I What is the current status of Many Worlds?

[PeterDonis]

tools that were origionally developed as tools of philosophy.

I don't think logic and inductive reasoning were originally developed as tools of philosophy. If they were originally developed as tools of anything, it was mathematics.

Philosophy, perhaps, has used these tools to try to discover propositions with more generality than any other discipline, which is both a strength and a weakness. It's a strength because the more general a proposition is, the more ways you can use it. It's a weakness because the more general a proposition is, the less it is likely to be able to tell you. In science, in particular, I think the weakness weighs much more than the strength.

 

[Demystifier]

I don't think logic and inductive reasoning were originally developed as tools of philosophy.

They were. https://en.wikipedia.org/wiki/Term_logic

 

[Lord Jestocost]

Remind me - what does ontology mean?

When it is allowed to answer:

Ontology is an easy concept. Maybe, some get mixed up with the term “ontology” because there is a meta-question which questions – so to speak – the basic question of ontology itself. David J. Chalmers puts it in a nutshell in “Ontological Anti-Realism”:

“The basic question of ontology is ‘What exists?’. The basic question of metaontology is: are there objective answers to the basic question of ontology? Here ontological realists say yes, and ontological anti-realists say no.” [bold by LJ]

That’s it. It’s only the meta-question which unsettles physicists who cling with ferocity to the concepts of daily life, which – in the scientific language of physics – may be refined to the concepts of classical physics.

 

[haushofer]

It is? Then what is considered the "modern" formulation of Bohmian mechanics? I thought the whole point of BM was to have a deterministic equation for the time evolution of particle positions, just like in classical mechanics. To do that you need the quantum potential.

No, determinism is not the primary concern of BM; ontology (a realist's form of it) is.

 

[PeterDonis]

They were. https://en.wikipedia.org/wiki/Term_logic

Aristotle may have been the first to explicitly codify these tools, but he wasn't the first to discover or use them. The Greek geometers were already using them (and the Greeks weren't the first to do geometry either), though they did not systematize their use of them until Euclid, who was a rough contemporary of Aristotle.

 

[Sunil]

I think many worlds is not even an interpretation. It is a big confusion. It would be better named "many words". Or, in more detail, "many not well-defined words". The basic objection: If there would be many worlds, then the notion of probability would simply not make sense.

 

[martinbn]

I think many worlds is not even an interpretation. It is a big confusion. It would be better named "many words". Or, in more detail, "many not well-defined words". The basic objection: If there would be many worlds, then the notion of probability would simply not make sense.

That is true for most interpretations. By definition an interpertation is many words added on to the core QM.

 

[gentzen]

I think MWI as a project was completed by Consistent histories.

I am familiar with both MWI and Consistent histories, and I get sad when MWI and Consistent histories are mixed up by people unfamiliar with Consistent histories. (I am not familiar with the Decoherent histories variant by Gell-Mann and Hartle, and I don't get David Wallace's point in section 3.7 Decoherent histories, and cannot see the connection to the Consistent histories framework I am familiar with.)

But you are the Consistent histories expert here, so if you acknowledge that there is that connection to MWI, then I guess it is really there, and it is my fault that I don't get it. From my perspective, Consistent histories seems to work with the statistical operator, and a "reasonably complete" mathematical description. MWI seems to insist on a pure wavefunction, doesn't use a "more complete" mathematical description compared to CI, and instead seems to invoke philosophical arguments for getting a "reasonably complete" description of the observed phenomenology.

Many people tried to get somewhat familiar with MWI. Finding people familiar with even the most basic notions of Consistent histories seems difficult. Understanding the basic notion of Consistent histories seems easy to me, compared to understanding the philosophy behind MWI. More fundamental, Robert Griffith or Roland Omnès don't seem to claim that Consistent histories is a complete interpretation of quantum theory, but MWI proponents seem to make exactly that claim with respect to MWI.

 

[Morbert]

I am familiar with both MWI and Consistent histories, and I get sad when MWI and Consistent histories are mixed up by people unfamiliar with Consistent histories. (I am not familiar with the Decoherent histories variant by Gell-Mann and Hartle, and I don't get David Wallace's point in section 3.7 Decoherent histories, and cannot see the connection to the Consistent histories framework I am familiar with.)

But you are the Consistent histories expert here, so if you acknowledge that there is that connection to MWI, then I guess it is really there, and it is my fault that I don't get it. From my perspective, Consistent histories seems to work with the statistical operator, and a "reasonably complete" mathematical description. MWI seems to insist on a pure wavefunction, doesn't use a "more complete" mathematical description compared to CI, and instead seems to invoke philosophical arguments for getting a "reasonably complete" description of the observed phenomenology.

Many people tried to get somewhat familiar with MWI. Finding people familiar with even the most basic notions of Consistent histories seems difficult. Understanding the basic notion of Consistent histories seems easy to me, compared to understanding the philosophy behind MWI. More fundamental, Robert Griffith or Roland Omnès don't seem to claim that Consistent histories is a complete interpretation of quantum theory, but MWI proponents seem to make exactly that claim with respect to MWI.

