Problems with Many Worlds Interpretation

Ken G

It sounds like there are two "me"s here, an "F" me and a "B" me. Will the real me please stand up! When did the "B tree of me" begin, anyway?

Fredrik

This isn't a property of the CI or any other interpretation of QM. It's a property of all theories of physics.

A bunch of statements is a theory only if it makes predictions about results of measurements. A measurement is an interaction with a measuring device, specifically an interaction of the type that will leave the measuring device in a state that indicates a number called "the result" of the measurement. The indicator component is always perceived as classical by a human. (A component that isn't would be of no use as an indicator).

So measuring devices, or at least their indicator components, will always be described in "intuitive" terms, no matter what theory we're dealing with. This means that there's a "cut" in every theory that doesn't describe the rest of reality in equally intuitive terms. I would say that there's a "cut" even in those theories, because measuring devices should always be thought of as essentially independent of the theory. Consider e.g. using a cesium clock to test the accuracy of predictions of classical special relativity. The theory doesn't describe the inner workings of the measuring device, but no one would say that this means that you're not allowed to use it.

Ken G

Exactly, the only way MWI achieves completeness is by defining anything it doesn't do as "not important." But CI is just as free to play that same game-- CI simply asserts that having nonunitariness in the actual reality is not important! Which is isn't-- it is not important to take the postulates seriously, any more than it is important to take Newton's postulates seriously. Physicists still routinely use Newton's postulates, long after relativity and quantum mechanics, in exactly the same way they did 200 years ago when then took them seriously as an ontological description of reality. So the lesson is clear: don't take your physics postulates seriously as an ontological description of reality, it is both unnecessary and dubious! That's why I have no issue with MWI as a permitted interpretation of quantum mechanics, it is using it to fabricate a world view that I take issue with-- the pretense of knowing what is not known. CI is more clear that it is not a world view, and it makes no claims on completeness other than that it is a kind of folly to seek it in the first place.
I appreciate your device of trying to create a specific example, but I don't understand how you differentiate what you are writing normally vs. in brackets. What do you see as the distinction between what is definite and what is indefinite? I would think that MWI would have no language to even make that distinction in any way. But I do understand what your device is attempting to accomplish-- as I said above, MWI has no difficulty at all with accounting for a model of "you". What it fails is a model of "me." For example, you cannot even have an experimenter, you have <experimenterH,experimenterT>. Try it again that way and it shows even more clearly that your problem is accounting for the experience of the individual! So your device works fine for any one who has a "birds-eye" view, and accounts for the perceptions that appear along the branches of the many worlds. That's what I meant by "a model of you." What it does not account for is the perceptions that I have when I do experiments (where by "I", I mean, of course, you!).

So in your scenario, if you really were there, how come you never actually hear someone speak the words "<heads,tails>", if that is what you claim is really happening? You always have to go find someone else to be the bird's eye view to describe what is happening to you, you never actually get to be the person with that birds-eye view. That's what MWI cannot account for, why there can never be an actual observer with a birds-eye view, if that is what is really happening.

That's why MWI is incomplete, it has no accounting for your own experience, it can only give you an accounting of everyone else's experience by adopting the rationalist dream that there is any such thing as a birds-eye view that no perceiving agent ever gets. Is that really the mission of physics, to explain everyone else's experience but never your own? If <heads,tails> is what really happens, why when you flip a coin do you perceive heads or tails? I'm talking about your actual experience here, not your way of rationalizing that experience.

Thus any claim that MWI is complete first must reject empiricism. Ergo, it is an interpretation that is complete for rationalists, and incomplete for empiricists. That's all I'm saying here-- MWI is not a philosophy that is a complete way of looking at physics, it is only a complete philosophy under the condition that one has already adopted rationalism and rejected empiricism. Only then does MWI seem to be complete, so it is very restricted form of completeness, and one with a poor track record.

PAllen

What about the proposition that there are many Hugh Everetts but only one Niels Bohr?

