Wednesday, November 20, 2019

Three foreign lands of Scott Alexander

New Hungarian boson resurrected: today, a Science Alert press release, RT, and others promote the claimed discovery by Krasznahorkay et al. of a new nucleus in which a new boson X17 (the magyaron) of mass 16.7-16.8 MeV seems to be the intermediate state. See a 2016 blog post about the previous beryllium-8 nucleus showing X17 and the new October 2019 preprint with 7-sigma helium-4 evidence. BTW Krasznahorkay is the Hungarian spelling of an adjective derived from the Slovak name Krásna Hôrka [Beautiful Little Hill], a Slovak castle near the modern Hungarian border (the non-modern Hungary included all of Slovakia or "Upper Hungary", you know LOL). It is one of the most preserved Slovak castles despite the big 2012 fire.
Gas, a user, has pointed out that Scott Alexander has written three stories whose punchline is an unusual argument in favor of Sean Carroll's many-worlds misconceptions.



Someone at the VW Group saw this crazy viral video and asked "Why don't we have a Škoda version of it" and accidentally, it ended with a nearly equally viral result, 7M vs 4M views.

To be more specific, Alexander wrote three stories about the visits to foreign lands. Kiki went to a land where they cared whether he added "and supernovae explode behind the horizon" to the list of his predictions.

Choo Choo, a master of racing, went to another foreign land where they couldn't decide whether "God planting dinosaur bones" was a simpler theory than "the fossils prove the evolution".



At the end, Kiki and Choo Choo danced, met in Taiwan perchance, and drove together where no one drove before. They met so many troubles and a man who wasn't good enough to be the orange man but he was good enough to have orange eyes. And he told Kiki and Choo Choo that if they appreciate the lessons that they had learned in their separate two lands above, they will also be ready to accept the Many World Interpretation of quantum mechanics.

And then the man whose name was Sean Carroll perchance vanished in a puff of smoke.



This story obviously can't work because – however attractive the idea sounds – a cheesy critic of quantum mechanics has a nearly zero probability to vanish in a puff of smoke. On top of that, Scott Alexander's childishly confused arguments fail to work, too. Let's look at the arguments more closely.

I Supernova

In Kiki's land, we are supposed to learn the moral that people care about the difference between two theories even though they make exactly the same predictions for testable situations.

But that's not what good theoretical physicists will agree with. When all the in principle testable predictions of two theories are exactly the same (e.g. Heisenberg's, Schrödinger's, and Feynman's path integral approaches to quantum mechanics; or S/T/U-dual theories in string theory or field theory; or the two sides of AdS/CFT), then the two theories must be considered physically equivalent. This is a valid "philosophical" rule that deep enough natural scientists do accept and it has been important in physics for more than a century.

Alexander's confusion is mostly the result of the mixing of actual science with pop science and assorted non-expert demagogues who have completely obfuscated the difference between "testable in practice" and "testable in principle". Only the latter characteristic, "testable in principle", is needed for science to be science. When something may be tested in practice, it's a nice bonus but it is in no way needed for someone's research to be science.

Do supernovae explode behind the cosmic horizon? When a theory implies that the region behind the horizon exists and obeys the same local laws as in the Milky Way, there are similarly high probabilities that supernovae explode there, too. So the answer "Yes" follows from any natural enough theory that respects locality at least to a minimum degree.

Can we test this prediction? By straightforward causality, we can't. I think that you should accept this statement – we can't observe things behind the horizon – as an undisputed statement in general, even in quantum gravity. With this definition of concepts, the question becomes whether the inaccessible regions behind the horizons exist at all – and how much they're inaccessible.

So one point of mine is that I really disagree with Alexander's "moral from the first land". I don't care about the difference between two theories whose observable implications are exactly the same. Such theories are perfectly equivalent. But that isn't the case of the Copenhagen Interpretation and the Many Worlds Interpretation. The former is well-defined and actually predicts all the experiments precisely; the latter isn't well-defined, is just a philosophical fable from cocktail parties, and can't properly and completely predict basically anything. It just doesn't work at all. The claims that the Everettian theory is at least as good as Copenhagen are pure lies of the marketing type and the buyers of these lies are pure morons.

Instead, we should discuss the actual reaction that a scientifically thinking person would make when exposed to the addition to the predictions, "and the supernovae explode behind the horizons". A scientifically literate person would notice that this question about the supernovae is almost totally separate from the question whether "acid plus base react to produce salt and water".

The chemical reaction and the supernovae explosions are almost completely independent branches of the human knowledge. So their mixture clearly looks fishy and it is almost certainly an example of demagogy: if you learn something about the reactions of acids and bases, you will learn nothing about the supernovae behind the cosmic horizons – and vice versa. So why would a person who wants to understand things mix them? Their mixing is clearly a trick by a demagogue who wants others not to understand, who wants others to be confused morons.

