Monday, November 26, 2012 ... Deutsch/Español/Related posts from blogosphere

Many worlds vs positivism and symmetries

About a dozen of TRF articles mention Hugh Everett and his "many-worlds interpretation" of quantum mechanics. Exactly three months ago, I showed that "many worlds" don't exist as long as one uses the standard rules of quantum mechanics to answer the very question about their existence.

If we use the same rules to answer the question "Do many worlds exist?" as we use for answering questions about the electrons' spins and other questions "obviously accessible to the experiments", the answer of quantum mechanics is a resounding No. There can't be any "multiple worlds". After all, the splitting of the worlds would correspond to a quantum xeroxing machine and that's prohibited by the linearity of the evolution operators in quantum mechanics. Also, the conservation laws would be violated whenever the worlds split, assuming that they were not split before the "measurement" or another critical moment. And if they were split in advance, the interpretation would violate causality because the "Everett multiverse" would have know about the measurements in advance.

Quantum mechanics unambiguously says that the linear superposition of orthogonal states, \(\ket\alpha+\ket\beta\), doesn't mean that "both the things described by \(\ket\alpha\) and \(\ket\beta\) exist at the same time". Instead, the plus sign means "OR", not "AND". The state says "only \(\ket\alpha\) is possible AND only \(\ket\beta\) is possible" but when we want to omit the words "possible", the only right translation is "Nature realized \(\ket\alpha\) OR \(\ket\beta\)". It's the usual probabilistic mixture. Well, there is a difference: in quantum mechanics, we first add the complex probability amplitudes, and then we square the absolute values of the results (the probabilities). In classical statistical physics, we sum up the probabilities directly so the "mixed" or "interference" terms would be absent.

So the "many worlds" are obviously prohibited when the rules of quantum mechanics are being used for all physical questions, including the questions that some people could be religiously prejudiced about. (I really think it's analogous to religious beliefs because many otherwise rational people abandon all rational thinking when it comes to questions that have the potential to unseat or otherwise disturb their God. Their boundary behind which rational thinking is prohibited is as arbitrary and surprising as it is for those who love to refuse the quantum character of quantum mechanics.) In the following text, I will discuss another part of this issue and explain that if you wanted to use some non-quantum, more classical rules in which quantum mechanics would be embedded, you would be forced to defend an indefensible theoretical framework, too.

When Shannon met Brian Greene a few months ago, he mentioned that he had a "little disagreement" with your humble correspondent about the foundations of quantum mechanics. It surely sounds nice and diplomatic and I endorse the diplomatic content of the quote. However, when it comes to the actual beef, I must say: A little disagreement? Galileo and the Pope had a little disagreement. ;-)

Let me discuss several points related to the indefensibility of the types of "many-worlds interpretation" that try to construct a "more realist model" into which quantum mechanics is embedded as an approximation:

  • it opens a previously non-existent question, "Which events may be classified as a good enough measurement so that they have the right to split the world?", one that can't have an objectively meaningful answer; while the Copenhagen quantum mechanics avoids such problematic questions because the presence of a measurement (or "collapse") is purely a subjective matter, any "realist intepretation" makes the conflict included in this question sharp, and therefore requires a qualitatively different treatment of "quantum objects" and "classical objects"
  • too finely grained a tree of the "many worlds" violates the uncertainty principle
  • the very existence of the "other worlds" violates the positivist, empirical approach to science
  • the Lorentz symmetry is inevitably violated in any "explicitly enough well-defined model" of the splitting of the Universe
  • none of the problems or signs of incompleteness of the probabilistic Copenhagen-like interpretation exists.
These are the reasons that force every impartial physicist who understands these things to conclude that the "many worlds" can't exist in any form and a non-realist probabilistic interpretation of the objects in quantum mechanics – the amplitudes are just tools to find probabilistic answers to questions that observers may subjectively ask, not a reflection of a fundamentally objective reality – is the only plausible outcome of a rational evaluation of the empirical evidence, the mathematical framework, and the logical relationships between them.

Proper QM doesn't need an objective definition of what a measurement is; MWI does

Three weeks ago, I argued that quantum mechanics is a tool to probabilistically answer questions that may be "subjectively" asked by observers, not a tool to describe the objective state of reality (which doesn't exist). Despite this fundamental rejection of quantum mechanics, this theory is fully compatible with objective science. It's because the subjective nature of knowledge adds no inconsistency to science – and because the equations of quantum mechanics guarantee correlations in the results that multiple observers obtain when all of them ask the same question.

But the key feature of the fundamentally subjective theory called quantum mechanics is that if no questions are being asked, no questions need to be answered. If the objects in our environments aren't asking any questions for us, we don't need to answer them and we don't need to imagine that the world is doing anything else than evolving the probability amplitudes according to the continuous Schrödinger's equation (or equivalent equations in other pictures). In particular, there's no "collapse" if there's no subject asking questions and learning answers! What is often called the "collapse" is the process of learning and it is a fundamentally subjective process.