The sentiment that consistent histories is continuous with many-worlds and in a sense completes it is expressed most explicitly by Hartle ( https://philpapers.org/rec/SAUMWE-3 )

"These conferences marked 50 years of Everett’s formulation of quantum theory. But they were only a year away from marking 25 years of the decoherent (or consistent) histories quantum theory that can be viewed as an extension and to some extent a completion of Everett’s work"

At the same time MW proponents see consistent histories research as in service of many-worlds. It provides machinery for exploring branching, assigning them probabilities, quantifying decoherence etc.

MW proponents regularly balk at consistent histories and have made plenty of objections, so it's certainly a hill that needs to be defended. I acknowledge the qualitative distinctions between the MW and CI camps.

 

[Quantumental]

Isn't Consistent Histories just MWI with more coherent decoherence math? I've always struggled to separate the two. Sure in CH/DH you can postulate that post-decoherence only one subset is 'real', but that does not seem to be derived from any of its math. In fact, I am not sure why they are even categorized as different. I don't see any bigger difference between CH/DH and MWI than I do with say the Oxford interpretation of Everett, classical Everett or Everett interpreted by Lev Vaidman for instance

 

[Quantumental]

As far as I know there is no way to derive Born's rule in MWI, so the theory cannot make any predictions. There is no way to ascribe any meaning for probabilities in MWI.

This I do agree with and in many ways were the inspiration for the title of this post; even among the most ardent proponents of MWI there is fierce disagreement on this topic. Decision-Theoretic approach by David Deutsch and David Wallace vs Lev Vaidman's view for instance, another one that comes to mind is Zurek's Envariance approach.

Some conclude that this renders MWI not only wrong, but 'not even wrong', since you literally can't call it science, but it could still be true, it would simply mean that we would have to accept that we are living in a universe where probabilities make sense and that there's infinite other versions of us where the Born Rule is violated repeatedly. This 'incoherence' problem would usually turn people off in the same way that Solipsism does: "Sure, it could be true, but it makes no testable predictions and its consequences mean that there is no science or logic to follow." which does not tend to sit well with either experimentalists or theoreticians. For some reason though, with MWI's fanfare over the past 20 years, this 'elegance' only reinforces their stance. It's fascinating.

 

[gentzen]

Isn't Consistent Histories just MWI with more coherent decoherence math? I've always struggled to separate the two.

You try to tease me? Let me check: Can you tell me the basic notion(s) of Consistent histories?

 

[Quantumental]

You try to tease me? Let me check: Can you tell me the basic notion(s) of Consistent histories?

Unless you're injecting a 'slight of hand' indeterministic element, how does CH derive a singular clear history which represents reality?

 

[gentzen]

Unless you're injecting a 'slight of hand' indeterministic element, how does CH derive a singular clear history which represents reality?

No, that is not related to the basic notion(s) of CH. Want to try again? By the way, CH does not claim to be a deterministic interpretation, and its indeterminstic elements are not 'slight of hand'.

 

[PeroK]

On a point of English spelling, it's "sleight of hand".

https://www.dictionary.com/browse/sleight-of-hand

 

[AndreiB]

Can you tell me the basic notion(s) of Consistent histories?

It seems to me that CH looks like a hidden variable theory (the measured values are "there" before measurement and measurements reveal them). This allows them to provide an explanation for EPR correlations. But then they add the idea that all frameworks are equally valid. I cannot understand this. If I measure the Z-spin, in what sense is the X-framework valid? The X-framework does not model the experiment being done and says nothing about what was observed.

 

[AndreiB]

This I do agree with and in many ways were the inspiration for the title of this post; even among the most ardent proponents of MWI there is fierce disagreement on this topic. Decision-Theoretic approach by David Deutsch and David Wallace vs Lev Vaidman's view for instance, another one that comes to mind is Zurek's Envariance approach.

I think the decision-theoretic approach is circular. It assumes that rational agents exist. I see no reason to grant them this assumption. It might be that in a world described by MWI rational agents cannot exist.

Vaidman introduces a "measure of existence". I don't understand the meaning of this. Are the copies in a world with a smaller amplitude less real? Do they develop a sloppy memory so that they "forget" their past measurements? This should not happen since the copies are supposed to be identical in all respects except for the observed measurement outcome.

I am not familiar with Zurek's Envariance approach. I'll try to look at it.

For some reason though, with MWI's fanfare over the past 20 years, this 'elegance' only reinforces their stance. It's fascinating.

Yeah, I agree. Even Weinberg was for some reason compelled to some extent by MWI. It's hard to understand why.

 

[gentzen]

But then they add the idea that all frameworks are equally valid. I cannot understand this. If I measure the Z-spin, in what sense is the X-framework valid? The X-framework does not model the experiment being done and says nothing about what was observed.

Nice, somebody who is somewhat familiar with CH. In fact, the answer from Omnès to your question about "in what sense the X-framework is valid" is quite in line with your comment. I once quoted from “Quantum Philosophy” by Roland Omnès parts of the replies to such objections:

Some of them will lead to the same conclusions and they are just as good. Others will be useless, not necessarily wrong but only idle talk of no consequence. Why bother? Asking questions about the existence of useless histories amounts to performing calculations that are of no help in solving a problem. They belong in the waste-paper basket.