Hurkyl

MWI is certainly capable of talking about whether a subsystem is in a pure state or approximately in a pure state or not.

The big idea is that nobody has a birds-eye view. Therefore, any philosophical assertions about how the birds-eye have no scientific basis. I assert the analogy:

CI is to MWI as Lorentz Ether Theory is to Special Relativity

I usually don't get to talk about the thing I really want to do, since in these discussions I'm usually involved in an uphill struggle to get people to admit MWI is valid, or (occasionally to get people to admit that CI is valid). But maybe you'll appreciate it.

What I really want to do is to do the very thing my analogy above suggests: once we can entertain the notion that unitary evolution of wave-functions can adequately describe our experiences, we can then take the next step and notice there are many wave-functions^* that are empirically indistinguishable, and so we can switch between them at our leisure.

In particular, collapsing a wave-function when you observe something just becomes an example of changing your frame of reference.

*: There's another bit of indefiniteness going on here if you want to pay careful attention to the bird's eye view



How would I tell the difference? Not only do I not possess any empirical evidence that I don't hear the words "<heads,tails>", I have no idea what such evidence would look like.

I would perceive "heads or tails" because <heads,tails> is what really happened.

Fra

I've made this point before but I'll make it again...
There are (at least) TWO main ways to understand what a theory is:

a) as a description (of the system, the universe, or whatever...)
b) or as an interaction tool (for learning about "the system", the universe, or whatever...)

in (a) you test the theory by clean poppian style falsification. A theory that in retrospect fails to "DESCRIBE the future" is wrong - there is no "theory" for how to produce a new hypothesis.

in (b) you test the theory by seeing "how much new information" you get from the system by applying it (ie by asking optimally rational questions as per the theories EXPECTATION of the future). A theory seen as an interaction tools is only "wrong" if it FAILS to serve the questioner in navigating in "hypothesis space".

Think about which of the two views that makes most sense in scenarious where for example the notion of ensembles and infinite repeats of the questions just can't be realized.

I agree with this, if we're talking about all mainstream physical theories, but this observation does not make it right.

This is pretty much a dogma that is apparently VERY HARD for most physicists to release oneself from, but I think at some point we wil lbe force to. Note that (b) by no means weakens "science", it IMHO rather provides a deeper understanding of it. I personally think is one point that people like for example Popper either didn't understand or for some outlandish reason disagreed with.

One may wonder what this has to do with the discussion here, but I think it very much has, at a subtle deeper level, since this is much in line with the "solipsism at it's best", or at least my view of it.

/Fredrik

Ken G

The place where CI puts the cut is not between the QM system and the classical me. Remember, Bohr said there's no such thing as the QM system, so there's no need to place any cut there. Instead, the cut is simply between what I experience, and what I think about what I experience. That is the standard cut of empiricism, it has nothing to do with any particular physical theory, and there is no theory that cannot have a cut put there, including string theory. So string theory is perfectly compatible with CI.
And the correct response of the CI person is: wave functions don't "behave" at all! They evolve in the parameter time, in exactly the way QM says they do, this is no issue for any CI person, nor does it have the least bit of connection to adopting a many-worlds view of reality. The very connection between the letter "t" you put in your expression for a wave function, and your experience of time, is just another type of collapse, in the CI view, and the only reason you can even associate the two is because you experience it to work. The experience is the justification for everything-- that's CI, that's empiricism.