In particular, quantum mechanics is compatible with numerous "theories" about the question "who possesses consciousness and is allowed to perceive things". As long as "consciousness" is a totally immaterial, spiritual process (i.e. as long as we don't talk about its measurable, material manifestations, which are clearly as accessible to science – especially neuroscience – as any other material processes), the difference between these different philosophical theories is unphysical. There's no experiment, even principle, that could answer this question. So one may give any answer to the question. In particular, solipsism is fully compatible with quantum mechanics (much like totally opposite philosophies, e.g. one in which you are "one soul" with all the macroscopic objects on Earth and your unified body perceives anything, it just doesn't allow the information to be sent from one part of the body to another). You may consider the whole external world – including other people – to be a collection of "dead machines" that just obediently evolve according to Schrödinger's equation.

Let me emphasize that quantum mechanics doesn't force you to believe that other people are "qualitatively different from you". Clearly, science – especially Darwin's evolution – makes it clear that there's no such qualitative difference between two people as all people share their ancestry. Instead, it allows you to dismiss questions about other people's consciousness – in a purely "spiritual", operationally inconsequential interpretation of the word – as unphysical questions.

The alternative may look more "materialist" which is intriguing for many people but it is a source of a terrible problem. Everett's interpretation suffers from this terrible problem just like any other "realist interpretation" of quantum mechanics, any other theory that tries to present the probabilistic, subjective, non-realist character of quantum mechanics as an illusion following from a hypothetical "realist [classical] theory".

Why is it a terrible problem? Because if you assume that the world is objectively found in a state, you need to prevent it from entering (or staying in) "unfamiliar" complex linear superpositions of macroscopically distinct states. If you assume the world possesses "objective reality" at the fundamental level, the process that does this job – "physical collapse", "splitting of the Universe", or anything of this sort – is a process that simply takes place at a given moment or it doesn't. In principle, all good physicists should ultimately agree whether this process has occurred or not.

But that means a catastrophe. It means that the coherent quantum mechanics has to objectively refuse to hold behind a critical line. One needs to abandon it, modify it, and so on. But such a modification is totally unjustifiable. There doesn't exist a tiniest glimpse of evidence that quantum mechanics could fail to work for some objects larger than \(X\). In fact, it seems obvious that all similar conceivable modifications of quantum mechanics would be either incompatible with the basic observed data or internally inconsistent.

At this point, and in many others, I am flabbergasted how upside-down the explanations by Brian Greene (and others) are. In his book The Hidden Reality, Greene tried to claim that the orthodox probabilistic interpretation of quantum mechanics has to establish an artificial boundary between the world of phenomena described by quantum mechanics and those described by classical physics.

But as we have seen two and three paragraphs above, the orthodox probabilistic interpretation of quantum mechanics is on the contrary the only conceivable explanation that does not have to do anything of the sort. In particular, quantum mechanics is always valid in this approximation. Bohr et al. just correctly said that in addition to quantum physics, the classical picture has to be also (approximately) right for questions whose answers we want to treat as pieces of classical information. But that doesn't mean that there's new physics that invalidates quantum mechanics. It just means that there are two descriptions, quantum and classical, that start to coincide and co-exist for large enough systems.

On the contrary, as I said, the assumption that objective reality exists means that one must believe that Nature has a very particular "critical line" behind which the coherent rules of quantum mechanics that we know from the microscopic world cease to hold. It means that the "objectivist interpretations" of any kind have to assume the existence of new phenomena that isn't observed and that isn't justified by anything whatsoever – except for philosophical prejudices.

When do the worlds split?

The "many worlds interpretation" is a typical attempt to find an objectivist reinterpretation of quantum mechanics – at least that's how e.g. Brian Greene presents it and I totally think that his presentation makes much more sense than any other tirade about Everett's picture (even though it's ultimately totally wrong, too).

It assumes that there must objectively exist "many worlds". In one of them, you measure Stalin to have won the war, in another one, you measure Hitler to have won the war, and so on. How many such universes are there? How finely do you have to divide the universes to make the MWI work? Roughly speaking, you want every small measurement or every small "macroscopic process" to produce new split worlds but what is counted as a measurement or a macroscopic process?

It's a very subtle question and it's easy to see (although some people may prefer not to look) that whatever answer you offer – except for the correct answer, namely that the world never splits – produces insurmountable contradictions. The basic problem is that if you assume that the objective "splitting of the worlds" is the reason why we perceive sharp answers instead of "fuzzy superpositions" (note that the actual reason is that the wave function describes probabilistic distributions with "OR", not a fuzzy shape of objects) and if your "splitting" occurs too rarely (more precisely, less frequently than infinite frequency because any observable may be potentially observed by "someone"), the world will be more fuzzy than the observations indicate. If your splitting will be "too frequent" (more frequent than zero frequency, to be precise), your picture will contradict the uncertainty principle. Note that theories with an "objective collapse of a wave function interpreted as a classical wave" suffer from exactly the same problem. One may say the same things about them.