 

[Morbert]

It seems to me that CH looks like a hidden variable theory (the measured values are "there" before measurement and measurements reveal them).

CH as a formalism is itself open to different interpretations. A realist interpretation is most commonly put forth by Robert Griffiths ( https://arxiv.org/abs/quant-ph/0001093 ), but an antirealist interpretation, where histories are only useful building blocks for constructing reliable logical relations between different experimental outcomes at different times, is also possible and probably a lot less contentious.

Griffiths's realist interpretation is still distinct from a standard hidden variable theory in important ways.
i) In CH, all variables are accessible by experiment, and all variables are the typical variables of standard QM/QFT (we don't need an additional state space where some physical state, distinct from the quantum state, resides). If we want to reveal spin-z, we measure spin-z. If we want to reveal spin-x, we measure spin-x. We cannot measure both spin-x and spin-z, but there is no subspace corresponding to "both spin-x and spin-z". It's not that it is hidden. There is nothing there to measure, as there is no framework that gives sense to "both spin-x and spin-z"
ii) The realism in CH is local ( https://arxiv.org/abs/0908.2914 ), while your standard hidden variable theory is nonlocal .

Gell-Mann has published an extended probability formalism which has an immediate realist interpretation ( https://arxiv.org/abs/1106.0767 ), and does invite comparison with hidden variables:
"Is this in effect a hidden variable theory? There are no variables involved beyond the usual quantum fields of sum-over-histories quantum theory — the {q(t)}. However their fine-grained values are not completely accessible to experiment or observation and therefore partially hidden"

 

[Fervent Freyja]

They were. https://en.wikipedia.org/wiki/Term_logic

I know you probably already know this. But for those who don’t: Plato first introduced the idea of using universals in his attempts to understand reality. His idea was to find something in nature that had unchanging properties. He believed that finding an unchanging property of nature was the only place that we could start if we wanted real knowledge about our reality. His idea of constants and universals was passed onto physics later. That’s what philosophy does and is supposed to do- it hands ideas over to the sciences and humbly accepts that it will never receive credit for it.

Also, Bertrand Russell and Alfred Whitehead were almost successful in describing much of mathematics using logic alone. The father of logic himself considered logic a branch of philosophy. There are probably a thousand more examples of where philosophy handed over ideas to the sciences. I am too tired to even try to argue this any further right now.

Edit: Descartes invented analytical geometry and introduced skepticism as an essential part of the scientific method. He is considered the father of philosophy.

 

[gentzen]

... the concept of "just take the theory seriously" and intriguing science-fiction concept of parallel worlds ...

... this controversial interpretation in 2021 seems only rivaled by orthodoxy ...

Many prominent MWI proponents like Sean Carroll, Bruce DeWitt, or John Wheeler simultaneously had interests in cosmology, general relativity and quantum theory. I believe that the orthodoxy (Copenhagen) doesn't even claim to be applicable to that combination. For the early universe, a splitting of worlds doesn't appear overly implausible or science-fiction to me, after all the entropy was probably nearly zero back then, at least compared to today.
I am not keen on using the MWI imagery to discuss Schrödinger's cat or Wigner's friend. A cat is not described by a pure wavefunction, and neither is Wigner's friend. But thinking about the Stern-Gerlach experiment or specific quantum computing setups in terms of MWI could be valuable, if the trap of too trivial explanations is avoided. (MWI did help David Deutsch to do groundbreaking work on quantum computing in 1985 and 1992.)
Sean Carroll's work on finding gravity inside QM as an emergent phenomenon goes back again to the roots of MWI. I find this work interesting, among others because Lienhard Pagel proposed that the Planck constant will remain the same even for emergent phenomena (in a non-relativistic context). But when time and energy (or length and momentum) are themselves emergent, what should it even mean that the Planck constant will remain the same.

... when you inquire about specifics it turns out that they all disagree on what platonism even means. In this spirit of confusion I'd love to hear what the thoughts on Many Worlds are in 2021 by everyone here at PF.

Everyone here at PF also includes me. As indicated above, what Sean Carroll tries to do with MWI got me interested, in his view on MWI specifically, and on how it relates to the views of other current MWI proponents. (My plan was to study the views by SEP authors Vaidman and Barrett, by decoherence experts Schlosshauer and Zeh, and by MWI experts Wallace and Zurek. Let me exclude Barrett, Schlosshauer, and Zurek, because I didn't find enough time yet to study their views.) Zeh and Wallace are quite careful and diplomatic, they avoid to make unjustified claims themselves, but are reluctant to distance themselves clearly from less careful MWI proponents. Vaidman seems to be the one who accepted the unthankful task to openly admit where other proponents oversimplified too much, or made claims that have been justly criticized.

Carroll's book "Something Deeply Hidden" has been reviewed by Sabine Hossenfelder, who later explained her two biggest objections in nice detail in The Trouble with Many Worlds. Vaidman already admited in 2002 (or even earlier) that both objections are somewhat justified:

... in his book, Barrett 1999, uses the name “MWI” for the splitting worlds view publicized by De Witt 1970. This approach has been justly criticized: it has both some kind of collapse (an irreversible splitting of worlds in a preferred basis) and the multitude of worlds.

