I Why is there no consensus about the meaning of probability in MWI?

[PeterDonis]

You keep trying to make me say things I have never said or that are the opposite of what I said

I am responding to what you post, and explicitly quoting your own statements that I am responding to. So I am responding to what you say. I might not be responding to what you think you meant to say, but that should be an indication to you that your statements are not doing a good job of saying what you actually mean.

 

[pines-demon]

I am responding to what you post, and explicitly quoting your own statements that I am responding to. So I am responding to what you say. I might not be responding to what you think you meant to say, but that should be an indication to you that your statements are not doing a good job of saying what you actually mean.

This time I have to take your approach, to say that I already said what I have to say. I have already explained where I think this conversation can go. I have also agreed with some of your points specifically that we cannot come up with a meaning of probability for MWI here in PF. You have also asserted that such a problem has no resolution, fine. I think that misses the point and we have already discussed it. We are circling so let us move on.

 

[PeterDonis]

We are circling so let us move on.

Indeed.

 

[kered rettop]

As a moderator (which is a mentor's job), the only option I have for "handling" a thread in general is to close it. Are you asking me to close it?

With moderator duties to be done, to be done
A mentor's lot is not a happy one, happy one
- G&S

 

[kered rettop]

Not at all. OP, yourself and I have made more than clear that we do not intend to solve the problem of the meaning of probability in MWI.

Not quite. I do not expect to but I haven't sworn off trying! Which is why I asked about the equivalence of different meanings (of probability) in a maths forum. Big mistake. Straight into Kolmogarov's axioms and a load more impenetrable theory.

 

[dextercioby]

After yet another >250-post discussion (apart from some 10 with metadiscussion), I am fairly convinced that MWI stands for the "many words interpretation" of Quantum Mechanics.

 

[kered rettop]

After yet another >250-post discussion (apart from some 10 with metadiscussion), I am fairly convinced that MWI stands for the "many words interpretation" of Quantum Mechanics.

Ooh, I know the one! Good wasn't it? I have high hopes that, by some sort of ergodic principle, it will eventually get back on topic.

 

[kered rettop]

Going back to topic, I just went a little bit deeper into Wallace references, some are not helpful at all as they discuss "Many minds" theories
and things like that. However there is this paper of Saunders that seems to be focused on the probability problem:

https://philsci-archive.pitt.edu/19362/

It raises some of the points already raised before and more against the definition of probability in MWI, but reaches a not so negative answer saying that probabilities are somewhat emergent.

That would imply that an equiprobable assumption (for example) does not need to be imported but is itself emergent. Does he address this?

 

[pines-demon]

That would imply that an equiprobable assumption (for example) does not need to be imported but is itself emergent. Does he address this?

What do you mean by "equiprobable assumption" ?
This is what he says about equiprobability:

Given the branching structure to the universal wave function, it is clear what is the intended interpretation of probability (ratios in branch weights). Are we sure there is no rival alternative?There is one that has been taken seriously even by those sympathetic to Everett’s ideas: the branch-counting rule.24 It is the rule that on any branching event, all outcomes, all histories that have ensued at any given time are equiprobable. If from repeated measurements a large slew of histories result, the number with a given relative frequency (divided by the total number) determines the probability of that relative frequency. The result, for the Everett interpretation, is mayhem.

To take again the measurement of spin with initial state (2) [$|\psi\rangle=c_+|\phi_+\rangle+c_-|\phi_-\rangle$], after $N$ measurements the vast majority of states have relative frequency of plus-outcomes equal to one half, and likewise for minus-outcomes one half, entirely independent of $|c_+|^2$ and $|c_-|^2$. When this ratio differs significantly from unity, only a tiny minority of branches after $N$ trials comply with the Born-rule.

He continues on with other branch countings, but finds other problems.

 

[kered rettop]

What do you mean by "equiprobable assumption" ?

An assumption of equal probability. What else could it mean?

 

[pines-demon]

An assumption of equal probability. What else could it mean?

I do not know, you are the one exploring the realms of axiomatic probabilities. I would just be careful with "probability". Do you mean equiprobable as the example I cited? That 50% branches one way or another independently of the amplitudes?

Then my question would be what do you mean by that it does not need to be imported?

