modern atheistic version of the "soul" concept, nothing more. Basically, we just gave up on immortality, and accepted that there are some physical processes underneath that cause it to start and stop - but we're still clinging to that self-perception of ourselves as a singular entity that drives self, as something that meaningfully exists in the physical world. But it is recursive - we think that we're conscious as a function of conscious, basically. You can't really distinguish consciousness from unconsciousness per se, because when you're unconscious, you do not observe or reason. In others, you can only observe the lack of prerequisites of consciousness (e.g. brain activity), or the lack of its side effects (communication etc), but you can't observe unconsciousness as such.

From that perspective, the idea of consciousness has no more place in science (or criticism of science) than soul does.

People ridicule the nonlocality of Pilot Wave theory. But I’d rather question locality than embrace Everett’s many-worlds theory. If you need to postulate such amazing complexity just to say basically nothing in terms of realism, then you’ve already lost.

Meanwhile as one good physics professor pointed out, none of these alternatives to PWT even rule out nonlocality. So why not just assume it and be done with the weirdness?

>If you need to postulate such amazing complexity just to say basically nothing in terms of realism, then you’ve already lost.

Defenders of the Many-Worlds Interpretation, and some other types of multiverse theory such as Eternal Inflation, would actually say that the multiverse idea is the simpler alternative, not the more complex one.

Multiverse theories produce an abundance of universes, but the underlying description of those universes is typically simpler than alternatives that suppose a single, unique universe. Concepts such as wave function collapse or hidden variables, which actually add complexity to the theory, have to be introduced in order to get the theory to produce a unique observed reality.

One could have argued with Aristarchus or Hubble that the idea that there were other solar systems, or other galaxies, added more complexity than was necessary to explain reality. From a broader perspective though, our reality would actually be stranger and more difficult to explain if the properties we observed of our solar system, or our galaxy, were the only ones that could exist. Ditto our universe, say the Everettians.

Kepler desperately struggled to explain the properties of our solar system as the product of a unique and divine geometry before he discovered the more general truth [1], which allowed the existence of the infinity of alternative configurations which are now known to exist around other stars.

[1] https://en.wikipedia.org/wiki/Mysterium_Cosmographicum

This is only true if you ignore the fact that there's no explanation, in detail, of how quantum splitting generates an entire universe - never mind why it happens for observed events only, and not for unobserved quantum events. (Is that how it works, or not?)

The photon bounced off a beam splitter - did that generate a new universe? How about if the photon moved one Planck distance through spacetime? The photon hit a detector? Okay - a universe. Yes? Or do you only get a new universe when the result is observed? By someone who's awake? And not on drugs, so they might be hallucinating?

The argument seems to be based on "Obviously it just does, because how else do you explain this weird math, QED, isn't our explanation great?" - which I personally don't find very convincing.

> If you need to postulate such amazing complexity just to say basically nothing in terms of realism, then you’ve already lost.

What about many worlds is complex? The Schrodinger equation predicts macroscopic decoherence. Macroscopic decoherence implies many worlds. Many worlds can hardly be called an interpretation. Rather, it is a prediction. It is just taking the Schrodinger equation literally. Every other "interpretation" is more complex, as they bolt things onto the Schrodinger equation, and often very ugly things, too.

Macroscopic decoherence is about the interaction of a quantum system with its environment, resulting in a mixture state for the subsystem of interest. But if you describe the whole universe as a quantum system there is no environment, no decoherence, no many worlds, the universe will remain in a pure state if it was originally in a pure state.

The evidence against Pilot Waves keeps stacking up. https://www.quantamagazine.org/famous-experiment-dooms-pilot...

That experiment really has nothing to do with actual pilot wave theories. It just rules out a fluid-mechanical analogue of quantum mechanics (which anyone could predict anyway since fluid mechanics isn't non-local in the same way).

The experiments described in that article have zero impact on Bohm's pilot wave theory, which is the only current pilot wave theory that is actually workable.

Could the anthropic principle taken into conjunction with our continued existence be used to show that the MWI is more probable?