The recent revival of this claim involving decision theory, Deutsch 1999, 2012, and some other symmetry arguments Zurek 2005, Sebens and Carroll 2018 also encountered strong criticisms (see Section 4.3) which might be perceived as criticisms of the MWI itself. Whereas the MWI may have no advantage over other interpretations insofar as the derivation of the Born rule is concerned, Papineau 2010 argues that it also has no disadvantages.

(Of course, Vaidman's admission in 2002 used different words: "All these results can be derived in the framework of various interpretations and thus the success or failure of these proofs cannot be an argument in favor or against the MWI. The MWI, like all other interpretations, requires a probability postulate." But the substance was the same.)

If prominent MWI proponents continue to present it in a way that invites justified objections, then this risks to damage the reputation of MWI and foundational studies more generally (in the long run). Let me be clear that Vaidman, Wallace, and Zeh are sufficiently careful from my POV to avoid that trap. Maybe it was partly also the fault of the critics, who failed to convince MWI proponents to take their objections seriously. Maybe the elaborations in Decoherence is Dephasement, Not Disjointness will change that slightly dangerous state of affairs. (From my POV, those elaborations are easy to follow and correct, except for the claim "This t_0 is approximately the mean free path time.")

MWI continues to be an inspiration for some people working on foundational topics. Other people are influenced by ideas and imagery developed in the context of MWI research, independent of whether they are aware of this or not. And David Wallace simply seems to do excellent foundational work on emergence, independent of whether he brands it as MWI research or not. Therefore, I accept that MWI is here to stay, but I am not one of its proponents. I prefer to think in terms of the statistical operator, and not in terms of a pure wavefunction.

 

[PeroK]

That’s what philosophy does and is supposed to do- it hands ideas over to the sciences and humbly accepts that it will never receive credit for it.

It depends how much of the job the philosophy achieves. A few years ago there was someone who posted on here claiming that medical research was "applied philosophy". This is the alternative view: that all scientific progress is directly related to what philosophers do.

It's impossible to reimagine the world today without the ancient philosophers. Equally, it's impossible to say how much of the credit for the revolutions in science, mathematics, heavy engineering, telecommunications, genetics etc. they and other philosophers deserve.

My point is simply that the real advances came largely when engineers, scientists and mathematicians developed their subjects independent of any direct intervention from contemporary philosophical thought.

 

[jbergman]

Despite its lackluster reception at its conception by Hugh Everett and subsequent advocacy by Bryce S. DeWitt the concept of "just take the theory seriously" and intriguing science-fiction concept of parallel worlds eventually gave it a major resurgence, much thanks to David Deutsch's pioneering work on the concept of quantum computation and militant advocacy of Everett. In the 90s the few people who refused to shut up and calculate started digging deeper into its intrinsic postulate and seductive simplicity and then in the mid-2000s the Everettian Relative State / Many Worlds Interpretation became mainstream as a consequence of the Oxford nucleus of proponents; David Deustch, Simon Saunders and David Wallace's work in the early to mid 2000s.

Now some 15 years later, hundreds of papers reference the interpretation yearly, tens of best selling books has gone into its depths and numerous high-profile sci-fi series has made it a pop-culture staple. Once left in the dustbin this controversial interpretation in 2021 seems only rivaled by orthodoxy and the more pragmatic "shut-up and calculate" crowd.

A lot of people have a visceral reaction, either favorably or negatively when confronted with the idea of wavefunction fundamentalism and its implied infinitely splitting worlds. This 'controversy' has just increased the frequency of discussion.

The promise of Everett was simple, in fact it was simplicity itself, in some sense it was a non-interpretation, at least that was its claim. Despite its elegant allure, now even ~70 years after its birth, the there are so many different Many Worlds Interpretations that they are hard to keep track of. Some proponents advocate for the view that the wavefunction is fundamental, other prominent advocates such as David Wallace and Christopher Timpson counter with more complicated views of Spacetime State Realism others yet argue over how to interpret probability and add axioms to justify their Born Rule derivation, then you have the more exotic perspectives of "Many Minds" (taken seriously by decoherence's father in Dieter Zeh and used as a counterpoint to Many Worlds by the likes of David Albert), and recently we've seen blends of Bohmian Mechanics and Many Worlds in the shape of several different "Many Interacting Worlds", then if you really want to go all-in you have prominent cosmologists like Leonard Susskind and others proposing that in fact Many Worlds is just another way to interpret the spatially separating Multiverse predicted by Eternal Inflation and String Theory.

As someone who's been following its developments for the past 17 years closely I am left feeling that "Many Worlds" is similar to Platonism. You have a lot of mathematicians who adhere to some form of Platonism, but when you inquire about specifics it turns out that they all disagree on what platonism even means. In this spirit of confusion I'd love to hear what the thoughts on Many Worlds are in 2021 by everyone here at PF.

