Tuesday, May 03, 2016

Gell-Mann's opposition to Bohr has been mostly a pose

A few days ago, a computer recommended Steve Hsu to watch a video of Murray Gell-Mann we discussed in 2014. (See also a 2009 blog post on Gell-Mann.)

And Hsu has interpreted all these things in a way that Gell-Mann was a big critic of the Copenhagen interpretation. And maybe Feynman was one as well, Hsu adds. This is particularly disingenuous given Gell-Mann's well-known opinions about the kind of "superluminal" stuff that Hsu was recently promoting.

What's going on?

Bohmist Sheldon Goldstein has considered Gell-Mann one of the most sensible critics of orthodox quantum theory while Feynman was one of its most sensible defenders (see Page 19). Is this split meaningful, or is it just another manifestation of the Gell-Mann vs Feynman rivalry? I think it's mostly the latter.

Murray Gell-Mann has been often quoted as saying:
Niels Bohr brainwashed a whole generation of theorists into thinking that the job (interpreting quantum theory) was done 50 years ago.
Great. And this wisdom that the generation of physicists was learning and endorsing was actually correct, unlike the tons of junk in which we are drowning in 2016.

Niels Bohr was given a hard time by Murray. Gell-Mann surely wanted to define the basic rules of quantum mechanics more clearly than Bohr has – which is not quite impossible because Bohr was arguably not the ultimate role model for clarity (although he was right about all the big issues and he was actually responsible for formulating some of them).

But what kind of an interpretation Gell-Mann has favored – and is still favoring? He is one of the big workers on the Consistent Histories interpretation – something that I used to call my "preferred interpretation" as recently as some 15 years ago, before I fully realized what are the "actual dividing lines" in this bogus controversy.

However, before Gell-Mann and Hartle wrote the papers on the histories in the 1980s etc. (in the early 1990s, I was influenced by similar papers by Roland Omnes but that was partly a coincidence), Gell-Mann could have been influenced by Hugh Everett. Gell-Mann talked about Everett sort of positively in the video that Hsu embedded and I mentioned at the top. Interestingly, however, I think that Everett's name doesn't appear at all in Gell-Mann's 1994 popular book, "The Quark and the Jaguar" even though quantum mechanics is discussed on almost 100 pages.

Instead, the chapter "Quantum Mechanics And Flapdoodle" mainly focuses on David Bohm and especially the irrational "interpretations" of Bell's inequality. Most importantly, Gell-Mann criticizes the absolutely wrong claims about nonlocality (such as those in Mark Alford's paper) which came after the verification of the quantum mechanical predictions for the Bell's experiment, and despite this verification. There's just no nonlocality. And the incorrect claims about nonlocality have produced all the garbage about quantum telepathy and similar paranormal phenomena caused by the "weird" quantum mechanics in the newspapers.

The wrong claim about the nonlocality was also the main target of Sidney Coleman's lecture about quantum mechanics in your face. The explanation of the Bell's experimental results requires either quantum mechanics or nonlocality, not both. There's not even an interaction term, let alone a non-local one, in the right description of the EPR measurements. So there's obviously no nonlocal influence.

What about Gell-Mann's attitude towards Everett's ideas? It depends on what Everett's ideas are supposed to mean. He wrote a long thesis (about the "relative-state interpretation") with lots of useless stuff – and the main claim often assigned to Everett, namely the claim that there are "many worlds", was actually invented later by Bryce DeWitt and Everett pointed out that DeWitt's stuff was "bullšit".

So this "fuzzy package of ideas" is extremely vague. But many otherwise sensible physicists like to pay lip service to Everett. Some of them say manifestly wrong things. But some of them are really sensible and what they mean when they credit Everett are the following principles:
  1. Quantum mechanics applies to all objects in Nature, including the large ones (such as apparatuses and human beings)
  2. The measurement is preceded by interactions that produce quantum entanglement between the observer and the observed system
  3. The collapse associated with the measurement isn't an objectively real process (let alone one that could influence other parts of Nature superluminally)
And I haven't even mentioned a more complex idea, Bohr's complementarity, which we will look at later.

Now, it's ironic that Everett ever gets any credit for those points because all of them were invented by the "Copenhagen school". They knew that quantum mechanics applied to large objects as well – and the school immediately started to investigated lattices, crystals, solids, and macroscopic amounts of gases (Bose-Einstein and Fermi-Dirac statistics) using the new formalism of quantum mechanics in the 1920s. So they didn't just agree it worked for large objects; they also did quite some detailed work that builds on this assumption.