Here's the situation that I'm thinking about: it's well known that we've come close to nuclear annihilation multiple times, either by accident or by tensions between the US and the USSR in the Cold War. If the MWI is correct, it is probable that a good chunk of these worlds have been reduced to ash. However, we currently find ourselves in a world where that has not happened, presumably due to the anthropic principle. Could we then give a numerical probability that MWI is correct, based on this fact?

I'm not a philosopher or physicist - just an interested layman.

I don't think that's how the anthropic principle works, but I'm not actually sure how the anthropic principle works.

To my understanding, the anthropic principle is basically: among the near-infinite set of possible physical laws, we can immediately rule out the ones that would prevent us from being around trying to figure out which physical laws govern the universe.

I think there's two varieties of this. The "weak anthropic principle" says, that doesn't mean those laws couldn't exist, there's no weird metaphysics around it, it's just that if the underlying physical laws didn't allow us to exist, you'd just have a universe that didn't have anyone in it and so the physics of that universe would be a moot point that didn't matter to anyone.

The "strong anthropic principle" says, wait, no, there is some metaphysical significance to this, and it's somehow not even logically possible for a non-anthropic universe to exist. I find that a little self-absorbed on the part of humanity as a whole, but that would tie in a little better to your quantum suicide idea. The weak anthropic principle, in contrast, would just kind of say, "yeah, there's a chance we all blow ourselves up, and if we do, I guess nobody will be around to care about physics anymore".

Even the strong anthropic principle doesn't prove much here. "Nuclear annihilation" probably wouldn't actually result in complete human extinction. Some people would survive. Individual quantum immortality doesn't stop you from shooting a hole through your brain and being neurologically disabled for the rest of your life, so collective quantum immortality also wouldn't stop you from living in a post-apocalyptic hellhole. Also, even the strong anthropic principle doesn't actually mean that humanity can't go extinct; after all, it didn't prevent the universe from existing for billions of years before humans started to exist. Maybe the real strong anthropic principle doesn't even apply to us, it applies to a species of avian hive-minded critters on a planet circling an unnamed star in the Andromeda galaxy, and as a weird side effect of living in a universe that is metaphysically required to permit for the existence of the Andromeda flying hivemind, humanity kind of shows up for a few hundred millennia and gets to do some physics before getting wiped out by a swarm of paperclip maximizers.

If the MWI is true, then there must be some universes that would experience multiple highly improbable events. These events would be a hint to residents of that universe that the MWI is correct. Why wouldn't the usual Bayesian analysis apply here?

Your personal probability of observing a universe that experiences multiple highly improbable events is the same whether or not MWI is true.

Imagine yourself as a resident of that "universe" (I prefer to think of it as "branch").

Why should a sequence of unlikely events be a distinguisher between different interpretations of a theory, which make the same probabilistic predictions as each other?

No, because you can come up for very good arguments a priori for why two or more rational nuclear powers leads to a very stable world, because every party is very interested in preventing nuclear war if they are materialists.

Are they good arguments though?

Some would say that we observe through declassified information after the fact, that the "rational nuclear powers" were not themselves in control, and a series of accidents at lower levels could have, but didn't, cause disaster, mainly through luck.

Similarly, there was a recent thread that had information about how the only reason the first moon landing succeeded was because of a series of accidents, where a computer bug compensated for another unintended defect in the lander. History could have been vastly different via small pivotal events.

Why, then, don't we find ourselves in universes where the atom bomb was never invented?

MWI seems more elegant than Copenhagen in many formulations because the problem of the relative magnitudes of the worlds is swept under the rug. If I perform a quantum coin-flip with 2/3 possibility of heads, then it creates two worlds - but I end up in one of them twice as often - why? The only solutions I've seen to preserve the relative probabilities are: 1) sufficiently unlikely worlds sometimes just disappear, or 2) some kind of appeal to consciousness. These are at least as ad-hoc and hand-wavy as waveform collapse.

> If I perform a quantum coin-flip with 2/3 possibility of heads, then it creates two worlds - but I end up in one of them twice as often - why?