Late to the party, but as I understand it, MWI, cannot explain the measurement probabilities we see when we repeat a state preparation experiment. Deustch's proposal for how to recover these probabilities is problematic and I think not widely accepted. I'm not sure this has been fixed even thought this interpretation is widely accepted.

 

[Demystifier]

Edit: Descartes invented analytical geometry and introduced skepticism as an essential part of the scientific method. He is considered the father of philosophy.

Did you mean father of science?

 

[Fervent Freyja]

Did you mean father of science?

I’ve never heard of anyone being called the father of science. I doubt it would be him. Only know that he is widely regarded as the father of modern philosophy. A portion of a basic introductory philosophy course is usually devoted to him.

In Meditations he introduces the method of doubt and insists that it must be used as groundwork in establishing truths. This attitude of skepticism was essential for any of the sciences to get off the ground and start impacting the world. It’s difficult to trace who influenced what. There are signs that Newton studied and drew from some of his work, but from what I know he never acknowledges that Descartes influenced him.

 

[Demystifier]

I’ve never heard of anyone being called the father of science.

Galileo? https://en.wikipedia.org/wiki/List_...ther_of_a_scientific_field#Science_as_a_whole

 

[PeroK]

Galileo? https://en.wikipedia.org/wiki/List_...ther_of_a_scientific_field#Science_as_a_whole

I wonder who the mother was?

 

[Demystifier]

I wonder who the mother was?

Science is not created by sexual reproduction, so the existence of father does not imply the existence of mother.

 

[hutchphd]

Galileo? https://en.wikipedia.org/wiki/List_...ther_of_a_scientific_field#Science_as_a_whole

I am astonished that William Gilbert appears nowhere in this list (did I miss him?)

 

[*now*]

CH as a formalism is itself open to different interpretations. A realist interpretation is most commonly put forth by Robert Griffiths ( https://arxiv.org/abs/quant-ph/0001093 ), but an antirealist interpretation, where histories are only useful building blocks for constructing reliable logical relations between different experimental outcomes at different times, is also possible and probably a lot less contentious.

Griffiths's realist interpretation is still distinct from a standard hidden variable theory in important ways.
i) In CH, all variables are accessible by experiment, and all variables are the typical variables of standard QM/QFT (we don't need an additional state space where some physical state, distinct from the quantum state, resides). If we want to reveal spin-z, we measure spin-z. If we want to reveal spin-x, we measure spin-x. We cannot measure both spin-x and spin-z, but there is no subspace corresponding to "both spin-x and spin-z". It's not that it is hidden. There is nothing there to measure, as there is no framework that gives sense to "both spin-x and spin-z"
ii) The realism in CH is local ( https://arxiv.org/abs/0908.2914 ), while your standard hidden variable theory is nonlocal .

Gell-Mann has published an extended probability formalism which has an immediate realist interpretation ( https://arxiv.org/abs/1106.0767 ), and does invite comparison with hidden variables:
"Is this in effect a hidden variable theory? There are no variables involved beyond the usual quantum fields of sum-over-histories quantum theory — the {q(t)}. However their fine-grained values are not completely accessible to experiment or observation and therefore partially hidden"

I think there seems to be the suggestion of a special status for some systems among other systems and I prefer interpretations concerned with maintaining equivalence of all systems. If so, regarding IGUS, is there a definition that answers questions like whether a system’s status is subject to change? (E.g., were Gell-Man’s frog suffering brain damage and losing calculating skills, would it lose the status to some extent or altogether?) Also, do these articles suggest a knowledge or mind dependent reality?

 

[Morbert]

I think there seems to be the suggestion of a special status for some systems among other systems and I prefer interpretations concerned with maintaining equivalence of all systems. If so, regarding IGUS, is there a definition that answers questions like whether a system’s status is subject to change? (E.g., were Gell-Man’s frog suffering brain damage and losing calculating skills, would it lose the status to some extent or altogether?) Also, do these articles suggest a knowledge or mind dependent reality?

For clarification, could expand on what you mean by special status for some systems?

 

[*now*]

From the Gell-Mann, 2011 paper you linked on page 5, section III, “Settleable bets, records, and decoherence”, there is a discussion of (some?) systems that might include frogs, ...”Information gathering and utilizing systems (IGUSes) like ourselves exploit the regularities summa- rized by physical theories to construct schemata, make predictions, and direct behavior [3]. Even a frog aiming to catch a fly can be said to be exploiting a rudimentary classical approximation to quantum mechanics (perhaps hard wired). The frog is in effect betting on these regularities”

 

[Morbert]

From the Gell-Mann, 2011 paper you linked on page 5, section III, “Settleable bets, records, and decoherence”, there is a discussion of (some?) systems that might include frogs, ...”Information gathering and utilizing systems (IGUSes) like ourselves exploit the regularities summa- rized by physical theories to construct schemata, make predictions, and direct behavior [3]. Even a frog aiming to catch a fly can be said to be exploiting a rudimentary classical approximation to quantum mechanics (perhaps hard wired). The frog is in effect betting on these regularities”