In the MWI approach, the splitting of the worlds is an objectively important moment. You know, it's not easy to create many new worlds that contain everything that our world does. ;-) An MWI advocate may be tempted to link this splitting to decoherence. However, decoherence is never perfect. In a basis of states we find natural (e.g. because it's local), the density matrix never quite becomes diagonal. There's always a nonzero (although expo-exponentially decreasing) risk of "recoherence" and if you want to avoid this risk, you're never allowed to mechanically divide the worlds.

A related problem – it's actually exactly the same problem – is that when you divide the worlds into "many worlds" too finely, your description will violate the uncertainty principle. Why?

Almost three weeks ago, I discussed consistent histories. Imagine that you observe the trajectory of a particle in a cloud chamber and you decide that the measurement of the position took place every microsecond, at \(t=0.0\), \(t=1.0\times 10^{-3}\), and so on. At each moment, you also measure the momentum of the particle with the best accuracy \(\Delta p\) that is compatible with the uncertainty principle. So far so good.

But the precise moments when the universe splits doesn't seem to be God-given and objective. So another person may think that the splitting occurred \(10^{-16}\) seconds after each splitting according to yourself. Now, there are 10 billion other people who assume the same microsecond spacing but whose "special points" are shifted by multiples of \(10^{-16}\) seconds. Can we reconcile the interpretations of all these people?

You are forced to say that the splitting actually occurs every \(10^{-16}\) seconds. The worlds split every time when at least one person thinks it does. Note that with a larger number of people, I could be forced to accept an arbitrarily fine splitting. But that's a problem because that would imply that in a particular branch of the "many worlds multiverse", the particle's position is measured pretty much at every moment. But if it's so, the uncertainty principle dictates that the momentum can't possibly be determined accurately at all. But each person above actually measured \(p\) rather accurately. By increasing the density of the "split world moments", we violated the uncertainty principle by an arbitrary factor. With 10 billion people, \(\Delta x\cdot \Delta p\) was \(\hbar/10,000,000,000\). Too bad.

So we're allowed to imagine that for macroscopic bodies, the histories are "coarse-grained" and the equivalence classes are treated classically. But we simply can't afford to make the graining too fine. At the same moment, there are clearly no "preferred boundaries" that would separate the space of possible histories into "little cubes" everyone may agree upon (much like the phase space isn't objectively divided to particular "only correct" rectangles of area \(2\pi\hbar\) that everyone must agree upon). It follows that to invent a rule answering the question "When do the worlds split?", you either violate the uncertainty principle or invent some arbtirary rules that contradict the Lorentz symmetry, rotational symmetry, and that simply depend on many totally arbitrary decisions that no one believes may have an objective significance.

Proper quantum mechanics, as pioneered by Heisenberg, Dirac, Pauli, and a few other friends, and as godfathered by Bohr, cleverly avoids all these problems. It doesn't have to divide the phase space to "canonical rectangles with the only right shapes" because it says that such divisions – and the choice of bases – depend on the situation and the questions that an observer asks. They are not objective in character. The same is true for the separation of the "space of possible histories" which is just a harder, infinite-dimensional version of the task to "divide the phase space to the only right rectangular boxes". The reason why we may dismiss the hopeless task of looking for the "precise boundaries in the phase space" is that no objective boundaries exist. They only arise subjectively, in someone's head. But that means that the transition from the linear superpositions to the sharply perceived outcomes of experiments – the "collapse" – is subjective, too.


Philosophy is never a good science. But when it comes to philosophies, many laymen in quantum mechanics – and even people not considered laymen in quantum mechanics by the society or by themselves – often think that "realism" is the right philosophy behind modern science. This viewpoint, based on millions of years of our everyday monkey-like experience, has strengthened by the 250 years of successes of classical physics and it was – unfortunately – energized by Marxism that repeated the untrue equation "science = materialist ideology" many times. Marx, Lenin, and related bastards surely belong among those who have encouraged people to never leave the mental framework of classical physics. But it is "positivism" which is the philosophy that is closest to the founders of the modern science, especially relativity and quantum mechanics.

Positivism says that all reliable knowledge – the truth we are allowed to become fans of – has to boil down to empirical observations and mathematical and logical treatments of such empirical data. It sounds uncontroversial among science types but many of them don't realize how dramatically it differs from the "materialist ideology". In particular, positivism assumes nothing about the "existence of objective reality".