I agree completely-- the difference between CI and MWI is entirely the way the window is dressed. But that's the whole problem-- since when did window dressing justify a world view? This has been my point all along, not that CI is demonstrably a better way to think about reality than MWI, but rather that CI is simply more honest about our motivations. We want to believe we know, but we can still resist the lure of wanting to create a world view based on what we only want to believe we know to be true. Again, look at the history of physics, and look at how scientists generally characterize the desire to believe in the absence of evidence one way or the other.
All physics theories are ad hoc, they just invent whatever they need. Need an electron? Invent one! That's how physics is, it's ad hoc. But you are right that CI and MWI are conditioned by the questions we feel are important. CI is motivated by the question "what explains what I perceive." MWI is motivated by the question "how can I describe a system in terms different from what I perceive, yet in a mathematically unified way?" Take yer pick-- mathematical unity or description of experience, it's rationalism vs. empiricism.
Correct, you get it-- to an empiricist, all physical theories are syntax, whereas reality is experience. To a rationalist, all experience is some kind of illusion, and theory is truth. So has it been for thousands of years. But note that the reason science is usually grounded in empirical truth is that it is the only type that is demonstrable in an unambiguous way, and when experience disagrees with what seems reasonable, experience always wins. Even so, we cannot understand our experiences without theories, so we need both, and the tug-of-war goes on.You are talking about the epistemology. Scientists have no real trouble with epistemology-- we all know we need to combine those two ways of knowing to get anywhere. Nor does either MWI or CI have any disagreement around how to do the scientific method. The place where we find the real issue between MWI and CI is not their identical epistemologies, it is their different ontologies. When someone asks you, do you think many worlds really exist, that is a classic ontological question, and that is exactly where the debate should center. What is the evidence you will call forth to make the claim that you are not just believing something because you like to believe it? Is there really anything more that can be said about MWI as something other than just doing quantum mechanics?

Hurkyl

Eh? This needs clarification, because I've never seen that in any of my science texts, and I've never observed any difficulty including myself in a system under study.

Hurkyl

Yes. That's all "rationalism" and "empiricism" are -- epistemological positions.

Er, that's a behavior.

Except that CI asserts collapse.

I don't want to believe I know. I just want to learn and understand a physical theory, and how it connects to observation.

Right. A world view to be based on knowledge (be it rational, empirical, or other) is more likely to be accurate and useful than one based on desire or biases.

As far as I can tell, those that see past their biases tend to prefer to do things like preferring Special Relativity to the equivalent Lorentz Ether Theory -- e.g. not to insist on the correctness of a particular frame of reference.



Ad hoc is necessary, but bad. The more ad-hoc a theory is, the less its ability is to make precise predictions, which in turn diminishes the value of any evidence favoring the theory.


No. MWI is motivated by the question "Can a mathematically unified method explain what I perceive?"

Individual people may have other motivations. I'm not talking about those people. If there was ever a branch of MWI that had a different motivation, I'm not talking about that branch.


I think you mean some sort of Platonism, rather than rationalism.

Formally, at least, theories are syntax and truth is semantics, there's no if's, and's, or but's about it. And I'm enough of a formalist to believe that anyone who claims otherwise really just hasn't learned to mentally separate the ideas of "theory" versus "interpretation".

(I'm fine with a conventional usage of 'theory' to refer to a theory together with a particular interpretation of the theory -- as long as a person can admit they are doing that!)

You would only ever catch me saying "theory is truth" if I really meant something about deduction or provability, but got lazy when faced with the fact natural language is poorly equipped to talk about the subtler details of the topic.


I strongly disagree. It is very, very difficult to consider an empirical "truth" without having first filtered it through one's intellect. I would be so bold to claim it impossible to do so completely.

The unambiguity only (seemingly) appears between people that filter their experiences in similar ways.

Fredrik

You seem to have a very precise opinion about what the CI is. There are many different ideas that are all considered to be "the CI" by someone. The one that you call "the CI" is the one that I'm the least willing to call "the CI" myself, because it involves an essentially supernatural element that actually contradicts the theory it's supposed to "interpret". An interpretation is supposed to tell us, in language we can understand, what the theory is saying is actually happening to the system. It's not supposed to tell us how the theory is wrong.

The idea that QM doesn't describe reality and just assigns probabilities to verifiable statements, is one of the ideas that can be called "the CI". However, people tend to call it an "ensemble interpretation", or a "statistical interpretation" these days.