I'm not sure if this answers your question but: From the decoherent histories perspective, if we want a quantum theory of our entire universe that includes a description of systems like frogs, we need the triple $(\rho,H,\{C_\alpha\})$. The terms $\rho$ and $H$ are the initial conditions of our universe and its dynamics respectively. These make up the core of our physical theory. $\{C_\alpha\}$ are a set of mutually exclusive alternative histories of our universe, each assigned a probability computed from $\rho, H$. There is great freedom in how we construct $\{C_\alpha\}$, and if we want to discuss frogs, we should construct histories of quasiclassical variables (i.e. averages of conserved quantities over small volumes) that make up systems like frogs. Let's call this set of histories $\mathcal{F}$. In this framework, we can identify frogs, and the environmental variables they correlate with (like flies, predators, weather, etc). What Gell-mann says is we can understand the behaviour of systems like frogs in terms of bets. E.g. A frog bets (instinctively), that if it does not eat flies, it will die, and our quantum theory could be used to show that this is true. Histories containing frogs that don't eat flies and live a long time will have a probability of almost 0. Our quantum theory supplies the reliable odds that systems like frogs or humans should adhere to for their self-interest.

But these frogs are not special as far as the theory is concerned. An alternative set of histories $\mathcal{F}'$ might not have any quasiclassical variables. No quasiclassical variables, no planets, no rain forests, no frogs etc. This alternative set is not mutually exclusive with $\mathcal{F}$. It's just that it's not suitable for the purposes of describing systems like frogs. Both $\mathcal{F}$ and $\mathcal{F}'$ are equally valid, and our choice of a set does not elevate that set to some privileged ontic status.

All Gell-Mann's EPE formalism in that paper really does is let us relate $\mathcal{F}$ and $\mathcal{F}'$ by a common refinement of fine-grained histories, each of with are assigned an extended probability by our theory.

 

[Demystifier]

our choice of a set does not elevate that set to some privileged ontic status.

My (and not only my) main problem with CH interpretation is that it is not clear what does elevate some set to a privileged ontic status. And if nothing does (or if only measurement does), then is CH interpretation any more ontic/realistic than the "standard" Copenhagen interpretation?

 

[gentzen]

is not clear what does elevate some set to a privileged ontic status

I would say that Roland Omnès simply accepts that CH is an incomplete interpretation. The advantage over the "standard" Copenhagen is a greater logical clarity, not being more ontic per se. But if you would decide to arbitrarily declare some set of histories (for example those based on particle positions) to be more ontic than others, then the resulting interpretation would still be consistent, and be more ontic than Cohenhagen.

 

[Morbert]

My (and not only my) main problem with CH interpretation is that it is not clear what does elevate some set to a privileged ontic status. And if nothing does (or if only measurement does), then is CH interpretation any more ontic/realistic than the "standard" Copenhagen interpretation?

That's the beauty. None are ever elevated. The use of one set over another is a matter of which set is fit for purpose, rather than which set is real. Quoting Robert Griffiths

"The choice of family is up to the physicist, and will generally be made on the grounds of utility, e.g., how to model a particular experimental situation with apparatus put together by a competent experimentalist. This choice has no influence upon what really goes on in the world, and alternative choices applied to the same set of data and conclusions will yield consistent results. The quantum world is of such a nature that it can be described in distinct ways which (in general) cannot be combined into a single all-encompassing description"

I would say that Roland Omnès simply accepts that CH is an incomplete interpretation. The advantage over the "standard" Copenhagen is a greater logical clarity, not being more ontic per se. But if you would decide to arbitrarily declare some set of histories (for example those based on particle positions) to be more ontic than others, then the resulting interpretation would still be consistent, and be more ontic than Cohenhagen.

Would you have a source re/ any thorough discussion of the completeness of CH by Omnes. I'm quite familiar with his early work ("Interp. of QM" and "Undestanding QM") but I have not read Quantum Philosophy.

 

[gentzen]

Would you have a source re/ any thorough discussion of the completeness of CH by Omnes. I'm quite familiar with his early work ("Interp. of QM" and "Undestanding QM") but I have not read Quantum Philosophy.

"Interp. of QM" was too technical for me, but I am familiar with "Understanding QM" (I didn't read it cover to cover, but I did read many parts and had the impression that I understood those parts). Much later I read "Quantum Philosophy" cover to cover. Two surprising claims from "Understanding QM" no longer occurred, but I will have to lookup specifics later at home.

In "Quantum Philosophy," R. Omnès is pretty clear that he is convinced that there is only a single world, even if CH doesn’t explain it, and cannot even disprove Many-Worlds. His defense is to admit that there is still a disagreement left between Reality and quantum theory, but that it would be hubris to expect otherwise. At least that is how I interpret his words on page 214:

… have reproached quantum physics for not explaining the existence of a unique state of events. It is true that quantum theory does not offer any mechanism or suggestion in that respect. This is, they say, the indelible sign of a flaw in the theory, … Those critics wish at all costs to see the universe conform to a mathematical law, down to the minutest details, and they certainly have reason to be frustrated.
…
I embrace, almost with prostration, the opposite thesis, the one proclaiming how marvelous, how wonderful it is to see the efforts of human beings to understand reality produce a theory fitting it so closely that they only disagree at the extreme confines. They must eventually diverge, though; otherwise Reality would lose its nature proper and identify itself with the timeless forms of the kingdom of signs, frozen in its own interpretation. No, science’s inability to account for the uniqueness of facts is not a flaw of some provisional theory; it is, on the contrary, the glaring mark of an unprecedented triumph. Never before has humanity gone so far in the conquest of principles reaching into the heart and the essence of things, but that are not the things themselves.

 

[Demystifier]

That's the beauty. None are ever elevated. The use of one set over another is a matter of which set is fit for purpose, rather than which set is real. Quoting Robert Griffiths

"The choice of family is up to the physicist, and will generally be made on the grounds of utility, e.g., how to model a particular experimental situation with apparatus put together by a competent experimentalist. This choice has no influence upon what really goes on in the world, and alternative choices applied to the same set of data and conclusions will yield consistent results. The quantum world is of such a nature that it can be described in distinct ways which (in general) cannot be combined into a single all-encompassing description"

Inspired by this any by what I already knew about CS, let me try explain the motivation for CH in my own words:

CH is not about what really goes on in the world. Instead, it is about what we can cay about the world, if we require that the things we say obey the following rules:
- It is logically consistent and precise. (Unlike some vague variants of Copenhagen and statistical ensemble interpretations.)
- We don't need any additional mathematical objects except those which are already there in standard QM. (Unlike Bohmian trajectories or GRW modifications of the Schrodinger equation.)
- There is no a priori preferred basis in the Hilbert space.
- Measurement and observation do not play any fundamental roles.
- There are no many worlds.

Would you agree?

 

[*now*]

I'm not sure if this answers your question but: From the decoherent histories perspective, if we want a quantum theory of our entire universe that includes a description of systems like frogs, we need the triple $(\rho,H,\{C_\alpha\})$. The terms $\rho$ and $H$ are the initial conditions of our universe and its dynamics respectively. These make up the core of our physical theory. $\{C_\alpha\}$ are a set of mutually exclusive alternative histories of our universe, each assigned a probability computed from $\rho, H$. There is great freedom in how we construct $\{C_\alpha\}$, and if we want to discuss frogs, we should construct histories of quasiclassical variables (i.e. averages of conserved quantities over small volumes) that make up systems like frogs. Let's call this set of histories $\mathcal{F}$. In this framework, we can identify frogs, and the environmental variables they correlate with (like flies, predators, weather, etc). What Gell-mann says is we can understand the behaviour of systems like frogs in terms of bets. E.g. A frog bets (instinctively), that if it does not eat flies, it will die, and our quantum theory could be used to show that this is true. Histories containing frogs that don't eat flies and live a long time will have a probability of almost 0. Our quantum theory supplies the reliable odds that systems like frogs or humans should adhere to for their self-interest.

But these frogs are not special as far as the theory is concerned. An alternative set of histories $\mathcal{F}'$ might not have any quasiclassical variables. No quasiclassical variables, no planets, no rain forests, no frogs etc. This alternative set is not mutually exclusive with $\mathcal{F}$. It's just that it's not suitable for the purposes of describing systems like frogs. Both $\mathcal{F}$ and $\mathcal{F}'$ are equally valid, and our choice of a set does not elevate that set to some privileged ontic status.

All Gell-Mann's EPE formalism in that paper really does is let us relate $\mathcal{F}$ and $\mathcal{F}'$ by a common refinement of fine-grained histories, each of with are assigned an extended probability by our theory.

This helps a lot, thank you

 

[Morbert]

Inspired by this any by what I already knew about CS, let me try explain the motivation for CH in my own words:

CH is not about what really goes on in the world. Instead, it is about what we can cay about the world, if we require that the things we say obey the following rules:
- It is logically consistent and precise. (Unlike some vague variants of Copenhagen and statistical ensemble interpretations.)
- We don't need any additional mathematical objects except those which are already there in standard QM. (Unlike Bohmian trajectories or GRW modifications of the Schrodinger equation.)
- There is no a priori preferred basis in the Hilbert space.
- Measurement and observation do not play any fundamental roles.
- There are no many worlds.

Would you agree?

Yes I think those statements are fair. The only one that might need to be sharpened a little is "CH is not about what really goes on in the world. Instead, it is about what we can cay about the world". If by this we mean that no specific ontology is an essential premise of, or determined by, the formalism then yes (even if some founders like Griffiths put forth an ontology). CH, as a minimum project, generalises the procedure for computing reliable probabilities and logical inferences so that it can be applied to closed systems without recourse to some external environmental context.

 

[Demystifier]

... (even if some founders like Griffiths put forth an ontology). CH, as a minimum project ...

So by "minimum project" you mean CH which does not put forth an ontology, am I right? Or are there also other ways how CH can be non-minimal?