For example, to be confused enough morons so that they're even ready to buy Sean Carroll's 100% irrational texts about the many worlds. But let's leave the orange-eye man who is bad for the third world.

The actual correct lessons from Kiki's world are: (1) two theories that produce the same predictions for all in principle testable things must be considered physically equivalent, (2) almost all pairs claimed to be equivalent by the pop science writers are actually unequivalent (like Copenhagen-Everett). Only when you can find complete enough technical proofs of (or persuasive enough circumstantial evidence for) a "duality" or "equivalence" in the physics literature, there is a chance that the equivalence holds.

II God bone-planter vs evolution: simplicity

When Choo Choo visits the second foreign land, Scott Alexander wants you to learn the lesson that you have to "understand philosophy" to decide whether Darwin's theory of evolution is a better explanation for the dinosaurs' bones than the theory that "God planted the bones into the soil". It's because, as Alexander says, the better theory is the "simpler one".

Just like in Kiki's land, all "morals" that Alexander wants you to learn from Choo Choo's land are wrong, too.

First of all, science isn't looking for simpler theories. Science is looking for correct theories. The statement – often described as Occam's razor – that simpler theories are true is just a rule-of-thumb. It is in no way guaranteed to be correct; and it is no way the most reliable tool that the actual scientists are using. It is right to summarize Occam's razor as a very unreliable popular-science description of some considerations that may sometimes legitimately influence scientists while picking theories. But to claim that Occam's razor is the "best" argument for preferring Darwin's theory over the divine planting is just completely and utterly wrong.

Scientists are looking for correct theories – conglomerates of correct propositions that may be generated from several basic axioms (that are also correct, primordially correct). And because scientists are rarely certain what is correct, especially in the cutting-edge research, they may quantify the degree of correctness by a continuous variable, the probability. Scientists are preferring theories that are more likely to be true.

"Darwin's evolution" and "God planting things into soil" are two competing theoretical frameworks to explain the dinosaur bones. The scientific method – or the Bayesian inference – tells us to assign these two rivals with comparable prior probabilities; and then replacing the prior probabilities with the posterior ones by taking the evidence into account.

Unless you choose the priors to be insanely unfair, favoring one of the frameworks, the second step will make a much greater difference on the result. Why? Because we observe many patterns in the bones, e.g. their similarity with each other (and the compatibility with the gradual evolution of life) that have a high probability to occur according to the evolution; but a tiny probability to occur within a generic "God planting bones" theory. The "God planting theory" cannot be predefined to be compatible with all the detailed experimental data – that would be cheating. It has to be defined neutrally and the probability that the bones will look like bones of dinosaurs who could have been relatives of each other within the evolution is infinitesimal assuming that God was responsible for the bones.

It would be much more likely for God to bury a bunch of bones that look unrelated. God could have deliberately deceived us, of course. Here, a humanities-loving person could mention Einstein's quote "Subtle is the Lord, malicious He is not" and I could serve it as well. It basically works for my purposes. But unlike Einstein, I suppose, I hate the idea that people accept "poetic slogans" as winning arguments in science. That's not really how science works and "poetic slogans" often contradict each other and humanities don't have any algorithms to systematically and correctly choose the stronger ones!

Science just isn't organized as a method to choose the truth that has some quotes and slogans at the top of the hierarchy of power. Science has rather technical tools at the top – Bayesian inference, scientific method, general postulates of quantum mechanics and other heavily verified axioms, and similar stuff. All these things are properly expressed in the language that requires some mathematics. Every person who understands the very spirit of physics must know that the replacement of the key criteria by some "slogans that guys from the departments of humanities could prefer" means to become less scientifically trustworthy. Einstein's quote about the non-malicious Lord is just a poetic commentary on the winning answer, a rationalization of a class of correct answers, not a reliable tool to be right in a completely new situation.

So instead, I would say that God could be malicious and deliberately deceive us into thinking that the evolution is right. (And in some cases, people have observed that God was malicious – for example when He sent a package with the muon although no one had ordered that. Only a malicious sender sends a package that no one has ordered.) But He could be malicious in googols of other ways. Even if God is malicious, the probability that He is malicious exactly in the right way so that we incorrectly choose the evolution instead of the correct Biblical explanation is still tiny! ;-) If He were malicious in the "right" way, you would still have to scream: What a coincidance! Even if God is subtle and malicious, the evolution theory still wins because there are many types of the mathematically possible malicious entities, or whatever I should call Him so that He doesn't get offended as an SJW snowflake. ;-)

OK, maybe God does things in a way that the correct (but hard to reproduce) calculation of the probability makes it reasonably likely or certain that He actually fakes the evolution deliberately. But you know, if God is this subtle, malicious, and focused on deceiving His subordinates, then we should simply submit to Him and allow Him to deceive us. I think that this is a manifestation of the scientific integrity. If God deceives us in a way that doesn't realistically allow us to rationally escape the deception, then I believe that God will reward us for being sweet and honest losers who will be allowed to admit "yes, you fooled us nicely, Goddie" in Heaven where the fooled scientists will get.