I think it's really hard to talk about "the CI" at all, because a lot of what we can read about it are silly misunderstandings of what Bohr actually said. For example, Bohr acknowledged the fact that we wouldn't consider something a "measurement" if we can't read a "result" off of an indicator part of the measuring device, and people have misunderstood this to be a claim that the laws of QM don't apply to measuring devices at all.

Fredrik

OK, I see that my statements look pretty weird in that context. Not sure if I can do much better though. One of the reasons is that I can't figure out why you think that string theory would be more compatible with MWI than CI. Since I don't understand that, I don't know what I should be saying in response.

Fredrik

I have some issues with the word "describe". I would choose to talk about making accurate predictions about results of experiments instead. More importantly, your (b) contradicts science as we know it. (That's not entirely clear from what you're saying here, but we have been talking about this before, so I know what you mean). I'm not saying that it definitely is a bad idea. I haven't ruled out that something similar to what you like to talk about can be developed into a useful generalization of the term "theory". However, I don't think it's necessary to mention this in every thread about quantum mechanics, which clearly is a theory in the sense of what I like to talk about: It assigns probabilities to verifiable statements.

I also wouldn't describe a speculative new idea as one of the two "main" things that the term "theory" can refer to.

Ken G

Not when the knowledge in question is about "what exists." This is the point of contact between epistemology and ontology, and it is exactly the place where CI and MWI differ. They just don't differ anywhere else. They certainly don't differ in the issue of how to know if a theory makes correct predictions or not, and they don't differ in the issue of whether or not observers and theorists need to confer on a joint agreement of what is a good theory. So they simply don't differ in any of the ways in which scientific knowledge is judged or obtained, except one: they differ on the issue of what constitutes true knowledge about what exists.
Certainly not, not to the empiricist. Your claim here is, categorically, that how a mathematical entity depends on a variable t, makes a one-to-one claim on how some system behaves with the experiential and measurable concept of time. That just isn't true. You are missing that we are already affixing an interpretation to a physical theory that the parameter t in the theory corresponds to the empirically measurable concept of time. But it is not a necessary part of any theory that we must fail to distinguish between an experimental measure and a parameter in the theory (note for example that t is not an operator of quantum mechanics). It is part of the judgement of the value of the theory to connect that variable t to the empirical observation of experienced time, but only in the same empirical context that all theories are judged. No one ever says "I criticize your interpretation that t is time on the basis that it just doesn't seem like time to me", no, any criticism of that interpretation must be based on empirical comparisons of how the parameter t acts under experimental conditions. But we already know that the t that appears in the Schroedinger equation will not correspond to any experiential version of time if t is small enough, so the entire idea that "t" in QM is a continuous reflection of real time is simply false.
MWI is just as ad hoc, but the ad hoc nature comes at a different level in the goals of the theory. CI adds an ad hoc physical postulate that some mysterious process, not covered by the theory, chooses which outcome we actually experience, MWI adds an ad hoc metaphysical postulate (as you did above) that our particular individual experience is a question that physics should not be interested in. CI sees that as cheating-- it's not surprising that greater unification can be achieved by allowing ourselves to cheat on what a theory should describe (rather than on how it should describe it, that was Einstein's objection about dice).
It can't be that, because it fails to do that. Also, note that CI is just as mathematically unified as MWI. It has to be, it's all the same mathematics. MWI is not content with mathematical unification, it wants ontological unification. And it can only accomplish it by dodging the question of why I perceive what I perceive. I still haven't seen your answer to that. Your scenario doesn't answer it, because we don't have an observer getting <heads,tails>, we have a bundle of perceptions connected with the sentient being I'll call observer1, and a bundle of perceptions (according to MWI) connected with sentient being observer2, whose perceptions do not overlap and they do not perceive each other. So what you'd have to say is, you have <observer1,observer2> reporting <heads,tails>. If you do that, the mathematical description is entirely unitary, but you have no way to account for why it did not come out <observer1,observer2><tails,heads>. All you could say is that you don't care about the difference there-- but try telling that to observer1 if "heads" means he loses his.
Platonism is quite a bit different from the way MWI is normally expressed. Indeed, I would have no problem with MWI as a form of Platonism, that is the sense in which MWI makes reasonable claims. The key difference is that Platonism draws a distinction between what is physically real and what is abstractly real. If people want to imagine that the many worlds are abstractly real, as some form of perfect concepts, I would have no issue with them other than being a bit idealistic. It is the claim on physical reality that I feel should require empirical demonstration. When someone can empirically demonstrate the existence of many worlds, only then would I find it appropriate for us to conclude that they are physically real.