 

[kered rettop]

I do not know, you are the one exploring the realms of axiomatic probabilities. I would just be careful with "probability". Do you mean equiprobable as the example I cited? That 50% branches one way or another?

Obviously branches are not equiprobable. An assumption made in one context is not a principle that applies everywhere else.

Then my question would be what do you mean that does not need to be imported?

If, as Saunders claims, probability is emergent, then there is no need to assume anything about it. Since, Zurek, Carroll and so on use arguments that depend on equal probabilities (or similar), they must either assume it or else derive it.

Edit, re "importing" I mean making the assumption that probabilities are equal brings with it the whole package of what is meant by probability. So you're not just assuming a value, you're importing the meaning. It's inevitable unless probability can be derived instead.

 

[Fra]

I hope we can find together what is the "consensus about the [lack of] meaning of probability in MWI?"

I am asking what the sticking point is?

I still think this question is as close to a consensus we will come.

Unless we or observer/agents define it's purpose/utility of this measure, the it's consensus will be similarly unclear; that itself is I think a reasonable consensus.

Who is the "user" of the probability in MWI? MWI supporters answer this, I have no clue.

As I see it in contrast, the "user" of the probability in an agent/obsever centerd view, is the agent itself. It is a guiding measure for otherwise random walks.

In other interpretations, the measure is rather a "description" of frequencies in fictive ensembles. Who is the "user" of these fictive ensembles?

I fear we may be missing both the meaning of probability and the "meaning of consensus" which is not too unrelated the the meaning of observer equivalence

/Fredrik

 

[kered rettop]

I still think this question is as close to a consensus we will come.

Unless we or observer/agents define it's purpose/utility of this measure, the it's consensus will be similarly unclear; that itself is I think a reasonable consensus.

Who is the "user" of the probability in MWI? MWI supporters answer this, I have no clue.

As I see it in contrast, the "user" of the probability in an agent/obsever centerd view, is the agent itself. It is a guiding measure for otherwise random walks.

In other interpretations, the measure is rather a "description" of frequencies in fictive ensembles. Who is the "user" of these fictive ensembles?

I fear we may be missing both the meaning of probability and the "meaning of consensus" which is not too unrelated the the meaning of observer equivalence

/Fredrik

If you are just saying that we need to state the role of probability in MWI before we ask why people disagree about its meaning, then yes, that's the approach I advocate. If you're saying any more than that then I'm afraid I can't see it for all the agents milling around. Do they do the random walk you alluded to?

 

[pines-demon]

If, as Saunders claims, probability is emergent, then there is no need to assume anything about it. Since, Zurek, Carroll and so on use arguments that depend on equal probabilities (or similar), they must either assume it or else derive it.

Agree, but I still have to understand what emergent means according to Saunders. He cites back Wallace's The Emergent Multiverse.

Edit, re "importing" I mean making the assumption that probabilities are equal brings with it the whole package of what is meant by probability. So you're not just assuming a value, you're importing the meaning. It's inevitable unless probability can be derived instead.

Maybe. I mean he agrees that branch counting/weighting is not enough, but I do not think he derives probabilities then. Take a look at the paper do not take my word for it.

 

[sbrothy]

The Many-Worlds Interpretation (MWI) cannot explain the probabilities, because the whole point of MWI is to reject probabilities. Thus, probabilities are “masqueraded” as “world splittings”, and all possibilities are realized in inaccessible alternate world branches. It reminds a bit of “philosophical flapdoodle”.

"Philosophical flapdoodle"! :)

What a wonderful addition to my vocabulary! That word is going to pop up whenever I have a question about QM interpretations. My favorite is still the Montevideo one but that's again my intuition playing tricks on me.

 

[Fra]

If you are just saying that we need to state the role of probability in MWI before we ask why people disagree about its meaning, then yes, that's the approach I advocate.

Yes that's the primary thing I meant for the case of MWI.

If you're saying any more than that then I'm afraid I can't see it for all the agents milling around. Do they do the random walk you alluded to?

But I also tried to say more by associating "the role of probability" to the "usage/utility" of this for a "user". For me, whenever we talk about a "theory" there is implicit that is has a value or utility to someone, a user.