 

[Quantumental]

It sounds to me like CH is then by default just a more elaborate "Shut up and calculate using these tools" rather than an actual interpretation of QM

 

[gentzen]

It sounds to me like CH is then by default just a more elaborate "Shut up and calculate using these tools" rather than an actual interpretation of QM

Would you also say that Copenhagen is not an actual interpretation of QM? And what is wrong with "... calculate ..."? All those calculations felt quite positive to me on my first encounter with CH, because they gave me a nice illusion of understanding:

In 1998, I had to endure a QM 1 course at university, and I didn’t manage to connect at all to QM. ... I couldn’t create a picture or film in my head of how to use this to describe nature. ... Occasional attempts to read material discussing interpretation of QM failed quite early, I couldn’t penetrate into the material and words at all. Around 2005, I read (or rather browsed) “Understanding Quantum Mechanics” by Roland Omnès, and it was the first time that I felt that the material was presented in a way that I would understand it, if I invested the time to work through it. It felt like “let me calculate and explain” as opposed to “don’t ask questions, nobody understands QM anyway”.

 

[Morbert]

So by "minimum project" you mean CH which does not put forth an ontology, am I right? Or are there also other ways how CH can be non-minimal?

Yes, in this convo, by minimum I just mean the development of the formalism without commitment to an ontology. So e.g. if we take a typical case of a microscopic system $s$ and a measuring device $M$ that measures observable $A = \sum a_i \Pi_{a_i}$ with pointer states $\{M_i\}$, the conventional approach would be to write down a Hilbert space for the microscopic system, and to compute the probability of a measurement result $a_i$ as $$p(a_i) = \mathrm{Tr}\left[\Pi_{a_i}(t)\rho_s\right]$$A consistent historian would be perfectly comfortable expanding the Hilbert space to include the measurement device and computing the probability for the result $a_i$ and immediately after, the pointer outcome $M_i$ $$p(M_i\land a_i) = \mathrm{Tr}\left[\Pi_{M_i}(t+\delta t)\Pi_{a_i}(t)\rho_s\otimes\rho_M\Pi_{a_i}(t)\right]$$They could then go further and show that $p(a_i) = p(M_i) = p(M_i\land a_i)$ which establishes the logical relation $a_i\iff M_i$ which identifies a measurement scenario in the interaction between $M$ and $s$.

What is normally an implicit measurement context is made explicit by the consistent historian. They can do all this without committing to $a_i$ as a real property that exists before (or after) measurement. They can limit their ontic commitment to $M$ if they like.

 

[Demystifier]

What is normally an implicit measurement context is made explicit by the consistent historian. They can do all this without committing to $a_i$ as a real property that exists before (or after) measurement. They can limit their ontic commitment to $M$ if they like.

That's very much similar to my solipsistic hidden variables, where one can choose objects for which to postulate the ontic Bohm-like trajectories: everything including the measured system, or the apparatus but not the measured system, or perhaps only the state of brain responsible for consciousness. And of course, all this has roots in von Neumann theory of measurement where one has similar freedom where to put the collapse postulate.

 

[gentzen]

A consistent historian would be perfectly comfortable expanding the Hilbert space to include the measurement device and computing the probability for the result $a_i$ and immediately after, the pointer outcome $M_i$
$p(M_i\land a_i)=\ldots$

The formula for $p(M_i\land a_i)=\ldots$ looks wrong: $\Pi_{a_i}(t)$ appears two times in the formula, and the rightmost occurence feels like a typo to me.

 

[Morbert]

The formula for $p(M_i\land a_i)=\ldots$ looks wrong: $\Pi_{a_i}(t)$ appears two times in the formula, and the rightmost occurence feels like a typo to me.

If I work in an explicit history space $\mathcal{H}_t\otimes\mathcal{H}_{t+\delta t}$ the rightmost term is not necessary. But otherwise, histories made up of chains of projectors like $\Pi_{M_i}(t+\delta t)\Pi_{a_i}(t)$ are not necessarily projectors themselves and so it is good practice to avoid any time-ordering commutation issues. This is why you will see authors like Omnes (in "Understanding QM") write the probability of a history like so

as opposed to just $p(a) = \mathrm{Tr}(C_a E_0)$

 

[Minnesota Joe]

Such a view is rather obsolete. The quantum potential is as obsolete as e.g. relativistic mass in modern formulation of special relativity. Wave function is fundamental but not ontic, in the same sense in which Hamiltonian in classical mechanics is fundamental but not ontic.

I find it really difficult to understand how the configuration space wavefunction could be real (not that my difficulties are an argument for anything), so I was intrigued to find this paper by Norsen, Marian, and Oriols (2015, Synthese, or https://arxiv.org/abs/1410.3676) that sounds a little like your description. They argue that the single particle conditional wave functions are real but the configuration space wave function is not; instead, it plays a role analogous to the role the Hamiltonian plays in classical mechanics (the part similar to what you wrote). Is this the kind of approach you are hinting at or something else?

Thanks,

Joe

 

[PAllen]

I think most people would consider logic and inductive reasoning to be general tools applicable to a variety of disciplines, not philosophy. If they are part of any particular discipline, I think most people would say that discipline is mathematics.

Well, most universities, at least when I went, offered formal logic as part of philosophy. That was how some geeks got to take a “math” course for a humanities distribution requirement. And many of the most eminent figures in the history of logic (e.g. Quine), certainly considered themselves philosophers.

 

[Demystifier]

Is this the kind of approach you are hinting ... ?

Yes.