I completely agree, that was the flavor of Godel's proof-- syntax and semantics can never be the same thing in any formal system rich enough to be suitable for our purposes. Indeed that is my entire issue with how MWI is normally expressed-- it improperly crosses that dividing line, mistaking a syntactic interpretation for a semantic one. That's also the place where ontology creeps in.
That's true, it's the bugbear of formal empiricism that brainless entities can't do it. All the same, it is pretty clear when a consistency of perception has been identified. That's why reading of pointers can be done by almost anyone, but predicting those readings can be done by rather few. The rationalistic perspective about what is true knowledge regarding what exists is quite elitist, and suffers the flaw that a much more intelligent species than we will likely form a completely different view, one that we simply cannot understand any better than most plumbers understand string theory. Yet the plumber knows what he experiences, so the empiricist version is a more accessible ideal about what constitutes truth. Empiricism also avoids the troubling "truth is only as good as your current theory" problem that dogs rationalism.

Still, I grant you that neither rationalism nor empiricism can make a self-contained case, and that's probably why we need a combination to do science. Perhaps we have more to learn from the tension between the CI and the MWI, than we have to learn by marrying one or the other. But I completely agree with Fredrik that you are mischaracterizing CI when you say that it adopts essentially supernatural claims about the ontology of collapse-- instead, CI adopts a solipsistic perspective on collapse, it merely accepts collapse as real on the basis that we experience it, and unitarity as unreal on the basis that we do not experience it. CI takes no other position on the matter, there is no sense that "a miracle happens" when collapse occurs-- instead, collapse is what happens, the two are exactly the same concept in every way. That's why I said that unitary evolution is not a behavior at all in CI, it is just the language of how to predict, statistically, the next behavior, the next collapse. There is nothing in the empirical meaning of our word "behavior" that is not also in the word "collapse," so there is no need to attribute any mystical or miraculous elements to the latter word.

Ken G

I suppose that depends on what one means by a subsystem. Many people, including on this thread earlier, talk about subsystems in MWI as associated with a particular particle that has some identifiable consistency across branches. In that language, no subsystem is ever in a pure state-- if I pass an electron through a Stern-Gerlach and accept all the ones taking the spin up path, none of them are in the state "spin up", they are all in a mixture of <spin up, accepted> and <spin down, rejected>. That's not a pure state, because there is no definite phase relationship between those terms, the phase is decohered. All the same, the experimenter doing quantum mechanics will deal only in the pure state <spin up, accepted> when he/she talks about the electrons in the accepted path, and MWI can deal with the fact that the projection onto that experimenter's acceptance decision will look like a pure state. But in MWI, that experimenter is under an illusion that the reality of the particle state is something different than what it actually is, because the experimenter's version cannot result from unitary evolution prior to the experiment, even as he/she uses unitary evolution to predict the subsequent behavior.
I take your point about nobody having a birds-eye view, so that the ontological description must in a sense be cobbled together from all the birds-eye views. However, I reach the opposite conclusion along that same logical path! I see the analogy as:

MWI is to CI as Lorentz Ether Theory is to Special Relativity

We agree that a consistent ontology needs to be patched together from objective observer's perspectives, but what goes into the patchwork? LET attempts to find a single truth that in some sense regulates all the observations, the aether, whose existence relies entirely on a rationalistic desire to unify the reality of all observers into a single description of "what is actually happening even though we never see it." That's a lot like what MWI does, it takes an ontological stance on what is happening in a way that we never actually see. But SR, like CI, rejects any ontology that involves entities we never see. SR elevates to the level of a metaphysical principle that if nature conspires to keep some ontological entity hidden, then it must be part of our interpretation that such an entity does not exist.

