Last Monday, Murray Gell-Mann celebrated his 85th birthday: congratulations!
As far as I remember, I have never published a blog post that would be primarily dedicated to Gell-Mann's comments about the foundations of quantum mechanics. So this is the first time. The 17-minute-long video monologue above was taken from multi-hour interviews with him (and analogously, many others) on the "Web of Stories".
Along with Jim Hartle, Gell-Mann is the co-author of their "strongly decoherent" version of the consistent histories, an approach to the foundations of quantum mechanics. For years, I would quote it as "my #1 favorite interpretation" of quantum mechanics. I still like it. On the other hand, I think that the very fact that one chooses similar "new interpretations" has helped to spread the misconception that there exists a legitimate business of "building new interpretations" because the original formulation of quantum mechanics is "wrong" in some way and the "consistent histories" is just one of them (and needless to say, the hopelessly wrong, intrinsically anti-Copenhagen interpretations always preserve their majority status).
In reality, there is nothing wrong with the original postulates of quantum mechanics – the "Copenhagen interpretation", as some people call them – and the "consistent histories" are nothing else than a way to optimize the Copenhagen interpretation to questions that involve several measurements at different moments rather than just one measurement. I wouldn't claim that the founders of quantum mechanics didn't know what to do with a sequence of measurements, however.
Because the specific features of "consistent histories" have pretty much nothing to do with the fundamental issues that make it hard for so many philosophers, people, and even physicists to understand and "accept" quantum mechanics, I would probably no longer choose to emphasize the "consistent histories" because their added value is very limited and much less "essential" relatively to the change of the basic paradigm that was brought by the quantum revolution.
But back to the monologue by Gell-Mann which I endorse almost entirely, at least 95% of it.
Gell-Mann says that he would work on the foundations of quantum mechanics in 1963 and 1964 and this thinking continued throughout his life although he would only start to publish papers about the foundations in the 1980s. In the 1960s, Gell-Mann and perhaps Feynman and another physicist Felix Villars [of the Pauli-Villars regularization fame; he became a pioneer of biophysics] essentially agreed about the fundaments of quantum mechanics – which were not really very different from Gell-Mann's more recent work with Hartle.
He talks about Everett. That guy wasn't passionate about quantum mechanics, just about problem solving, Gell-Mann said. Quantum foundations were just one of these problems and Gell-Mann suggests that the weapons problems he would be solving for most of his life were at least equally interesting for Everett.
At any rate, Gell-Mann says that back in the 1960s when his modern picture of quantum mechanics was pretty much born, he and Felix Villars didn't know about Everett's work. I emphasize this because some people love to paint Everett as the ultimate source of all new ideas about quantum mechanics like decoherence etc. So just to be sure, there is at least one Nobel prize winner who wrote some of the most sensible papers about the foundations of quantum mechanics who rejects the idea that he was building on Everett's work. Maybe, subconsciously, some ideas of Everett could have gotten to Gell-Mann's head through Feynman although he doesn't really have any evidence for that speculative possibility. ;-)
So I really do think that the evidence indicates that Hugh Everett has actually inspired only those who remained deeply confused about quantum mechanics.
Gell-Mann spends several minutes by arguing that the feature of Everett's ideology that there are "many worlds that are equally real" is operationally meaningless. The comment may only mean that the theory treats the possible alternative histories on equal footing, except for their generally different probabilities. But only one of those actually takes place in the "real", experience-based sense of the word "actually". ;-) I completely agree with Gell-Mann. There is absolutely no other physics behind the claims that the alternative histories are "real". At most, it is a new meaningless salvo arguing for a particular interpretation of the word "real" or a paradigm that wants you to imagine how the alternatives look from a meta-observer's viewpoint. But we are not such meta-observers (who "see" un-realized outcomes as well) and because physics is about things we can detect, hypothetical observations of something from an entirely different agent's viewpoint just don't belong to physics.
After 3:00, Gell-Mann mentions Roland Omnes and Robert Griffiths as guys who kickstarted decoherent histories etc. in the late 1980s so they "scooped" Gell-Mann although he had known these things for decades. He explains that others preferred to study the "minimal conditions" and "minimal amounts" of decoherence while he and Hartle always cared about the opposite extreme, the very "strong decoherence" and "lots of decoherence" because in the real world of classical observers, we never really experience any "shortage of decoherence".
Gell-Mann talks about the quantum mechanics' freedom to change the representations at each moment – plus to change the degree of coarse-graining – so the number of potential realms is gargantuan. In practice, we use the "hydrodynamic" limit – histories are defined by intervals of observables which are nothing else than volume integrals of (approximately or exactly) conserved densities. In this quasiclassical limiting "realm" or treatment of the histories, we want this description to be fine enough (if not "maximally fine") so the volumes shouldn't be too large but these volumes can't be too small because we demand some equilibrium in those volumes. The word "quasiclassical" means that the classical equations of motion are a good zeroth approximation for the behavior of the observables (fluctuations, branchings etc. may be added as small corrections).
This picture allows one (as Hartle wrote in some papers) to include the gravitational fields – with large quantum fluctuations of the metric – to the formalism that may define the histories. So Gell-Mann suggests that in this quantum-gravity regime, the "decoherent histories" picture may be giving us something new, a generalization of quantum mechanics. At this point, Gell-Mann somewhat interestingly suggests that the "consistent histories" and "Feynman's sum over histories" are morally the same thing so Feynman's path-integral approach to quantum mechanics could be a "generalization" of quantum mechanics, after all – Feynman's "completely new theory" that he always wanted instead of working on theories initially designed by someone else.
After 9:30, Gell-Mann is asked whether quantum mechanics is taught correctly. When the sum-over-histories as well as (especially) the decoherence histories are perfected, they should perhaps be taught. He says that the Copenhagen picture is correct for all contexts but it is not "convincing". It's hopeless for quantum cosmology.
Here I would say, right, *we* can't prepare many copies of the Universe and measure the probability by repetitions etc. But note that the previous sentence contains the word *we*. It's a limitation of *ours*, not a limitation of the theory. A theory may predict probabilities of events in cosmology that only occurs once but it's clear that we can't measure those probabilities in any quantitative way and no improvement of a theory will ever change that!
There are clearly things happening without human observers, he says. Well, right, if one puts it in this way. However, there are no properties of the events in Nature that humans know without human observers! ;-) The previous sentence is a tautology and it's really the only thing one needs to accept that the fundamental description of physics deals with observers. Without observers, no one can know what the events were so he or she or it or they cannot think about them rationally!
OK, you could sense a slight degree of disharmony between Gell-Mann's words and your humble correspondent's emphasis here.
However, that changes totally after 11:50 when Gell-Mann starts to talk about the "foolishness" often associated with the entanglement ("Einstein-Podolsky-Rosen-Bohm effect", using his words). He treats this issue at some length in his book; I hope he meant The Quark and the Jaguar.
OK, where did the "foolishness" come from? Gell-Mann says that the bulk of John Bell's work was right but he introduced words that were prejudicial such as "nonlocal". People often say that there is something nonlocal about the EPR phenomena but the only correct similar statement that they could mean, Gell-Mann emphasizes (and I often do, too) is that a classical interpretation of what is happening would require nonlocality (or negative probabilities). But the world is not classical, and no nonlocality is needed because the world is quantum mechanical. As far as Gell-Mann can tell, it's like giving a bad name to a dog and sticking with it.
When the quantum mechanical predictions for Bell's experiment were experimentally verified, Gell-Mann would have expected everyone to scream "Great, we can go home!". Instead of this rational reaction, people would start to say that "there is something seriously peculiar". But the only thing that is seriously peculiar about all of this is the new foundation of physics discovered in the 1920s – quantum mechanics.
As Gell-Mann described in that book, the entanglement between particular polarizations of two photons is no different from Bertlmann's socks (real guy!) and not strange or requiring any nonlocality. Quantum mechanically, the character of the correlation may apply to many types of measurements, so quantum mechanics allows deeper correlations. But the conceptual interpretation is still the same as in the mundane classical case of the socks and no nonlocality is needed. People say that the measurement on one side has to do something to the twin photon. But it doesn't do anything! The only thing that happens is that you measure one property and you also learn the corresponding property of the other particle.
The people "who want to confuse us" sometimes mention that "we choose which property of the first particle we measure", and therefore we affect what kind of states the other particle in the pair may acquire. But the catch and point is that the different types of the measurement (e.g. the linear polarization vs circular polarization) can't take place simultaneously on the same branch of the history. They are on different branches, decoherent from each other, and only one of them may occur. So it's simply not true that we are nonlocally influencing something in our world. In a particular branch of the history, we made some particular decisions "what observable is going to be measured" and some corresponding basis of the Hilbert space is therefore more useful than others – and the information we learn about the other photon is no stranger than the information about the other classical Bertlmann's sock.
And, as Murray added, Einstein's original thesis that if a quantity has a potential to be measured with certainty, then it has to be "real" and have some objective real properties at all times, is simply wrong! It directly contradicts quantum mechanics. If two quantities' commutator is nonzero, non-intrusive measurements of these two quantities could only be performed on different branches of the history. And that's all there is to it.
As he reminds us, Gell-Mann has presented it in his book and so did some other people but this simple point doesn't seem to get across, Gell-Mann complains (and so do I). People keep on being mesmerized by the confusing language involving "nonlocality". What they do isn't necessarily wrong – lots of people do correct stuff – but the vocabulary makes it sound as something totally different than what it is. I couldn't agree more.