This question is impossible to answer until you unambiguously define "I" and "end up" first.

My answer is that there's no singular you. There's one you for each of the worlds, just as there's one coin for each of the worlds. The only difference is that the coin doesn't wonder about its "other selves", because it doesn't have the concept of self. Humans do have one, but we have no reason to believe that it is correct or meaningful in this context.

Or it actually creates three worlds, two heads worlds and one tails.

Being able to count worlds in the manner you describe is apparently not easy to justify. If I recall correctly, this is actually a pretty thorny epistemological problem with the theory. IIRC the current response among Many Worlds people is that either (1) it doesn't really matter because there is no reason to believe that probabilities are "real" in the first place, or (2) attempting to show that if you did insist on finding a metric that looked like probability, it would provably have to be the Born probability (see https://arxiv.org/abs/1405.7577). All this stuff is way over my head so I have no way of evaluating whether it's persuasive or not, but it does at least seem to be contentious among physicists.

Or an infinity of worlds, with two thirds of them with heads, and one third with tails.

I enjoyed reading this article. But at the beginning the author has this to say about objections to the MWI that are based on feeling:

> "This objection is rightly dismissed by saying that an affront to one’s sense of propriety is no grounds for rejecting a theory. Who are we to say how the world should behave?"

But really at the end of the article the conclusion about the problems with MWI feel similar:

> "But an idea that, when we pursue it seriously, makes that view inchoate and unspeakable doesn’t fulfill the function of science."

Of course it seems weird to think of Many Worlds as somehow manufacturing universes for each fluctuation. That seems weird and wrong ... because it probably is.

A better way to think of it is as the equivalent of the relativistic block universe. All these different spaces already exist in some superspace, and ‘random’ events take you from one space to a neighboring one. Nothing is manufactured.

You can also think of it (at least I think of it) as a domain-specific subset of modal realism: https://en.wikipedia.org/wiki/Modal_realism

From a certain point of view, these kinds of metaphysical ideas are pointless in the sense that it's impossible to actually determine whether they're true or false, and even if you could ask God and God told you, the answer wouldn't matter. It's to some extent a completely meaningless distinction that doesn't describe anything that's directly useful to us. On the other hand, as an illustrative model, it can be an intuitive way to understand nondeterminism. If it isn't intuitive to you and some other model is more intuitive, use that instead. Until the differences between these models become somehow falsifiable, it doesn't matter anyway.

Well put; this is more or less what Everett actually proposed, and it's the interpretation that mainstream Everettians like Sean Carroll and Max Tegmark subscribe to.

Yes, before reading Tegmark I thought that MW was silly because "splitting" the whole universe for every tiny event just seemed crazy. Thinking of the multiverse as more like a probability distribution where everything exists in a kind of overlay seemed to make sense.

Unless I'm missing it, his argument about why MWI is untenable in terms of physics is:

- If the universe splits with each measurement (or interaction), then every possible outcome does occur

- Therefore nothing could be predicted based on probabilities, and that proves MWI is wrong

This seems to have a few logical errors:

Just because everything that can happen does, in MWI, happen in at least one universe, that doesn't mean that the probability of each outcome is identical.

Probabilities still apply: higher probability outcomes mean that one is more likely to find one's self in a universe in which that outcome takes place.

In this way of describing things, it's still a very low (but non-zero!) probability that all the air in your room will move to the corner, leaving you to suffocate in a vacuum. But the way we state it would be, "it's a very low probability that you will find yourself in one of the universes in which this happens".

Philip Ball has previously written very similar articles on his objections to Many Worlds:


Whilst I share his frustration with how readily it is being taken up and the dogmatic glamour surrounding it, I feel like he consistently misses the primary issues with it.

Copy-pasting a previous comment trying to describe succinctly what I see as the central problem:

> Although there are many versions of MWI, the general claim is that a single axiom (a unitarily evolving wavefunction) is sufficient to explain our observations of the universe. Without getting into what it means to "explain" something, it seems fair to demand that a sufficiently intelligent agent with no prior knowledge of our physics should be able to predict what the theory says about their future observations. However no one would be able to make any experimental predictions based on the above axiom. There's nothing there to suggest that if the agent becomes entangled with a quantum system then they would only experience a single branch of the entanglement. Even if you add that in, there's nothing to suggest which branch the agent will experience (this is roughly the preferred basis problem). Even if you add that in, there's nothing to suggest with what probability the agent will experience that branch.