On the other hand, the role of decoherence as seen in MWI sounds an awful lot like how Lorentz imagined that the aether must be monkeying with our experimental apparatus to fool different observers into seeing different things, all within a single unified mathematical description. In other words, Lorentz interpreted his contractions ontologically, whereas Einstein made it a core value that if there is no empirical imperative for the contraction, then the contraction was not ontological at all, it was merely relative to the observer. In short, the set of observers that can share notes with each other, and their observations, is what defines the reality, reality is not tricking them or leading them into illusions. Yet that's just what it is doing in the MWI, so to me, the MWI sounds a lot more like LET in its stress on a consistent ontology instead of just querying a demonstrable collection of observers and take their reports at face value.
I can agree that what reality does not distinguish as different is not different. I don't yet see why that has a different flavor in MWI than in CI.
I will grant you that is a valid insight, I can see elements there that would justify your analogy. If, for some reason, all the observers in our universe had to always be in the same one inertial frame, they might quite likely have come up with something like LET rather than SR. Their frame would have seemed very special, and when elementary particles at high speed took a long time to decay, they might have assumed something was affecting them, like a Lorentz aether. They would never have encountered an observer in a different frame, so they would not have cast the result in terms of invariants for that other observer, but rather in terms of physical effects on the elementary particle. But to that I have two immediate reactions:
1) if all the observers in the universe had to all be in the same inertial frame, there would be good justification for LET over SR, because there would seem to need to be a darn good reason for being all in the same frame, and
2) if we will think of MWI as a kind of transformation to observers in other reference frames, then what is the equivalent to the invariants that SR builds reality from? There is no equivalent to the norm of a 4-vector, which is empirically constructed. Instead, the only invariant I can see is the unitarity, but that is not empirically constructed, it is purely a rationalistic invariant. There is no function that takes as inputs each observers findings and generates a common scalar out of them, such that we could say our theories must only refer to those scalars.
But that's just the point-- that you have no idea what such evidence would look like is the proof that you do not experience it.That's not what you perceive when you flip a coin one time-- you do not experience "heads or tails", you experience heads, or you experience tails, which is something different .

Fra

Yes you're right, but I deliberately used the word describe to sharpen the point.

But there is no difference in what I refer to as "describing the future" and "predicting the future"; except in the way you understand what the point of "prediction" IS, and that's exactly my point (a) vs (b).
I don't mention it every thread. But the discussion in this thread is IMHO not so much about physics as it is about philosophy of physics. After all these discussions about the same old interpretations often (to me) seems non-constructive as noone rarely makes a point that aims to make a difference.

Also, it seems to me that I'm one of the few that represent these ideas on here, so for the benefit of a health discussion at least I feel I'm providing a fresh (possibly constructive, each one can judge that on their own) perspective to the discussed topic.

/Fredrik

Eqblaauw

I think it would be philosophy if we talked about the consequenses on our daily lives if mw where true. Wich are pretty big, especially if you talk about the splitting variant, that lives on while being denied by prominent proponents of mw. For example when someone dies in a quantum (chance) event you will feel a lot different when you believe he/she lives on in most of the (many, many) worlds. Sure it's sad you don't see him/her anymore, but that certainly isn't all we grieve about at funerals,
also it makes every history book seem pretty trivial
But making speculative claims about the physical existense of things isn't philosophy, it isn't even bad philosophy. Just speculative science (good or bad, it's up to you. Bad I would say, but hey).

Dmitry67

Sure, but please provide a definition of "being real" first.

Your tree had started when you were born. In some alternatives you were born earlier/later and in different conditions, I dont know how operation = (equals) works for the consiousnesses, so I can't say if they also belong to "you"

The same issue you have in Infinite Universe even without MWI, just alternative/copies are separated spacially.