But the Copenhagen men also knew that in order to apply quantum mechanics, some "adult in the room" is needed. One needs an observer who observes things and may verify the laws of quantum mechanics, so at some moment, it's silly to describe everything (including all large objects) in terms of state vectors. The observer who ultimately applies the theory must assume that his own reasoning is composed of classical facts, so he must make the cut and say that things on his side basically follow the rules of classical logic. They knew that in any feasible application of quantum mechanics, the human feelings may be approximated as the classical facts, so the humans and their macroscopic machines may be assumed to be basically classical objects that interact with the observed microscopic systems (even though their precise working could be described by quantum mechanics as well – if they were observed by yet another external observer).

Also, the Copenhagen school knew about entanglement although the name was only coined by Schrödinger in 1935 (as a more moderate and less misleading synonym for Einstein's 1935 "spooky action at a distance") and they surely knew that the interaction of two objects – even if one of them is an apparatus – may be described by quantum mechanics and some entanglement between all the degrees of freedom emerges. And the Copenhagen school emphasized that the "collapse" isn't an objectively real process but a quantum counterpart/description of the "change of the observer's subjective knowledge" caused by the measurement.

This last point, that no information is really being sent, is primarily an opposition to Alford-Bricmont-style claims that there is a non-local influence caused by the collapse. There's none. The Bohmian and many-worlds "interpretations" are often presented as allies but they're only allies to the extent that they call themselves different than the "shared enemy", namely proper quantum mechanics. But from most perspectives, they're actually further from each other than either of them is from quantum mechanics.

If I simplify things even more, the sensible physicists who pay lip service to Everett today mostly want to say that "no ugly process such as a collapse" is needed at all. We're just being divided to branches. And that's why there's no non-local influence. They deny that "anything" special aside from the unitary evolution takes place during the measurement at all.

So this "innocently Everettian interpretation" is nothing else than the Copenhagen interpretation modified by the assertion that "nothing happens during the measurement". Except that during the measurement, the wave function clearly has to change as the observer is acquiring a new knowledge. In the "innocently Everettian picture", they need to assume that "parts of the wave function" start to behave as different worlds, and the probability is interpreted as some subjective self-locating probability.

Except that no one has ever written down any well-defined rules about "how the wave function should generally be decomposed to pieces" and "how the Born rule may be derived without assuming it". So the "innocently Everettian interpretation" is nothing else than the Copenhagen interpretation (i.e. proper quantum mechanics) from which a part of the rules – the rules connected with the measurement and the collapse – is completely removed, while no viable replacement has ever been found (and I find it obvious that it cannot exist).

(In the comments, Tony makes it clear that some Everettians deliberately want to omit "everything linked to the measurement" partly to make the "whole theory" including the measurement reversible. What they're missing is that the measurements demonstrably are at least psychologically irreversible – we learn facts but never unlearn them during the measurement, for example – so if they erase this well-known irreversibility, which is nothing else than the difference between the past and the future known already to babies, they will end up with a theory that contradicts absolutely basic facts about the world. But theories of science should be compatible with the facts about Nature and not only the basic ones. It sounds unbelievable but these people's proclamations really do suggest that they're not capable of distinguishing the past and the future.)

It is spectacularly clear that for quantum mechanics to be connected with the observations, one needs to admit that there exists an independent input concerning the specification of the observation – the observer must know that it is taking place, what is being measured, and he must feel the result when it's measured – and this observation also induces the collapse of the wave function, and the probabilities of one outcome or another may be calculated from the equations. There just can't be any replacement. All attempts to "derive" the Born rule for probabilities from something that doesn't talk about probabilities are absurdly hopeless.

You simply can't "derive" the word "probability" in the fundamental laws of Nature from something that is not probabilistic! The fundamental laws either talk about real calculable values of probabilities, or they don't. If they don't, they never will and there's no way for the probability to suddenly "emerge" out of something that isn't probabilistic. If they do, all "derivations" of the Born rule are unavoidably circular. Proper quantum mechanics does need to talk about probabilities at the fundamental level and it authoritatively states that this rule can't be derived from anything more fundamental.

Gell-Mann, Hartle, Omnes, Griffiths, and others have talked about the "Consistent Histories" interpretation of quantum mechanics. Is it a variation of the "Everettian ideas"? Is it a competitor of the Copenhagen school? I think that it's not a competitor of the Copenhagen school and whether it arises from the "Everettian ideas" depends on your definition of the notoriously ill-defined "Everettian ideas".

In "The Quark And the Jaguar" book, Gell-Mann uses the "histories" that sound Everettian to make claims like these:
In the EPR experiment, experimenters may decide to measure different properties of the entangled particles. For each choice of the quantities that the people want to measure, there is a branch of the history. For example, if they may measure \(j_x\) or \(j_y\) in the GHZM experiment, there are 8 branches \(xxx,xxy,\dots, yyy\). And on each history, we may find correlations whose explanation is the same as the correlation of Bertlemann's socks.
This is not a quote but an equivalent sequence of ideas appears in Gell-Mann's 1994 book. Now, it's a bit funny to ask whether the "branch of the history" is something old-fashioned that was supposed to be understood by physicists in the 1920s or earlier; or it is "Everettian branches of the history"; or the "Gell-Mann-Hartle branches of the history". The added value of all the "newer ideas" is so debatable and fuzzily bounded that no answer exists, I think.

If you say that they're "many worlds" in some Everettian sense, be my guest. The different branches are the "many worlds". And the correlation is completely mundane and not mysterious – it has the same reason as correlations of the color of 2 socks of Dr Bertlemann (a nutcase in Vienna who always avoids wearing 2 socks of the same color). The perfect anticorrelation in \(j_z\) as well as \(j_x\) or \(j_y\) may follow from the angular momentum conservation law, to mention an example. But for this "non-mysterious" description of the correlation to make sense, we must first divide the possible histories in the "branches" differing by the macroscopic question what quantities the experimenters decided to measure.

And Everett gets a credit for that point. Once you split the wave function to these branches, you may use the same explanation of the correlations on each branch as you had in classical physics.

I find it absolutely breathtaking that Everett – presented as a warrior against the Copenhagen interpretation – gets credit for this idea because the actual physical content of this idea is exactly the Bohr's principle of complementarity! So why should someone who is sold as a warrior-in-chief against Niels Bohr get credit exactly for this big observation that Bohr did several decades earlier and that the top physicists have worshiped Bohr for? It just leaves me speechless.

Bohr's principle of complementarity exactly says that an object – e.g. an EPR entangled particle pair – may be studied in different ways or with different apparatuses. These different measurements can't be performed at the same moment. They are mutually incompatible – and they give us different type of information about the measured object. We can measure the particle-like or wave-like properties of the field quanta, and so on. Within each allowed method to probe a physical system (a way that is taken from a list of mutually complementary options), the laws of logical reasoning hold pretty much just like they did in classical physics.

Bohr's complementary descriptions are exactly the same thing as the "many worlds" sometimes attributed to Everett.

Bohr just found it wiser not to adopt this silly terminology involving the "many worlds" but the "separation" (and therefore not simultaneous existence) between different, complementary Bohrian ways to probe the system is exactly the same apparent "separation" that makes the more sensible post-Everettian people talk about the "many worlds".

Niels Bohr is often criticized for being incomprehensible. But he did it decades before Everett expressed a similar idea. And Bohr actually did it in a much clearer way than Everett. If Everett meant this valid idea – that the different branches of the wave function are "de facto separated worlds" and "it's not necessary for them to be logically consistent with each other for that reason" – it was an idea found by Bohr decades earlier, and Bohr was rightfully celebrated for that before many physicists got crazy. It is plain insane to celebrate Everett – let alone anti-Copenhagen warriors – for those insights.

And Everett was actually much less clear about all these things. His disciples remain split on the very basic question whether he actually argued that the "different worlds actually exist". I don't know whether he believed in their real existence, in the classical sense, and I don't really care because Everett was just another confused person. This question is only interesting for his worshipers – from my perspective, Everett doesn't make it to the top 1,000 physicists of the 20th century so I really can't care much whether he understood some particular basic insight about quantum mechanics.

But everything that actually makes sense about these Everettian ideas was previously found by the Copenhagen school. And all the "genuinely added value" is just a pile of incoherent nonsense and falsehoods.

Is the "Consistent Histories" interpretation a ramification of Everettian ideas? It can't be answered, "Everettian ideas" are ill-defined. Are the "Consistent Histories" rebuking Copenhagen? I don't think so.

Gell-Mann wanted to be clearer than Bohr and I think that at the end, he was. But when you look at what the consistent histories are, they are just sequences of measurements. You may formulate quantum mechanics in such a way that the "elementary building block" is not one separate measurement – and you ask how it is affected by the previous measurements; and how it affects the next measurements. Instead, you may study the "sequences of measurements", the whole histories, from scratch.

But quantum mechanics calculates the probabilities of different histories in the set of consistent histories. This is nothing else than when quantum mechanics calculates the probabilities of different measurements. The consistent histories get reduced to the usual wisdom about the measurement if the history only includes one moment with one measurement. And when the consistent histories are longer, you may build them from pieces – from individual measurements – according to the Copenhagen rules. So there's really no difference.

In some sense, (Gell-Mann's and pals') "Consistent Histories" interpretation is constructed out of the Copenhagen interpretation exactly in the same way in which Feynman's sum-over-histories (that actually computes the results, not just interprets things) is constructed from the Heisenberg or Schrödinger moment-by-moment picture of quantum mechanics.

You may slice the spacetime and histories to pieces or not. But the two types of descriptions are physically equivalent and the equivalence should be self-evident. A difference between Feynman's and Gell-Mann's attitude was that Feynman has never tried to claim that he had discovered a new theory by reformulating it in the path-integral language. On the other hand, Gell-Mann and pals did often present the "Consistent Histories" to be a replacement for the Copenhagen postulates of quantum mechanics. I think it's just a pose. The two situations are completely equivalent. They're just ways to reformulate the same rules in a slightly different language.

Gell-Mann may have wanted to find something completely different about the foundations of quantum mechanics but when his work was mostly over, it was clear that he didn't. He has confirmed the Copenhagen principles.

Just to be sure, the "Consistent Histories" approach may avoid the Copenhagen's favorite word "observer" but it in no way avoids his "beef" (even if we don't assume that the observer is either a cow or the founder of McDonald's). Just like in the Copenhagen interpretation, the observer may in principle "feel" many observables describing the physical system and he must know what he feels when he measures something, the "Consistent Histories" approximation depends on your selection of the "set of consistent histories" which isn't unique, either.

In the "Consistent Histories", someone has to make the choice – what is the collection of the consistent histories you want to consider. And this "someone" is exactly the same agent as the observer in the orthodox, "Copenhagen" quantum mechanics. Someone who cares about the "which history in the set" information because he can feel it, it affects him. So the "Consistent Histories" description may look more elegant and perhaps more impersonal but the content is exactly the same and the claims of "progress" are pretty much spurious.

Also, it's been often said – by Gell-Mann and others – that the Copenhagen quantum mechanics isn't well-defined in quantum gravity etc. The Hilbert spaces have to behave in some way under the general coordinate transformations and it causes problems. Except that the problems with the description once you allow the diffeomorphism (gauge) symmetries are clearly genuine. You want to avoid them. You may avoid them by gauge-fixing. (AdS/CFT and Matrix theory avoid all these problems by giving you totally non-gravitational theories that are equivalent to a background of quantum gravity in string/M-theory.)

My interpretation of the recent 10 years in quantum gravity that I have often stressed (and I sometimes called it "background indifference") is that you must pick a classical spacetime background – and the Hilbert space of a consistent theory of quantum gravity is pretty much always able to adapt itself to the background and be interpreted as excitations of the background. The closer the classical background is to the "real geometry" of the situation, the more natural the description will be, but the description is possible in other cases, too.

Different spacetime-geometry-changing dualities are examples of this principle of background indifference: T-duality, U-duality, mirror symmetry, AdS/CFT, and even ER-EPR correspondence. There are different classical geometric backgrounds and the same Hilbert spaces may be represented as excitations of those backgrounds. The very existence of "coherent states of gravitons" is an example of the fact that the backgrounds may be changed continuously, too.

In all cases, you ultimately want a theory in which you may define some gauge-invariant operators (which is harder in the case of a theory with the diffeomorphism symmetries but it's needed). Once you have those, quantum mechanics may be applied to this theory just like it may be applied to non-gravitational theories! The only special feature of the quantum gravitational theories is then the choice of the right Hamiltonian (or a related larger algebra) or the S-matrix. But the foundations of quantum mechanics aren't really affected by quantum gravity in the most general sense! In particular, the usual Copenhagen postulates of quantum mechanics are good enough for quantum gravity as well. Observers are needed and they observe some operators which are acting on a Hilbert space. A theory of quantum gravity is not "different", it's just a special class of quantum mechanical theories much like gauge (quantum field) theories.

While I have been calling myself a fan of the "Consistent Histories" for quite some time, I am annoyed by the fact that despite the "correct" beef of this "interpretation", it has mainly helped to "proliferate the interpretations" and strengthen the utterly incorrect meme that there are many inequivalent versions of quantum mechanics and every physicist may freely choose his favorite one, as if he were choosing his favorite soda.

That just isn't the case. There's only one correct theory of quantum mechanics. It may be formulated in many equivalent ways and there's always some freedom in the language one chooses; and then there are many inequivalent candidate theories and they may be shown wrong. I urge the proponents of the "Consistent Histories" to endorse this statement and admit that the "Consistent Histories" are just a reformulation of the Copenhagen rules of quantum mechanics.

No comments:

Post a Comment