> You could always add in the above as additional axioms. You could postulate some physical content to the Born-rule, or to something weaker from which to deduce the Born rule. But this would break the illusion that MWI requires fewer axioms. Instead, a great deal of energy has been expended into deducing the Born rule purely from that single axiom. There have been many very clever attempts at this, but none convincing enough to settle the matter.

If you're interested in reading more, Adrian Kent has written some fairly thorough critiques on arxiv:


> there's nothing to suggest which branch the agent will experience

isn't this accurate modelling though? as far as we can tell the collapse picks a branch at random. I just don't see how this is a problem of the model.

My description there was pretty crude, apologies. The measurement problem is often loosely expressed in terms of metaphors like "live cat vs dead cat". But if you actually look at the quantum state of such a complicated system, it's this huge tangled mess of Hilbert dimensions lumped together into one big global wave function. The "alive vs dead" separation is just one possible perspective (technically it's a slicing of the Hilbert space into mutually orthogonal sub-spaces). There are infinitely many possible such ways of splitting the space up, i.e. infinitely many possible basis sets. It's not directly clear why, when we physically interact with the system, one specific basis is chosen for us to "collapse" into one branch of.

Perhaps it could be better explained as "there's nothing to suggest which branching the agent will experience splitting into".

I should note that the process of decoherence goes some way towards removing this obstacle (although not the following question of explanation of probabilities).

The Many-Worlds Interpretation is seriously mischaracterised in the summary at the top of the article (presumably not written by Ball himself) when it says this:

"The idea that the universe splits into multiple realities with every measurement has become an increasingly popular proposed solution to the mysteries of quantum mechanics."

This is a misrepresentation of what the Many-Worlds Interpretation actually suggests, and it is contradicted in the article itself by Ball:

"In effect, this implies that the entire universe is described by a gigantic wave function that contains within it all possible realities. This “universal wave function,” as Everett called it in his thesis, begins as a combination, or superposition, of all possible states of its constituent particles. As it evolves, some of these superpositions break down, making certain realities distinct and isolated from one another. In this sense, worlds are not exactly “created” by measurements; they are just separated. This is why we shouldn’t, strictly speaking, talk of the “splitting” of worlds (even though Everett did), as though two have been produced from one. Rather, we should speak of the unraveling of two realities that were previously just possible futures of a single reality."

The error in the summary is a common misrepresentation, and it leaves many people (including me, when I first heard of the Everett Interpretation) with the mistaken impression that the interpretation proposes some mysterious new mechanism that causes the entire universe to 'split' every time a physical interaction or observation occurs. This is pretty much the opposite of what the interpretation actually suggests, since its purpose is to dispense with the significance of observation and 'wavefunction collapse' in the Copenhagen Interpretation.

The physicist Max Tegmark (an adherent of the Everett Interpretation, who made a film[1] about Everett with Everett's son; a sad story, as Everett was unable to make a career for himself in physics due to the unpopularity of his most famous idea) I think puts it best when he says this about the Many-Worlds Interpretation in his book, Our Mathematical Universe:

"The rumours I'd heard suggested that Everett proposed crazy-sounding stuff like parallel universes and that our Universe would split into parallel universes whenever you made an observation. Indeed, even today, many of my physics colleagues still think that this is what Everett assumed. Reading Everett's book taught me a lesson not only in physics but also in sociology: I learned the importance of going back and checking the source material for yourself rather than relying on secondhand information. It's not only in politics that people get misquoted, misinterpreted and misrepresented, and Everett's Ph.D. thesis is a great example of something that, to first approximation, every-one in physics has an opinion about and almost nobody has read.[2]"

"The notion that at certain magic instances, reality undergoes some sort of metaphysical split into two branches that subsequently never interact isn't only a misrepresentation of Everett's thesis, but also inconsistent with Everett's postulate that the wavefunction never collapses, since the subsequent developments could in principle make the branches interfere with each other."