I am not going to worship positivism because in the most general sense I have just mentioned, it represents the absence of any knowledge with beef rather than knowledge itself (this absence is important at the beginning of research – which has to begin with a tabula rasa without prejudices – but it's bad if someone's mind stays tabula rasa even a long time after that). That's also why e.g. Werner Heisenberg heavily criticized positivism as a philosophy – it seemed vacuous to him. It's kind of paradoxical because his beliefs in physics were absolutely positivist. But he rightfully criticized people who don't know or believe anything and who make living out of not discovering anything. ;-)

Incidentally, Auguste Comte – despite his being the founder of this "pro-science" positivist philosophy – had crazy opinions about science, too. He once declared that the chemical composition of the stars would forever remain outside science because we couldn't travel to the stars. It took less than a decade after his death before spectroscopy told us everything about the composition of stars (at least their surface). It's fun to be eager to say that "questions such as XY are inevitably outside science" except that many (or most) statements of this kind (but not all of them) are wrong. Whether some question is scientifically meaningful is a subtle question and difficult science research is needed to resolve it (the question is difficult because we don't know all conceivable experiments and all of their conceivable relationships to the "interesting statements" in advance): philosophical prejudices are never enough. They're not enough if your answer is Yes and they're not enough if your answer is No!

But let's return to Everett's picture. It assumes that there objectively exist "other worlds" except that pretty much by definition, the "splitting" is an irreversible process. Because you can't return to the past, you're also unable to return to the "crossroads" from which other worlds originated. But that's the only way to get to the parallel universes, so you can't "get there" (or interact with them) at all.

Now, if this is universally the case, even in principle – and MWI seems to be crucially based on this assumption – then the question whether the other branches exist is outside science, because of the basic positivist definitions of a reliable truth. We can find new ways to study the sunlight but we can't find ways to return to the past or change the facts or events in the past which is why we're sure that the other worlds will remain empirically inaccessible.

It's an OK argument but I still find it more important that there can't be any sensible "set of rules" that would tell you "when the worlds split". Even if we can't observe other worlds, we may still accumulate very strong indirect evidence that they do exist or they don't exist. I need the previous discussions about the "too finely grained histories" and other things to collect data that are relevant for this question whether the other worlds exist. And the answer is No, they can't exist. Whether they're in principle observable or not, the question about their existence turns out to be available to scientific reasoning – one that boils down to the empirical data and their mathematical and logical treatment – and the answer is, inconveniently enough for MWI and other "realist" advocates, that these constructions can't exist.

Lorentz symmetry is doomed in MWI, too

Consider an EPR-style entanglement experiment. We measure properties of two entangled particles. The two events – two measurements – are spacelike-separated. So which of them occurs first depends on the reference frame.

However, if you literally imagine that the number of worlds is increasing at the moment of each measurement and the "tree of parallel worlds" gets ramified, it is possible to find out whether the first measurement occurred first, or the second measurement occurred first. Each measurement is a "vertex" in the branching tree of worlds and one of them has to be closer to the "root" of the tree in the past, so it occurred first. Because the branching of the worlds is an objective processes, all observers should agree about which of the two spacelike-separated measurements occurred first. But relatively implies that different observers won't agree which of the two spacelike-separated events occurred first. That's a contradiction between MWI and relativity.

Again, proper orthodox quantum mechanics cleverly avoids this problem because nothing "objective" is changing during the measurement. The measurement is a change of the subjective knowledge which is why the ordering of the two changes of knowledge may depend on the observer – and on his reference frame. The subjective interpretation of the wave function and the "collapse" (learning) is a property of orthodox quantum mechanics that is necessary for it to agree with relativity in all physically meaningful, operationally answerable questions. This consistency is fully preserved because the "ordering of the collapses" is identified as a question that is only meaningful subjectively which is why different people (in different inertial systems) don't have to agree about the answer. And indeed, they don't agree.

The splitting of the worlds, much like "objective wave function collapses", inevitably leads to contradictions with relativity.

Quantum mechanics is complete and consistent

But the main reason why all the research into hidden variables, de Broglie-Bohm pilot waves, Ghirardi-Rimini-Weber collapses, and many worlds – among a few other, less widespread "realist approaches" – make no scientific sense is that their main motivation, the hypothetical "flaws of the Copenhagen quantum mechanics", is completely invalid.

There is nothing wrong or incomplete about quantum mechanics – the new framework discovered within the Copenhagen school. The intrinsically probabilistic meaning of the wave function and related insight aren't a matter of "interpretations": they're inseparable properties of quantum mechanics, they're general postulates of quantum mechanics, they're really what makes quantum mechanics quantum. The very moment when someone starts to talk about "interpretations" as something that should be built "on top of quantum mechanics", he is already deluding himself and refusing all the insights that are actually summarized by the term "quantum mechanics". Quantum mechanics as a theory doesn't have any other interpretation than the Copenhagen interpretation, with the Born rule, or newer reformulations of pretty much the same thing.

All the other ideas that are called "interpretations of quantum mechanics" are really totally different theories – either toy models that only work as OK descriptions of some extremely limited and special situations but that can't be generalized to "full physics" or, which is more usual, just hypothetical theories whose existence is a wishful thinking many people protect as gold (even though one may easily show that no such theories compatible with the basic empirical data may exist). People who sell "interpretations of quantum mechanics" nearly as a whole new subdiscipline of physics are analogous to people who want to "interpret Darwin's evolution" for a long enough time so that the creator and the events from the Bible reappear again. They just don't like what the theory is saying and be sure that it is saying that the wave function has to have a subjective probabilistic meaning.

As long as one is doing genuine science and not prejudiced ideology, quantum mechanics, a new framework for physics clarified by the Copenhagen school in the 1920s, isn't negotiable. It's as established as evolution, heliocentrism, or any other important pillar of science, and the people who try to relabel basic postulates of quantum mechanics as "illusions" are as deluded as creationists or geocentrists.

Redistribution of probabilities among many worlds

Two paragraphs earlier, I mentioned that most of the theories in the MWI business and related businesses don't really exist: their existence is a wishful thinking. The dreamed about derivation of the Born rule for the probabilities from the MWI framework is a great example of this unlimited, breathtakingly irrational wishful thinking displayed by the MWI advocates and other "realists".

After the quantum revolution, we know that all empirical evidence coming from repeated experiments may be summarized as measured probabilities of various outcomes of diverse experiments. Once again, all the empirical knowledge about the physical processes that we have may be formulated as a collection of probabilities. Probabilities are everything we may calculate from quantum mechanics (and from other parts of science, too). So they're surely not a detail.

Orthodox quantum mechanics promotes probabilities to fundamental concepts and uses the standard probability calculus – which existed a long time before quantum mechanics – to give you rules how to verify whether the probabilistic predictions of a theory are right. The basic laws of quantum mechanics are intrinsically probabilistic. The MWI framework tries to deny this general point. It has many worlds – something we would normally call "a priori possible results" are pretended to be "real new worlds somewhere out there" – but at this moment, we haven't even started to discuss the empirical evidence yet. We only start to discuss the empirical evidence once we start to compare the measured probabilities with the theoretically predicted ones because everything we empirically know are probabilities!

But when it comes to the probabilities, MWI has absolutely nothing coherent to say. If the probabilities of "spin up" and "spin down" are 64% and 36% respectively, the MWI framework just gives you two worlds. How do you actually extract the numbers, 64% and 36%? There doesn't exist any proposed answer that makes any sense. It's clear that there can't exist any answer that makes any sense because probabilities can't be "derived" out of something that has isn't probabilistic.

You may try to say that there are 64 worlds with "spin up" and 36 worlds with "spin down" and all of them are equally likely (or 16 and 9 worlds, or other multiples). Except that almost all probabilities predicted by quantum mechanics (and verified by experiments) are irrational numbers so you would need infinitely many "many worlds" to account for these irrational probabilities: a division to fractions just isn't good enough and all "rational approximations" would look so awkward that no one would believe they're right. Or you may just have two worlds and assign the unequal probabilities to them (and justify it, for example, by saying that the observer doesn't know "in what universe he is", and the right answer is just one that happens to agree with the Born rule). But a new, "deeper" derivation of the quantum mechanical predictions – and all such predictions have the form of probabilities – is what you wanted to achieve. So if you mysteriously assign your fictitious worlds the probabilities by hand, you have clearly derived nothing. You have just invented a childish story compatible with your religion (one that may still be shown incorrect if you try harder).

When someone tries to offer an allegedly "deeper" explanation of the probabilistic rules but he ignores the fact that he doesn't have any remotely conceivable explanation what the probabilities could emerge from, and this is a superserious problem because all the data in science may be formatted as probabilities, it seems to me that the person is so prejudiced that it makes absolutely no sense to discuss these issues with such a person. What drives him is pure bigotry, a metaphysical, quasi-religious fanaticism ready to overlook all empirical evidence we have or we may ever add. Indeed, that's exactly what the anti-quantum zealots are doing all the time.

And that's the memo.

Add to Digg this Add to reddit

snail feedback (37) :

reader Robert Rehbock said...

Do you consider it ironic to be attacked by some for pointing out the rationality and utility of String theories but attacked by others for pointing out the irrationality and lack of usefulness of MWI?

reader Fred said...

How can the MWI be possible given that it violates the conservation of mass/energy ?

reader Luboš Motl said...

Great question, and one never gets a coherent answer to this one, either. The MWI advocates either say that

1. the conservation laws are just violated - which is bad because they're tested and should follow from the translational symmetries

2. the total mass was always bigger, there were many universes even before the measurement - which is also bad because the multiverse would be "split in advance". It would know that the measurement would occur which is acausal.

At any rate, they can't give you any mathematical description of what actually happens - they just ignore this problem, much like others, and pretend that you shouldn't really believe the law too much etc., but they don't say why and so on.

reader Luboš Motl said...

The difference is that the people who doubt the validity of string theory are crackpots, much like the people who doubt the validity of the Copenhagen school's quantum mechanics, which is the only right analogy between the two situations. If you repeat your indefensible crackpottery again, you're banned, deal. Despite my permanent struggle and banning of aggressive cranks, I am still receiving about 10 times more of this garbage than what I would find tolerable.

reader Shannon said...

Awh ;-) Can't wait to read this :)
MWI is psychedelic !

reader Dilaton said...

Hm Robert, you are probably confusing two things. The MWI of QM (an unneeded and wrong interpretation of QM) has nothing to do with the ST landscape (the legitimately allowed vacua in ST), so there is nothing ironic or contradictory about the fact that there sadly exist these two kinds of very annoying trolls you mention ...

Anyway, I am very grateful that our Chieftain ;-) keeps any kind of trolls out of TRF !

Reading the comment sections below articles of some otherwise quite nice physics blogs (CV, Strassler, Phil Gibbs, Jester) so often annoys and upsets me, as these comment sections are regularly taken over by horrible trolls which as a result, prohibits any serious and reasonable physics discussion :-/

reader Fred said...

Yep. For that simple reason it makes no sense. I would only consider it if they had some specific falsifiability tests so it could be tested as a real alternative physical theory. However without that, it is just an "interpretation" that adds nothing to science, indeed it violates known "laws", and so is not needed. We don't need it.

reader Luke Lea said...

Would it be wrong to say that whereas MWI doesn't really explain anything -- in fact violates the principles of relativity and QM -- anthropopic reasoning could at least in principle explain a few things? You've at times suggested the motivation of anthropic reasoning was quasi-religious (haven't you?). I presume you meant in a different sense -- as in giving up or trying to bring about closure prematurely based on psychological needs.

Anyway I think I got what you were saying in this post: there's no there there (as Gertrude Stein once said about Oakland) behind the laws of quantum mechanics. What you see is what you get -- and it's all you get.

reader Mikael said...

Dear Lubos,
being a longterm reader of this blog it is clear that your position on these things is familiar to me. Still I would say that what you write about these things is as clear or even more clear as what I can understand from anybody including people such as Sidney Coleman.

I would only add that there are maybe two observation which also may belong to the full picture although I have no clue how.

1. I am all for thought experiments in theoretical physics which make you worry about something which can be done in principle irrespective of the fact whether it can be done in practice. Still when I think about Schroedinger's cat I cannot overlook the fact that interference between a living cat and a dead cat may be forever beyond the capability of any human or other experimenter. So maybe in the best positivistic sense we are worrying about something artificial which can never be observed. This could be similar as the classical physicists who were worrying about the recurrence paradox in Thermodynamics,

2. I don't think that any dead or alive person has much useful to say about the relation between the physical world and our conscious experience. Still I would expect that it is not possible to construct a conscious entity out of 10 atoms but you will need something like 10^23 atoms for this. Thus it may be no accident that consciousness happens in the realm of classical experience.

reader Orpheus said...

>>In particular, quantum mechanics is compatible with numerous "theories" about the question "who possesses consciousness and is allowed to perceive things".<<

Just one question: While you are obviously right to regard such questions as unscientific, I am pretty sure you have some opinion on such matters, and (provided "you" exist in some sense) maintaining this blog would make little sense if you were a solipsist. So, if there is no such thing as "objective reality", in what sense can you meaningfully talk about "other people/observers"? I mean in the "unphysical" sense of them subjectively perceiving a state of the world just like you do.
(Let me emphasize that this is not meant to be a counterargument to anything you have written, just a curious question about whose answer I'm not quit sure so far.)

reader James Gallagher said...

This is a wonderful post, but MWI adherents won't be swayed by any of the arguments because there is no current experiment which would distinguish their "interpretation" from the probabilistic one.

I think it will eventually come down to an occam razor type decision, since in my view, MWI is equivalent to a discrete time seeded probabilistic evolution (at planck time resolutions), and we'll never be able to distinguish between the two experimentally unless we ever get finer than 10^-43 secs resolution experiments. Of course the Born rule is easier if you assume probability from the start rather than inventing obfuscating type arguments to derive it, so MWU adherents have to haul along a lot of unnecessary baggage.

Also I think a probabilistic seeded evolution solves the arrow of time arguments rather easily, and also allows our existence to be meaningful since it enables free-will (in a sense that is what most of us really hope free-will means)

reader Robert Rehbock said...

No confusion. Those with "not even wrong" nonsense like MWI ask one to accept their unscientific interpretations of QM that demand one disregard experiment for a "not even wrong" sci fi notion that ignore causality, experiment and the consistency just to preserve a fantasy. Others insist that any extension of SM consistent with QM and including insight to gravity and other BSM physics is " not even wrong" merely because one cannot necessarily prove all yet by experiment. Too me that is ironic.
However, I did not intend any rigor nor did I intend to suggest that the two kinds of trolls meant any scientific connection between theories. I agree with you that TRF is the best. I have been following the blogs only after April. I have found Lumo to indeed provide the only site that focuses on the

reader Robert Rehbock said...

I shall not repeat it. I apologize. I actually was intending to agree with exactly your view.

reader Quantum said...

What about black hole complementarity? Does it matter which observers have "consciousness"? If not, which observer's account should we choose?

reader Luboš Motl said...

Each observer chooses his or her own account and none of them is "objectively better than others". Indeed, this is the whole point and meaning of complementarity, and not only black hole complementarity.

reader Quantum said...

So, it's OK if Alice observes Bob raising his hand if her consistent histories are chosen, while Bob never remembers raising his hand if his consistent histories is chosen?

Alice is freefalling. Bob is an external observer. Bob has decided beforehand to only raise his hand if no firewall is detected.

reader Luboš Motl said...

Dear Quantum, your instructions for Bob are ill-defined. What does it mean for an observer outside to "detect a firewall"? Tell me exactly what he should do before he decides to raise or not to raise his hand.

Otherwise he won't know whether he should raise his hand and your thought experiment makes no sense.

reader Quantum said...

The whole point about firewalls is that

a) they are located outside the event horizon
b) they are located at scales much greater than the Planck scale above the horizon.

reader Luboš Motl said...

Excellent, then there can't be any contradiction involving objects constrained in this way and the equivalence principle guarantees that no observations done in the region you have described will deviate from observations done elsewhere in nearly empty space.

If you want to reconstruct the observations of Alice done in that region, you need to follow all the observables all the way up to the actual event horizon. This was shown e.g. here

The conclusion is obvious and expected - there can't obviously be any contradiction between the two viewpoints.

reader Quantum said...

If Alice detects no firewalls in this region, then according to QFT in curved spacetime, there has to be entanglement between the outgoing Hawking radiation and the infalling Hawking radiation.

By the monogamy of entanglement, the outgoing Hawking radiation can't be entangled in any way with earlier outgoing radiation. This violates the assumption of strong scrambling.

reader Luboš Motl said...

No, this argument is wrong because the infalling Hawking radiation and the earlier Hawking radiation are described by the *same degrees of freedom* written in different, complementary bases, so the outgoing later Hawking radiation that is entangled with "both" is actually entangled with one set of degrees of freedom only, with one wife, and the monogamy holds perfectly.

reader Quantum said...

The infalling modes and the earlier radiation are the same degrees of freedom???

This includes the infalling modes just before crossing the horizon. When they are still a little bit outside by more than the Planck scale.

This violates locality between degrees of freedom which are both outside the horizon by more than the Planck scale!!!

reader Dilaton said...

Dear Lumo,

want to become a Chieftain at Physics SE too ?

I would vote for you :-) !

(Sorry for the off topic, I'm still reading this nice text)

reader Luboš Motl said...

Yes, they're the same degrees of freedom. That's what the term "black hole complementarity" meant in the first place.

Yes, locality is violated by the black hole evaporation – that's always been known to be a necessary condition for the information to get out of the black hole, as a simple look at the Penrose diagram shows.

However, the violation of locality goes to zero when one studies ordinary low-energy modes with wavelength much longer than the Planck length but much shorter than the black hole radius.

reader Quantum said...

No, my surprise is that I always though complementarity is the identification of modes inside the horizon with external modes. At best, this might include modes up to a Planck scale distance above the horizon.

You're claiming something much stronger. You're claiming the identification also applies to infalling modes above the horizon by more than the Planck scale!!!

reader Luboš Motl said...

Sorry, this is a deep misunderstanding of yours.

Black hole complementarity relates two sets of observables, internal and external ones, and the external side has always included the degrees of freedom of the Hawking radiation, pretty much all of them – degrees of freedom very far away from the black hole.

If the identification involved just some "closer than Planck length" degrees of freedom near the black hole, the whole concept of "black hole complementarity" would be pretty much vacuous.

reader Quantum said...

I meant identification of the external outgoing modes with infalling modes which are outside the horizon by more than the Planck length. I wasn't referring to the external side, which can be arbitrarily far away. I was referring to the infalling side.

reader Pablo said...

Hi Lubos, I did not study Physics but read your blog regularly. I have a question that may be interesting for laymen like me. When searching in the Wikipedia for methods of generating entanglement, I saw the following figure in the article "Spontaneous parametric down-conversion":
The figure tells us that photons in the points are entangled. If you measure for example all the time the photon emerging at the left, you will see 50% of the times the polarisation "up" and 50% of the times the polarisation "down". However, could this case in particular be explained by "classic" means? Why not consider that in this case the photon at the left may come from the up of the down conus, 50% of the times, and its state was perfectly defined before detection? The final results are the same, don't they?

If any other layman can give an explanation for this (or Lubos himself, better) it would be great for readers of this blog like me (hope not to be the only silly reader...)


reader Luboš Motl said...

Hola Pablo, thanks for your interest and kind words. The pre-existing classical correlation - referred to as "Bertlmann's socks" by Bell because they have the opposite color "objectively" even before we look - is enough to explain one particular correlation. However, the funny thing about the quantum entanglement is that the same photon pairs may show perfect (or very high) correlation in a different quantity you decide to measure.

For example, 2 electrons may have spin "up" and "down", respectively, or vice versa. And if you measure the spin to the left or right, they will show "left" and "right" or vice versa but never left-left or right-right. This correlation in "multiple quantities" already contradicts any conceivable local classical description.

For example, if your electron 1 is objectively "up" and the other is objectively "down", then you can derive that if the left/right measurements of the spin are done with both of these electrons, you will get left-left, left-right, right-left, right-right possibilities in 25% of cases each - but quantum mechanics shows that only left-right or right-left will occur, each in 50% of cases. The correlation survives to many more pairs of observables that can be measured than classical physics could explain.


reader sadpandabear said...

what do you think of the recent 't Hooft ideas on cellular automata and QM?

reader Luboš Motl said...

They're not recent ideas, first of all, he began with them 15 years ago or so, when I began as a grad student, and I have already won some bets that they would remain considered wrong.

They're wrong and the paper you mention that tries to "apply" these wrong ideas to string theory is even more wrong. It contradicts pretty much every important principle here. Superposition principle of quantum mechanics, symmetries of the theory, everything.

I know that it's tempting to think that a famous author name on the title page implies a correct or valuable paper but it ain't the case.

reader Robert Sykes said...

Steven Hsu, now at Michigan St. U., believes MWI is true and that the proof (not given) is trivial.

reader Olaf Simons said...

Hi Luboš, could we win you to post a blog entry at our site on the question: "is positivism dead?" - or
the question we do not know and you would find far more interesting?

reader Pablo said...

Hi Lubos, thanks for your explanation. However if we go back to the Wikipedia article, the fact that the two photons have to show opposite polarization is always guaranteed, because of conservation laws and as they both come from one photon that gets split in the crystal. If one gets polarized up it belongs to the upper conus, and the other must get polarized down and belongs to the lower conus (and viceversa). I don't get yet why we need of quantum explanation for this case.

I suppose you see this reasoning is wrong at some point. Would you mind explain where?

Cheers and thanks for your blog.


reader Luboš Motl said...

Dear Pablo, right, the correlation, whenever it's perfect, always follows from the angular momentum conservation law. But in local classical physics, it would simply be impossible to produce the correlation regardless of the choice of the axes that the experimenters choose for the measurement of the two photons. I have already explained why and have no time to do so again.

reader Pablo said...

Dear Lubos, thanks for clearing up, I read too quick and confused your "measuring to the left or right" with the electron emerging to the left or to the right in the experiment. Now everythinig is clear and it's easy to understand and accept, except one thing. We know the correlation results are independent of the axes chosen for the measure, but, what if the photons are located in any other point of the upper and lower conus, respectively? The situation would be the same, we get the same results and correlations indpendently of the axes chosen. What difference does it make with the photons emerging at the "entangling" points? The only difference I see is that in this case, the photon to the left will belong 50% of the times to the upper conus (spin "up") and 50% of the times to the lower conus. Is entanglement so a simple idea?
Thanks and cheers

reader Luboš Motl said...

Dear Pablo, chances are high that I may misunderstand your question.

If you're asking about the correlations in positions where photons from a pair etc. are detected, they're generally detected in the opposite positions pretty much by momentum conservation. Well, the momenta are opposite and they get "classically" translated to the directions of future position, at least if they interact with the environment.

On the other hand, for electron-positron pairs, the entanglement of the momenta and the entanglement of the spins are two independent entanglements, mostly. So the spins are entangled in the same way regardless where you measure them and which axes you choose - for any relative orientation of the spin axis and the momentum.

For photons, the polarization vectors used for measurements must always be orthogonal to the momentum direction.

(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){ (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m) })(window,document,'script','//','ga'); ga('create', 'UA-1828728-1', 'auto'); ga('send', 'pageview');