I think conventionally the "user" is always science, as in the community of human scientists and the measurement devices we control and monitor. Ie the "user" is not ONE agent, but more like a community or an equivalence class of agents. For example all macroscopic systems, as in Bohrs CI. This is why probability measures are usually statistical in their nature, and valid only at short time scales as noone can prepare ensembles of the universe etc.

There is no "gaming" concepts in that perspective, or inter-agent interactions that are analysed in depth, except that of special relativity and the class of intertial systems. This I personally think is a problem as I don't think the consensus between observers can be attained without interaction.

Do they do the random walk you alluded to?

This part is part of my reflection based on a agent based qbist style interpretation, not on mwi. But yes, in my view, all observers are in some unifiction domain "random walkers" on par with "fre falling" but in a more abstract space. But their random walk are guided, and that game would conceptaully have effective laws emerge, to the point that they become lawful including "effective conserved measures" etc. But this use of "probability" is not descriptive in the statistical sense, it is only normative. It's the angle that I find more useful for my own understanding. The real problem is that actions must always be decided, before "perfect" or "ideal" statistic exists. So in this sense the probability spaces must be emergent and subject to evolution.

But I think the "emergent probability" that Saunders talkes about must be of a different kind. The whole lineup even rejects me from looking into it in depth. It smells bad, but that's just me,

/Fredrik

 

[pines-demon]

But I think the "emergent probability" that Saunders talkes about must be of a different kind. The whole lineup even rejects me from looking into it in depth. It smells bad, but that's just me,

Yeah I do not know if it is worth it to focus on this emergence. However the paper touches on most of the problems with probability in MWI, I think we should focus on those.

 

[gentzen]

Yeah I do not know if it is worth it to focus on this emergence. However the paper touches on most of the problems with probability in MWI, I think we should focus on those.

For me, Saunders is symptomatic of a "MWI school" which claims "pilot-wave theories are parallel-universes theories in a state of chronic denial" and similar claims for consistent histories theories. Maybe that sort of claim is even successful in hitting close to real shortcomings of BM and CH. But it fails spectacularly in clarifying the relations between BM, CH, and MWI. It is always just the others who have shortcomings. This is the reason why I am no longer very keen on studying papers by Saunders in any detail.

 

[pines-demon]

For me, Saunders is symptomatic of a "MWI school" which claims "pilot-wave theories are parallel-universes theories in a state of chronic denial" and similar claims for consistent histories theories. Maybe that sort of claim is even successful in hitting close to real shortcomings of BM and CH. But it fails spectacularly in clarifying the relations between BM, CH, and MWI. It is always just the others who have shortcomings. This is the reason why I am no longer very keen on studying papers by Saunders in any detail.

Thanks for providing some background. While reading Saunders, it feels his arguments are for the total rejection of MWI but he still manages to a get a conclusion that goes into supporting a more obscure version of it.

Do you have any other source that discusses the problem of probability in MWI?

 

[Fra]

May I ask - If one entertains the notion of "multiple worlds" that does not interact, why do we care about consensus between them? It seems it's not a problem as long as they don't interact.

What is wrong with the more obvious: We have "many observers" in the same world, but that DO interact. And the quest is to understand how.

QM as we know, does not describe interacting observers. This why therer is a "problem" whenever we put a "classical measurement device" as part of the "quantum side".

I say lets solve the real problem, I dont see how then many worlds concept is constructive in any direction? How does it help?

/Fredrik

 

[kered rettop]

I say lets solve the real problem, I dont see how then many worlds concept is constructive in any direction? How does it help?

/Fredrik

It doesn't. It's not even a well-defined concept. It's primarily a name given to an interpretation, where things which it would be reasonable to call "worlds" emerge from the model and scare the living hell out of us. They also invite a catchy, if somewhat misleading, name for the interpretation. But these worlds are not used in the arguments of the model as far as I know, and they are certainly not postulated a priori.

 

[Fra]

It doesn't.

I think this all the consensus I need.

/Fredrik

 

[kered rettop]

I think this all the consensus I need.

/Fredrik

Really? You must be the only person on PF who regards my opinions as being a sufficient consensus. Even I take them with a very large pinch of salt!

 

[PeterDonis]

QM as we know, does not describe interacting observers.

How do we "know" this?

 

[PeterDonis]

Thread closed for moderation.

 

[PeterDonis]

After moderator review, the thread will remain closed.