[1] https://en.wikipedia.org/wiki/Parallel_Worlds,_Parallel_Live...

[2] https://www-tc.pbs.org/wgbh/nova/manyworlds/pdf/dissertation... (Everett's original thesis)

Thankyou for clearing this up. I've been confused about this for a long time. It also explains why the decoherence theory fits in with this story.

My own problem with many worlds is (a) it seems to me to require a ridiculous amount of energy and (b) what happens when two world splits occur "at the same time" on opposite sides of our planet? Also (c), it just doesn't appeal to intuition, but it seems like much of physics does not :)

One thing that others me about the MWI: Is the so-called "set of all possibilities" a well-defined set?

Yes - the universe is still described by a single, pure wavefunction. I think this kind of confusion is inevitable when people talk about different "universes" or "worlds" being created, when nothing of the sort is occurring in the theory.

This feels a bit like JIT processing.

There may be a number of states possible but there is no need to 'process' which one it is until that information is needed.

In some universes it has no problems.

This is a very philosophical / meta-physical discussion of quantum mechanics, and not in line with our modern understanding of the measurement process. For example, there really is no sudden wave function collapse in quantum mechanics, as every measurement process can be described as a purely quantum-mechanical process that produces an apparent "collapse" of a wave function through entanglement and decoherence. There's nothing magical or abrupt about it, today we can even perform quantum measurements that precisely control the amount of decoherence that is introduced to a system (see e.g. some of Irfan Siddiqi's seminal quantum feedback experements like https://www.nature.com/articles/nature11505, or the earlier work done in Serge Haroche's group). Taking aside the philosophical problems that people have with "pure" quantum mechanics (i.e. an interpretation that does away with the second postulate of wave function collapse) it perfectly describes our experimental results.

The largest problem with wave function collapse (which the author only briefly skims) is that you need a way to explain how it works: Basically when the wavefunction collapses we go from a fully reversible, deterministic quantum system to an irreversible, stochastic system. At which scale is this supposed to happen? Let's say one day we can build huge quantum computers (think billions or trillions of qubits) that we can perfectly control. We could then prepare a single qubit in a superposition state. We can entangle this qubit with the other qubits (the measurement system) and manipulate the state of the measurement system using a deterministic but highly chaotic control program. If we'd then measure the state of the single qubit (e.g. by preparing many identical systems and performing quantum state tomography) we'd find that the wave function has collapsed, i.e. there is no more coherence between the two qubit states. Now, after entangling the qubit and evolving the large system we could just reverse its deterministic evolution (as we have perfect control over it) and bring it back to the state it was in directly after entangling it with the qubit. If we then perform quantum state tomography of the qubit we should see that the coherence is back. Now, if we believe that wave function collapse is a phenomenon that occurs independently and not as an effect of the evolution of the large quantum system, we would expect to observe a loss of coherence in the qubit even after perfectly reversing the state of the large quantum system. The question is then of course: At which scale does this happen, and what is the physical theory that governs this behavior? Are one billion qubits enough to produce it? One quadrillion? To my knowledge, no one came up with even an idea of how to describe this. If someone finds a good theory for describing wave function collapse I'll consider it as valid theory, until then I'll stick with the "many worlds" interpretation, though I'd really prefer calling it "plain" quantum mechanics instead.

> This, he says, is contrary to what we generally learn from physics: that most of the fine details make no difference at all to what happens at larger scales."

Oh? Since when have we disproved the butterfly effect?

> Since when have we disproved the butterfly effect?

There never was such a thing, not the way you are using it.

The butterfly effect is an observation in chaos theory that small changes in initial conditions can make large changes in output.

The "popular" usage of butterfly effect "A butterfly flapping its wings can eventually cause a hurricane" is not a thing that ever existed.

The difference is that one is directly fed into the output, the other is not, the other is filtered by huge amounts of noise.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact