Problems with Many Worlds Interpretation

[Ken G]

Lorentz Ether Theory represents maintaining prior views -- e.g. shaping the interpretation to retain the notion of absolute time that was present in previous physical theories.

I don't think the "priorness" of the views matters here at all, that is like saying any theory is better if it seems newer. There's no such principle of science, we all just have to watch out for intellectual inertia. The real issue here is, when we need to explain some surprising result, do we attach invisible ontological constructs that we cannot detect just to achieve a sense of cognitive resonance (that's what Lorentz did, I don't see it as making any difference if he expected there to be an aether-- he just wanted a mechanistic model for length contraction. I'll be that is what he would have said was his goal, not the desire to adhere as closely as possible to old modes of thought, he might have found that suggestion insulting), or do we, like Einstein, assert that an ontology that cannot be detected simply does not exist. Had we expected many worlds because we for some philosophical reason expected reality to be unitary, but then when we, for the first time, started doing observations of single particle systems and discovered they kept collapsing every time, you would be the one saying that we were holding onto old ideas to stick to the ontology of many worlds. Einstein would also have agreed-- he would have said that if observations keep collapsing every time we do one, perhaps it should be regarded as a physical law that this is what they do, and scrap the ontology of the unitary universal wave function. That's my point-- I claim a unitary universal wave function acts much like an invisible aether, and it doesn't make any difference to me which one is closer to old modes of thought.

CI is the same -- shaping the interpretation so as to retain the prior notion of definiteness.

Perceptions are definite, all CI does is recognize that and take that undeniable truth at face value. You say we are fooling ourselves that when we perceive definiteness, it means reality is definite. But it's all about what you take at face value-- the mathematical structure that you like for aesthetic reasons, but which changs every time we get a new theory, or the observations, which don't.

Special Relativity, on the other hand, represents taking the physical theory seriously enough to warrant reshaping our views on reality and reject^* absolute time.

Yes indeed, and was absolute time an empirical, or a rationalist concept? It's very important that you answer this to see what I'm saying here.

Or he might say "why do you care if the universe is definite or not, aliens on alpha Centauri are using a more advanced theory right now that dropped definiteness a millennium ago.

But are those aliens still doing measurements that get definite results? Why would they drop that? As I said, the theories change, the observations do not.

How can you tell? In what perceptible way does a unitary experience differ from a non-unitary one?

If you flip a coin, and you see "heads", that's a nonunitary experience, because it is a clear break in the symmetry. If you see a superposition of heads and tails (along with everything else that is coupled to it), that's a unitary experience. Are you saying that my perceptions are not viewed as being on a single branch of the many worlds? How are they "many worlds" if my perceptions are unitary, that's just one world and I'm perceiving it.

 

[Ken G]

But if, instead, we are considering a qubit which I had reason to believe was spin up in the x direction and you measured it around the Z axis and said "I measured spin up", I would fairly confidently believe that, conditioned on my knowledge the original state was in the x direction, we are now currently in the indefinite state of a mixture of one half "you saw spin up" and "you saw spin down". And I would agree with you: "I see spin up", and by the MWI, have high confidence that both of our usages of the word "up" is indefinite.

This seems to be an important issue, because I cannot see what you are saying here. The order of events here does not seem clear-- if you had reason to believe the particle had spin up, and I did a measurement that got spin up around some other axis, then it would not make any difference what your knowledge was, that knowledge has been superceded and become completely irrelevant to the state of the system. Or, if your knowledge came after my measurement, then my measurement is equally superceded and irrelevant. I see no indefiniteness in the scenario as you described it-- there is nothing indefinite about an outcome <observer 1 | outcome A>.

I don't need to keep two separate notions in my head of "quantum states evolving by unitary evolution" and "things collapse onto a definite state".

But there is every reason to keep those separate, because one is part of a theory, and the other is an outcome that is independent of theory. This is the fundamental difference between a rationalistic and an empirical entity, so not noticing that difference is like not being able to distinguish rationalism and empiricism.

You do perceive the unitary evolution of the wave-function.

By relative states and decoherence, unitary evolution of the wave-function looks like a transition to a mixed state.

By probability theory, mixed states look like definite outcomes obeying statistical laws.

You would describe your perceptions as being definite, and obeying statistical laws, right?

This is an important point to address, because it is framed like a syllogism that leads to the conclusion that I perceive unitary evolution, when it seems quite clear to me that I do not. So we must find the disconnect in the syllogism. It comes as no surprise to me that MWI generates mixed states, that is what I meant by the decoherence just being quantum mechanics. That all happens before you even get to MWI vs CI, we have no need of either yet. Where we find a distinction is when we take the mixed state that decoherence and quantum mechanics gives us, and ask the key question: why don't we perceive a mixed state, why do we perceive a definite outcome?

In other words, I flip a coin, and I don't look at it. It's either heads or tails, I just don't know-- that's the standard description of the mixed state of the coin. No MWI yet, no CI yet. Now I look at the coin. Now I need MWI or CI, because something weird happened-- I went from "it's either heads or tails but I don't know which" to "it's heads" or "it's tails". That is when I need an interpretation of just what the heck happened there. So there are really two very different "collapses" going on, and I think it causes a lot of confusion-- the first collapse is just decoherence, it's pure quantum mechanics (and doesn't even need to be called collapse, nothing collapsed it just decohered). The second collapse is what "collapse" really means-- the perception of a definite outcome in a way that breaks a symmetry between heads and tails that has always persisted right up until the moment I perceived one or the other.

So there's the misstep in your syllogism: mixed states do not look like definite outcomes that obey statistical laws, they look like "one or the other but I don't know which", whereas definite outcomes look like "one" or "the other"-- and I do know which. That doesn't look the same, not in a single experiment. A symmetry was broken, and MWI has no accounting for that, whereas CI calls it a "collapse". And many repetitions of the experiment does not recover the symmetry unless I throw away information-- if I just look at the total number of heads vs. tails I get, it will look the same as a mixed state, and MWI will account for it just fine. But I have perceived more than the ensemble outcome-- I have a sequence HTHTTHT or some such, and that sequence is still something that MWI has no account of at all excdpt to say "who cares." That's what I called the "cheat" of MWI-- it chooses not to care about what it cannot account for, then turns around and claims to be a complete description.

And in various circumstances, I would find it perfectly natural to paraphrase that with words like "t behaves like a continuous variable" or "t behaves like a real number".

That is saying that the individual outcomes are decided randomly! That's what collapse is-- that is just precisely what CI says. If MWI is making that claim, I say it's CI in an elaborate disguise. A random variable is allowed to generate a definite sequence, which breaks a symmetry and no such definite sequence is unitary. It's only unitary when no particular set of outcomes of the random variable is referenced-- and that's the step before we even get to the distinction between MWI and CI.

Did you mean for "time" and "t" to be different?

Of course, they are obviously different. "Time" is something that we experience and measure. Look up the definition of time in any physics glossary, there is not one that is going to say "time is a parameter in quantum mechanics." But the latter is what t is. If you don't see that difference, it's not surprising I'm having such a hard time getting you to understand what empiricism is. You seem to be so programmed to think rationalistically that the alternative is almost inscrutable to you! That may be the best service I can do here-- to get you to see this whole other way of thinking about physics, called empiricism.

 

[Fra]

Whille my own view is much more radical than CI, I often defend CI traits since I see my own view as a further development along the CI philosophy.

Then one can always argue that "than it's not the old CI anyway", but set aside from that agreed point I'll add some of my own views on the below question.

Really? CI has found a way to unify unitary evolution and collapse?

In my view, it's reasonable to say that they are "unified". Unification is made in the context of general inference and learning models. This also very relates to the point I tried to made that there are several ways to understand a "theory".

CI reasoning IMO makes better sense if we understand the "predictions" to simply be rational expectations suggested by our theory seen as an interaction tool. Ie. the purpose is to always learning more about the system. Thus, the intersting thing isn't just the prediction, but rather the feedback from the system from the "measurement".

If this was EXACTLY on expectations, it would hardly contain and New inforamtion.

So in the inference picture, where a theory is undertsood as an interaction tool (that is evolving as we learn) it's highly natural that the collapse is an information update, and the unitary evolution is the EXPECTED evolution, that holds until we get new information.

In this picture it's also clear that it does not make sense to talk about "predicting" WHEN and WHAT new information that arrives. That is even a logical contradiction. IF we could do that, we obviously already HAD this information!

What can be done is that sometimes ANOTHRE observer, can predict things. But then we aren't talking about the same inference system.

So my point is that all these well known arguments of CI, that it's "just an information update" that I know maks no sense for hte MWI'ers, IMO at least does make much MORE sense if you see the theory as an interaction tool.

That said, Ihave issues with "classical CI" too but I'll stop there.

/Fredrik

 

[Ken G]

I believe I see what you are saying, though I would frame it a little differently but might be saying much the same thing. To me, the heart of CI is an acceptance that "what happens" and "what is observed to happen" are just precisely the same things. That's why the "evolution with t of a wave function" would never be seen by CI as either a "behavior of a system" or a type of "happening to the system." The t evolution of a wave function is simply a mathematical script for generating expectations, when the parameter t comes with an expectation of corresponding to time intervals between measurements, and the Born rule gives us our statistically expected outcomes. I think your insight is how empty that description is, compared to the actual experience of existence, which is much more alive and much more surprising. I like the view that the surprise is really the point of the whole exercise, I think that puts determinism into its proper place. It also shows why MWI does nothing to account for the surprise, and why CI at least recognizes the presence of surprise but doesn't really know quite what to do with it. Above all, this shows that quantum mechanics is still incomplete as a description of our experience, or at least a description that accounts for its most essential element: the surprise.

 

[Fra]

the "evolution with t of a wave function" would never be seen by CI as either a "behavior of a system" or a type of "happening to the system." The t evolution of a wave function is simply a mathematical script for generating expectations, when the parameter t comes with an expectation of corresponding to time intervals between measurements, and the Born rule gives us our statistically expected outcomes. I think your insight is how empty that description is

I agree. Your rephrasing is fully in line with what I said.

However even in this "trivial picture", there are several a remaning non-trivial parts which is yet an open issue (I'm working on it though):

1) To show, explicitly, how the SPECIFIC unitary evolution FOLLOWS rationally merely from consistency requirements of the existing information set. With this I mean, to show exactly how the information structure folloes purely from an interaction history, and how a condensed "state of the history" ENCODES the future expectations.

That is - WITHOUT referring to classical input such as classical hamiltonians etc.

Noone has yet shown this. If I'm right this is possible, but it remains to be worked out.

2) We also need to show, if unitariy evolutio nis "just an expectation", HOW COME many observers very often does agree on the statistical outcome after a finite evolution in time? This is a big open issue too, on which there are ideas. Ie. to explain the de facto objectivity that exists. Here I think the issues of equilibration between observers (thus tuning of expectations) is the way to understand it.

Also no one has to my knowledge explained this. Decoherence is only a part of hte answer.

3) there is more too...

I like the view that the surprise is really the point of the whole exercise

Yes, and one cna take this yet another level and trace all the way down to the scientific method. I don't want to expand on it, but shortly this relates (indirectly) to what happens when a theory is falsified: you obviously find a new hypothesis - but HOW? This is the core point of learning. When you are right, it's somehow "trivial" from the point of vie of learning.

/Fredrik

 

[Dmitry67]

That's indeed the question. To a CI proponent, "being real" means "demonstrable by experiment"-- what is real is what is experienced by objective observers that we can (in principle) communicate with. To a MWI proponent, being real means fitting in with a mathematical structure that is generated by the minimal and most unifying conceptual principles. So to be consistent, both of the interpretations must build their models of what "I" am from similar stuff. That's going to be a lot easier for CI, because "me" is whatever I perceive myself to be, and the experiences of "me" are nonunitary.

This is a very strong claim, because you've just claimed that (real) Universe is finite and everything behind the Cosmological Horizon is not real.

Adding "is experienced by objective observers that we can (in principle) communicate with" you made your claim even stronger, declaring the interior parts of Black Holes not real.

Do you want to change your mind?

 

[Eqblaauw]

This is a very strong claim, because you've just claimed that (real) Universe is finite and everything behind the Cosmological Horizon is not real.

Adding "is experienced by objective observers that we can (in principle) communicate with" you made your claim even stronger, declaring the interior parts of Black Holes not real.

Do you want to change your mind?

define 'change your mind'

 

[Eqblaauw]

by saying 'what we know is real', instead of what's real, i think ken g's first claim is very sensible.The rest are speculations

 

[jensa]

I don't think it's quite that simple.

Depending on how far into philosophy you would like to go I am sure you are right.

I think the "assignment of reality" to the whole state operator can be considered a definition of a MWI, but I don't think it really makes sense to "assign reality" to a single term. If we do, we have to imagine an unknown process called "collapse" that eliminates the other terms. (Other people may have less of a problem with this than I do). So I consider the main option to the MWI to be the idea that nothing in the theory is "assigned reality". This is to assume that QM doesn't describe reality, that it just tells us how to associate probabilities with verifiable statements.

I think this is more or less exactly what I tried to say but was somewhat sloppy/inarticulate: When I talk of "branches" I refer to the sequence of labels $j,j',j'',\ldots$ which are associated with (perceived/not perceived) measurement results. And when I talk about assigning "reality" to one of the branches I only talk about the sequence of measurement results, not necessarily the corresponding parts of the mixed state operator. As such the state operator and the laws that govern its (unitary) evolution is simply a calculational tool to obtain possible outcomes of a sequence of measurements (branches) and their associated probabilities $p_jp_{j'}p_{j''}\ldots$.

Of course, IF you choose to consider the state operator as a part of reality then certainly all the branches must be equally real, and MWI naturally follows (unless you also introduce "true" collapse).

So to me it looks something like this:

MWI: State op.=Reality -> Deterministic (Unitary ev.); Perception -> Probabilistic

CI(w.o. collapse): State op. -> Deterministic (Unitary ev.); Perception=Reality->Probabilistic

Because it would be nicer to have a theory that in addition to telling us how to associate probabilities with verifiable statements, also describes reality. Since QM assigns probabilities so well, it would be weird to not even consider the possibility that whatever QM describes is reality. Without a physical process that eliminates the other terms in the post-measurement state operator, or a radical change in the laws of logic, I don't see how QM can be said to describe a single world.

To be honest I think the issue is whether we can accept the probabilistic nature of QM: How can a theory describe reality and yet not predict (with arbitrary accuracy for a single experiment) what we observe. If you say that what we observe is what we mean by reality, then I don't think such a theory can ever be constructed. The options are to say (CI:) QM does not describe reality but we may use part of this construction to probabilistically describe reality, or (MWI:) QM describes reality but what we perceive is only a (probabilistically selected) part of reality.

 

[Eqblaauw]

when you have an explanation that explains everything, it doesn't mean you have the right explanation. It just means you have a hypothesis that needs to be tested. Thinking having an explanation that explains everything is enough led to people taking freud seriously, even up to this day.

 

[Hurkyl]

Had we expected many worlds because we for some philosophical reason expected reality to be unitary, but then when we, for the first time, started doing observations of single particle systems and discovered they kept collapsing every time, you would be the one saying that we were holding onto old ideas to stick to the ontology of many worlds.

I can't respond to this since the details don't seem to make sense. Let me make a different scenario:

Suppose we had expected reality to be unitary, but then we discovered a new physical theory that made better predictions, but introduced a non-linear correction term into the Schrödinger equation so that evolution was no longer unitary and tended to collapse -- however we also quickly found some ad-hoc way to reproduce those calculations using unitary evolution. In this situation, I would be likely to favor the new collapse interpretation.

If we could naturally reproduce the calculations with unitary evolution, my opinions would depend on the particulars of each approach.

But the situation you proposed where we had started with the MWI and then someone discovered "Hey, you can reproduce all of MWI's predictions with an ad-hoc collapse interpretation", I would probably think "hey, that's pretty neat", and possibly "are there any useful features of that approach I can extract and adapt into the MWI?", but not really much more than that.

Einstein would also have agreed-- he would have said that if observations keep collapsing every time we do one, perhaps it should be regarded as a physical law that this is what they do, and scrap the ontology of the unitary universal wave function.

And, to the best of my knowledge, there hasn't been much progress in trying to find such a physical law. CI might become interesting again if progress is made.

That's my point-- I claim a unitary universal wave function acts much like an invisible aether, and it doesn't make any difference to me which one is closer to old modes of thought.

Really? I got the impression you cared very, very strongly about the old "reality is definite" mode of thought, because you keep saying things like

Perceptions are definite,

Yes indeed, and was absolute time an empirical, or a rationalist concept? It's very important that you answer this to see what I'm saying here.

Assuming such a classification even makes sense, it falls in the same category that one would place its negation.

If you flip a coin, and you see "heads", that's a nonunitary experience, because it is a clear break in the symmetry.

What if I see tails? Is that a nonunitary experience?

So you have proposed an experiment: I flip a coin.

• If I view heads, I proclaim my experiences non-unitary
• If I view tails, I proclaim my experiences non-unitary

Now, if unitary evolution predicts decoherence into a mixture of heads and tails, we have the following fact:

Unitary evolution predicts that I proclaim my experience non-unitary​

Because of this fact, I dismiss your experiment as not being useful to determine what it claims to determine.

If you see a superposition of heads and tails (along with everything else that is coupled to it), that's a unitary experience.

Ah! The state of the coin being pure is thermodynamically impossible! MWI most certainly does not predict that. In sufficiently "small" systems, though, we can and do see the superpositions. But there isn't much interest to that, because for sufficiently "small" systems, which pure states one calls a superposition is quite arbitrary.

How are they "many worlds" if my perceptions are unitary, that's just one world and I'm perceiving it.

You remember my game with the indefinite universe? I mentioned that the mathematics of the setup are identical to a different scenario with two universes and two experimenters. Worlds are a useful way to analyze a mixture, in the same way that Euclidean geometry is a good way to analyze vector spaces. (and that vector spaces are a good way to analyze Euclidean geometry!) Call it the theoreticians habit of actually naming the concepts they study, and sometimes naming after whatever other idea gave them inspiration.

 

[Hurkyl]

mixed states do not look like definite outcomes that obey statistical laws, they look like "one or the other but I don't know which"

Only in the bird's eye view. (and even then only in an ignorance interpretation of probabilities)

The relevant question is how they look to the frog's eye.

 

[Ken G]

Only in the bird's eye view. (and even then only in an ignorance interpretation of probabilities)

The relevant question is how they look to the frog's eye.

Yes, that's right, and what I am talking about is the frog's eye, because I'm distinguishing the moment before you look at the coin, and the moment after. Those are what is different from the frog's eye, they are not different from the bird's eye, and that's the whole problem when MWI gets explained from the bird's eye as you did above-- it misses our experience as frogs. As Fra made me see, what's different is the surprise. That's also why MWI is good rationalism and poor empiricism (and CI is good empiricism and poor rationalism), this is exactly the structural difference in those approaches.

 

[Ken G]

This is a very strong claim, because you've just claimed that (real) Universe is finite and everything behind the Cosmological Horizon is not real.

That is just exactly what empiricism demands, yes-- the cosmological principle is a rationalist principle when taken as an ontology (it is an empiricist principle when you say "as far as we have seen so far, the universe behaves like this, and we have no reason to expect otherwise at the moment"). Note empiricism also does not assert that what is beyond the horizon is "unreal", it merely asserts agnosticism around what is beyond that horizon. Empiricism is being awake to the things that might change, i.e., what has not been observed yet that might be different from what has. It's true what Hurkyl mentioned-- this already requires some form of rationalism to be mixed in there, empiricism does require logic. But it tries to stay honest about what we know, and what we don't.

Adding "is experienced by objective observers that we can (in principle) communicate with" you made your claim even stronger, declaring the interior parts of Black Holes not real.

True, as long as you bear in mind the distinction between "not real" and "unreal." In other words, empiricism does not say there is nothing inside an event horizon, or that the material we see crossing it disappears into a pandimensional void, or even that GR can't make predictions about what is in there. It simply says we don't know what is in there, because the universe has put up a "no trespassing" sign. There are other competing theories to GR, by the way, about what is going on inside event horizons, so this is the point of empiricism: we resist pretending we know what we have not empirically established. That is one of the clear lessons of the history of physics.

Do you want to change your mind?

Why would I, these are quite clear tenets of empiricism. To be clear, I do not wish to claim that empiricism has it all right and rationalism has it all wrong-- rationalism has had its successes too. My actual point is we need to be aware when we are adopting one perspective or the other, because I've found that rationalists often think they are "just doing physics", and seem to have completely missed the two-headed nature of this discipline.

 

[Ken G]

So to me it looks something like this:

MWI: State op.=Reality -> Deterministic (Unitary ev.); Perception -> Probabilistic

CI(w.o. collapse): State op. -> Deterministic (Unitary ev.); Perception=Reality->Probabilistic

I completely agree with essentially everything you said, but let me suggest a different construction for the CI case. Since you started the MWI situation with what we will call reality, we can do the same for CI:

Perception = Reality <-- limited prediction by unitary evolution <-- QM formalism

and MWI:

QM formalism = Reality <-- limited perception by observer

This should make it clear the fundamental difference in priorities that we call empiricism and rationalism. It's true that these are actually attitudes about what is the most proper path to truth, but each approach forms a very clear opinion about the proper ontology, which goes hand-in-hand with the proper way to know truth about what exists, or even what we want to imagine exists.

 

[Eqblaauw]

yes al very true (perhaps) now respond to my simple post (both of you)

 

[Eqblaauw]

I will quote it again for you, with some modifications (the post was a little bit unclear)

when you have an explanation that explains everything if it where true, it doesn't there by mean you have the right explanation. It just means you have a hypothesis that needs to be tested. Thinking that having an explanation that explains everything if it where true is enough led people to take freud seriously, even up to this day.

 

[Ken G]

I
But the situation you proposed where we had started with the MWI and then someone discovered "Hey, you can reproduce all of MWI's predictions with an ad-hoc collapse interpretation", I would probably think "hey, that's pretty neat", and possibly "are there any useful features of that approach I can extract and adapt into the MWI?", but not really much more than that.

Except that is not what I proposed at all. The issue keeps coming back: you are not getting that the reality we perceive is nonunitary, even though unitarity does tell us how our statistical expectations toward reality evolve while we are not looking at that reality.

I'm not sure how to make that distinction more clear, until you accept that something happens, perception and information-wise, when I take a flipped coin that I haven't looked at, and look at it, and that neither MWI, not unitary evolution, make any accounting of that separate something, because they both take the birds-eye view of a closed system rather than the perception view of a part of that system. If no parts in any system had perception, and there was somehow an all-knowing god that had your birds-eye view, then MWI would make perfect sense, and empiricism would not even exist. I see it as rather odd that the rationalistic camp has now in effect taken over the view that it makes sense to talk about the perspective of god.

Really? I got the impression you cared very, very strongly about the old "reality is definite" mode of thought, because you keep saying things like

I don't believe you understand anything I'm saying about "definiteness", because you don't seem to be able to "get" empiricism at all. Empiricism has a beautifully simple view of what is definite-- what is definite is what is measured. Yes there is much difficulty around issues like measurement uncertainty, and the role we can attribute to hypothetical observers, but the core idea is pretty simple in practice.

Assuming such a classification even makes sense, it falls in the same category that one would place its negation.

I haven't the vaguest idea what you are saying here, the classification makes obvious sense. The classification we are talking about is distinguishing a t parameter in a theory from our experiences of measuring time. What I can't get my head around is why you can't get your head around that distinction.

So you have proposed an experiment: I flip a coin.

• If I view heads, I proclaim my experiences non-unitary
• If I view tails, I proclaim my experiences non-unitary

Correct. Either outcome is nonunitary, that's just exactly right.

Now, if unitary evolution predicts decoherence into a mixture of heads and tails, we have the following fact:

Unitary evolution predicts that I proclaim my experience non-unitary​

Because of this fact, I dismiss your experiment as not being useful to determine what it claims to determine.

You are again missing the distinction between an experiment that results in a mixed state of outcomes, and one that results in the perception of a definite outcome. Your syllogism overlooks a key assumption you made-- that the perception of a definite outcome is possible. So you assumed the evolution was nonunitary. So actually what you just proved is "assuming evolution is nonunitary, unitary evolution predicts a non-unitary outcome." I'm not surprised by that.

I think we need to figure out what a unitary outcome looks like. Let's say we start out with a particle in a spin up state. We subject it to a field that should precess it 90 degrees, and ask what it's state is now. It will be in an equal superposition of spin up and spin down, that's a unitary evolution. To "see" that unitary final state, we need to be able to do a measurement that can see a superposition of spin up and spin down. Any other outcome of that measurement will be nonunitary, because we just said what the unitary state was, and so that's what we would need to measure. This is the point I made earlier, if we had some philosophical expectation for particles to evolve unitarily, but no one had ever figured out how to measure a spin, we would have expected such a measurement, when it became possible, to yield the superposition we expected.

But then when people actually tried to figure out ways to measure spin (think Michelson-Morely here, in the LET analogy), they kept finding that reality foiled them-- they kept getting spin up or down, rather than the unitary superposition they expected. So they invented "many worlds" to explain how the state could still be unitary, but they weren't getting it in the experiment. Of course they also had to invent some mechanistic way that the experiments kept giving one result (think aether). But then Bohr comes along (think Einstein), and says, "if every experiment is confounded into giving a definite result, maybe we should drop the whole unitary view of what is really happening. Maybe we should take what we see as what happens, and make it a principle that observers see collapsed results, and build our theory around a way to cobble together the results that observers we can communicate with actually get." And that's CI. MWI is clinging to the "outmoded" (in this scenario) way of thinking that "what happens" must be unitary.

In sufficiently "small" systems, though, we can and do see the superpositions.

Absolutely not, it's your rationalism creeping in again. What do we actually see, and what is the jump of inference you are making? What superposition do we actually see? Never, we never actually see a superposition of anything, that's the whole point. We see interference patterns in an ensemble, but the individual outcomes are always collapsed-- we never actually see a superposition. Reality confounds us every time-- just like it did with the aether.

 

[Eqblaauw]

ok, but could you respond to my post please?

 

[Hurkyl]

Yes, that's right, and what I am talking about is the frog's eye, because I'm distinguishing the moment before you look at the coin, and the moment after. Those are what is different from the frog's eye, they are not different from the bird's eye, and that's the whole problem when MWI gets explained from the bird's eye as you did above-- it misses our experience as frogs.

If you're talking about the frog's eye view, then why did you bring up the bird's eye view?

You've confused yourself. Look carefully at what you have argued -- you denied my description of the frog's eye view on the grounds that the bird's eye view looks different. And then you complained about the very idea of discussing the bird's eye view because it's not the frog's eye view!

If you want to consider the frog's eye view of a mixed state, try looking through the frog's eyes. :tongue: (hint: it looks like a definite outcome)

 

[Eqblaauw]

If you're talking about the frog's eye view, then why did you bring up the bird's eye view?

You've confused yourself. Look carefully at what you have argued -- you denied my description of the frog's eye view on the grounds that the bird's eye view looks different. And then you complained about the very idea of discussing the bird's eye view because it's not the frog's eye view!

If you want to consider the frog's eye view of a mixed state, try looking through the frog's eyes. :tongue: (hint: it looks like a definite outcome)

your pretended arrogance is appalling

 

[Hurkyl]

I don't believe you understand anything I'm saying about "definiteness", because you don't seem to be able to "get" empiricism at all.

This last post makes it look like you're talking about a notion roughly synonymous with "precision". That's so totally not what I'm talking about -- it is not what the "definite" in "definite outcomes" refers to.

Maybe "indeterminate" would be a better word to use? It's pretty similar to the notion of an indeterminate variable. The indeterminate real number X is a real number. X is not multiple real numbers we haven't chosen between. It is not an imprecise real number. It is not some number we are ignorant of. It is simply X.

And even while X is indeterminate, and X+1 is indeterminate, other things like X-X or X/X are determinate real numbers.

From the bird's eye, the result of the coin flip is indeterminate -- and this is formally represented by writing the result as an indeterminate variable C over the values {heads, tails}. C is not imprecise. C is not something we are ignorant of. C is simply C.

And it's fairly easy to see that given the hypotheses:

• If "C = heads", I proclaim my experiences non-unitary
• If "C = tails", I proclaim my experiences non-unitary

That "I proclaim my experiences non-unitary" is a theorem. It's truth is determinate.

 

[Hurkyl]

Never, we never actually see a superposition of anything

Do we actually see "spin up around the x axis"? Then in exactly the same sense, we actually see the superposition "$|z+\rangle + |z-\rangle$".

 

[BruceW]

Do we actually see "spin up around the x axis"? Then in exactly the same sense, we actually see the superposition "$|z+\rangle + |z-\rangle$".

Good point. We measure the component of spin in the x-direction. And this doesn't mean we see the spin up around the x axis. (so we never actually see eigenstates or superpositions), we only see the eigenvalues.

 

[Ken G]

Is this the post you mean:

when you have an explanation that explains everything, it doesn't mean you have the right explanation. It just means you have a hypothesis that needs to be tested. Thinking having an explanation that explains everything is enough led to people taking freud seriously, even up to this day.

It seems to me that an explanation that explains everything is more or less the only definition of the "right" explanation I can think of. But I agree that often when we think we explain everything, it is only because we have carefully chosen what we want to explain such that it matches what we do explain, and that is a trap.

 

[Ken G]

If you're talking about the frog's eye view, then why did you bring up the bird's eye view?

Um, to contrast them.

And then you complained about the very idea of discussing the bird's eye view because it's not the frog's eye view!

Here's how I translate everything you said: "you are noticing a difference between the frog's eye and the bird's eye view, and asking which perspective is the one science should be framed from." Yes, guilty as charged.

If you want to consider the frog's eye view of a mixed state, try looking through the frog's eyes. :tongue: (hint: it looks like a definite outcome)

No, not at all. The frog's eye view of a mixed state looks like a coin that the frog hasn't looked at yet. That's just exactly what it looks like. I think this is your primary stumbling block to seeing basically everything I'm saying, so we should look closer. You are actually saying that there is no difference in the experience of knowing that a coin has been flipped, and adopting the natural attitude that it is in a definite state, but not knowing what that state is-- and looking at the coin to know what state it is in. I would say those are very clearly different experiences, and each experience has a name-- the first experience is called the experience of a "mixed state", and the second is called the experience of a "definite outcome."

 

[Ken G]

Good point. We measure the component of spin in the x-direction. And this doesn't mean we see the spin up around the x axis. (so we never actually see eigenstates or superpositions), we only see the eigenvalues.

Yes, that works fine for me, we only see eigenvalues. That is very much the point I'm making.

 

[Eqblaauw]

Ken G, because of the faults in the post I posted this:

I will quote it again for you, with some modifications (the post was a little bit unclear):

when you have an explanation that explains everything if it would be true, it doesn't there by mean you have the right explanation. It just means you have a hypothesis that needs to be tested. Thinking that having an explanation that explains everything if it where true is enough led people to take freud seriously, even up to this day

but your reply response to this quite nicely, but feel free if you have something to add to your additional comment

 

[Travis_King]

Eqblaauw. That post merely describes skepticism. Something science accepts and allows. But skepticism doesn't mean you should go around not believing the most likely solution just because it might "just look like it's right".

Not that I am taking stance here either way, since it seems to me that we have peanuts in the way of data to making such grandious hypotheses about the universe.

Also, offtopic but, Freud was taken seriously because his arguments were compelling and the majority of people (psychologists included) are not incredibly self aware and cannot, themselves, readily examine their own psyches. It's not because his theories so accurately described psychological phenomena that one would have to have been crazy to not accept them.

 

[Hurkyl]

Um, to contrast them.
Here's how I translate everything you said: "you are noticing a difference between the frog's eye and the bird's eye view, and asking which perspective is the one science should be framed from." Yes, guilty as charged.

So are you finally adopting the long-standing tradition of formulating physics from the bird's eye perspective because it's much easier to understand and compute with, and deriving the frog's eye view from it?

Or maybe... are you going to stop making assertions about the definiteness of your experiences, because definiteness is a topic about the bird's eye view?

The frog's eye view of a mixed state looks like a coin that the frog hasn't looked at yet.

The indefinite outcome position allows for it to be a mixed state after the frog has looked at it.

You are actually saying that there is no difference in the experience of knowing that a coin has been flipped, [STRIKE]and adopting the natural attitude that it is in a definite state, but not knowing what that state is[/STRIKE] looking at the coin to know what state it is in, but having no idea if the state is definite -- and looking at the coin to know what state it is in.

I've corrected the above quote to say what I'm actually talking about.



Maybe I should take a different tack and stop talking about states and outcomes. You like propositions and reasoning about our experiences, right? You would agree with all of the following statements, conditioned on the hypothesis of an ideal^* coin flip?

• If I saw heads and look again, I will see heads again
• I saw tails or I saw heads.
• I did not see both tails and heads.
• If I saw tails, then you saw tails

Does the claim that these conclusions hold capture your notion that our experiences are definite? Any other things you want to add to the list?


*: meaning outcomes like "lands on edge", "a bird flew away with it", and "spontaneously converts into pure energy" don't happen.

 

[Ken G]

This last post makes it look like you're talking about a notion roughly synonymous with "precision". That's so totally not what I'm talking about -- it is not what the "definite" in "definite outcomes" refers to.

I agree, it's not precision, we're both idealizing precision. We have heads or tails, there's no measurement uncertainty here. Those are the eigenvalues, they are all we ever experience (the mixed state before we look is an experience of other eigenvalues from which we infer a kind of missing information about the coin, and from the eigenvalues we do get we assemble a notion that the coin is in a definite state and we just don't know what it is). Here's the key point: eigenvalues are nonunitary. MWI gives us no account as to why we only experience eigenvalues, and I described a hypothetical progress of scientific knowledge where that might have come as rather a big surprise, had we expected evolution to be unitary in the first place. In that scenario, clinging to unitariness would have been clinging to the "god frame" of Lorentz's aether, justifying the way the LET analogy works for me.

And even while X is indeterminate, and X+1 is indeterminate, other things like X-X or X/X are determinate real numbers.

Yes, you can take information-free expressions involving X if you like. I can also wash all the paint off a Van Gogh using bleach. So what?

From the bird's eye, the result of the coin flip is indeterminate -- and this is formally represented by writing the result as an indeterminate variable C over the values {heads, tails}. C is not imprecise. C is not something we are ignorant of. C is simply C.

That's not the bird's eye view, that's the rationalist view. I believe I see the problem here-- there are really 3 views we are talking about, but thinking that there are only 2 is creating confusion. Here are the three views, I will express them from the empiricist perspective:

frog's eye: this is our own view, we see only eigenvalues, but we don't see every eigenvalue that has been decohered, we only see the ones our experience has brought us into contact with. This is like my own knowledge of Chicago-- I have seen it myself, I do not just need to talk to others who have seen it. But there are many things about Chicago I haven't seen or heard about, and I treat them as mixed states-- they are the happenings that didn't happen to me and I am not privy to.

bird's eye: this is a view no living thing actually has, but we can imagine it easily enough, while still being true to empiricism. This is the union of all the eigenvalues that observers whom I could talk to (in principle) have experienced. This is "objective reality" for the empiricist (where to avoid issues about trees in woods, we put in the "in principle" part-- we don't care if anyone was really looking, what matters is that the information is accessible in principle, we have a way to test if the tree fell or not even if we never do the test.)

god's eye: this is the view no living thing could have even in principle, it is fundamentally meaningless in the context of perception as we know it. It is the view of a supernatural being that has abilities we lack-- the ability to see a superposition of eigenvalues. This is what an empiricist would label a rationalist fantasy, expressly because it requires supernatural abilities to perceive (not think about, perceive in an experiment). The rationalist is perfectly fine with building the primitive elements of reality from this kind of view, and the empiricist demands a demonstration of this view, a demonstration that never comes.

Now here's the point. You imagine that the many worlds make sense at the level of the bird's eye view (or even the frog's eye, I can't tell because that part really mystifies me). That's why you don't see that it is rationalistic. I claim that the many worlds exist nowhere except in the god's eye view, and that's why it is so very rationalistic.

And it's fairly easy to see that given the hypotheses:

• If "C = heads", I proclaim my experiences non-unitary
• If "C = tails", I proclaim my experiences non-unitary

That "I proclaim my experiences non-unitary" is a theorem. It's truth is determinate.

I already dispelled that syllogism by pointing out where you assumed that it was only possible to perceive "heads" or "tails", which is the entire reason that the outcomes are nonunitary. So you embedded the assumption that the only possible outcomes were non-unitary, and then used that to prove a theorem that was in fact part of your assumptions. See if your theorem works if the possible experiences of the state are "heads", or "tails", or "a superposition of heads or tails." Where is the proof now?

 

[Ken G]

The indefinite outcome position allows for it to be a mixed state after the frog has looked at it.

You have a bizarre interpretation of what a mixed state is, that's all I can say. Most likely it is so rationalist a stance it can't even conceive of empiricism. So let's concentrate on eigenstates and superposition states, it seems we can get to the heart of the nonunitarity of our experiences perfectly well in that language.

Maybe I should take a different tack and stop talking about states and outcomes. You like propositions and reasoning about our experiences, right? You would agree with all of the following statements, conditioned on the hypothesis of an ideal^* coin flip?

• If I saw heads and look again, I will see heads again
• I saw tails or I saw heads.
• I did not see both tails and heads.
• If I saw tails, then you saw tails

Yes, I'm fine with all those, they are essentially the empirical bedrock of objectivity and the meaning of an eigenvalue. The second one is most key, that one is equivalent to the statement "experience is nonunitary."

Does the claim that these conclusions hold capture your notion that our experiences are definite? Any other things you want to add to the list?

Yes, I can't think of anything to add at present.

 

[Travis_King]

You'll have a hard time convincing me, personally, that the uncertainty that is present in subatomic particles does or can manifest itself on the macro scale. Superposition of up and down spins works because the particle experiences the effects of quantum uncertainty. A quarter does not, because it is not a particle, but a structure of interacting particles.

 

[Hurkyl]

Yes, I can't think of anything to add at present.

Good. All of those propositions are also valid in the sort of indefinite outcome interpretation I'm talking about.

Thus, when wearing his "empiricist" hat, the scientist will admit that none of his experiences can distinguish between definite and indefinite outcomes -- on this topic, empiricism has nothing to offer.


For the sake of streamlining the argument, I'm not going to go through the rehashing of the logical consequences of the above point and that they deny the assertions you've made. If you can't agree on the above point, there's not much reason to explain its consequences.

 

[Ken G]

Thus, when wearing his "empiricist" hat, the scientist will admit that none of his experiences can distinguish between definite and indefinite outcomes -- on this topic, empiricism has nothing to offer.

I am not following the significance you attach to this statement. If no experience can distinguish between definite and indefinite outcomes, then the empiricist says there is no difference, and labels all experiences as definite outcomes. Yes, that is indeed just what empiricists do. So? The question remains: why is a superposition experienced neither at the frog's eye level, nor at the bird's eye level, but only at the god's eye level? Note that to the empiricist, there is no god's eye level of perception, because perceptions are done by perceiving agents that we actually understand.