At any rate, the main points of this section of mine were that scientists (1) pick the more likely theories/frameworks, and the probabilities of the hypotheses are at least approximately calculable by Bayesian inference etc., (2) scientists don't need to rely on "philosophy". Philosophy – in the sense of the "stuff in which the folks from the philosophy departments are considered 'better' than physicists" – simply doesn't have any capacity to legitimately influence conclusions of good physicists or other natural scientists. In the history, philosophers have inspired physicists (especially the positivists have inspired both Bohr and Einstein to help to kickstart the two main revolutions of the 20th century physics) but philosophers could never be treated as trusted leaders of physicists.

If someone says that physicists should listen to the philosophers (in the sense of promoting philosophers into principal investigators who determine the big picture of the physicists' research groups), he or she doesn't understand what physics or natural sciences mean at all. Philosophers don't matter and in the same way, the "counting of Devils and Gods" doesn't matter at all. What matters in the selection of the preferred theories and theoretical frameworks are the probabilities, not the number of Devils, and probabilities are being computed, instead of taken as mandatory dogmas from would-be authorities like physics-wise crackpots calling themselves "philosophers" and from their ad hoc slogans that are becoming more or less popular for totally unscientific reasons.



At the last moment, Franz Liszt lost this place again and Brahms' Hungarian Dance No 5 won.

III Combining wrong lessons to advocate Sean Carroll's misconceptions

OK, in three stories by Scott Alexander, Kiki and Choo Choo finally traveled where no one drove before. They could really drive, that was a nice Škoda, and I hope you will tell me if you buy a Škoda because of this blog post on the philosophy of science. I will surely inform the CEO of Škoda Herr Bernhard Meier if that goal succeeds LOL.

Because I rejected the morals from the first two lands, Kiki's and Choo Choo's, and because the third land combines the morals from those two lands, it's obvious that the moral from the land of the man with the orange eyes is invalidated, too. But I still want to say something more because the reasoning in Scott Alexander's third land is even more muddy than the reasoning in the first two lands (but despite the muddiness, it ironically has many more correct statements!).

I suppose that Scott Alexander is joking in most sentences. I am also joking but I believe that the intelligent readers of The Reference Frame can kind of distinguish which formulations of mine are chosen to be entertaining and which are serious. A person with the IQ above 120 should be able to see "this is supposed to be Motl's joke" within 20 seconds and after seeing the previous and next 2 sentences. In 40 more seconds, they should be capable of sanitizing my playful sentences so that the blog post is turned into a serious boring text. I don't do this sanitization because (1) I am playful and more amused by writing things entertainingly, (2) I think that the readers are also more attracted to playful texts. But this playfulness has some limits: I still do believe that the jokes don't obfuscate most of the message for most readers.

On the other hand, I think that Scott Alexander also wants his final conclusion to be taken seriously. He wants you to believe that the "supernova lesson" plus the "bone lesson" imply that you should allow Sean Carroll's "many worlds" theoretical framework to survive despite its apparently being "untestable". A problem is that the logical reasoning leading to this conclusion seems to be wrong. As I mentioned, I think that the morals from Kiki's land and Choo Choo's lands are wrong. But even if you considered them correct, the reasoning in Sean Carroll's land that is supposed to merge the previous two is even more irrational – and apparently needs the sentences that seem like jokes to actually work.

OK, in his third land, Sean Carroll starts by accusing his critic that he fought against something abroad. But the doors were locked. Great. Carroll's criticism is irrational and "abroad" was clearly a poetic repetition of the "behind the cosmic horizon" from the first, Kiki's land. But aside from these poetic links, I have no idea what role this joke – and therefore Kiki's land – is supposed to play in Alexander's argumentation. What does the invalidity of Carroll's book have to with the accusation that someone fought against something abroad? I think that even more clearly than in previous cases, this is just a noisy addition of some question that has absolutely nothing to do with the topic under consideration. Just like it was irrational to combine the "acid-base reactions" with the "supernovae behind horizons", it's irrational to combine the evaluation of Carroll's book with the stamps in the Carroll's critics' passports.

The lesson from the supernova world is meant to say that we care which "interpretation of quantum mechanics" is right even though they are said to be empirically equivalent. Although the moral of the supernova world was wrong, I agree with this conclusion. I surely care about the difference between the Copenhagen rules and the Many Worlds framework because they demonstrably aren't the same. To say that the wave function is observer-independent or observer-dependent, collapsing or non-collapsing are two absolutely fatal differences between types of theories that really can't be bridged at all. In either of these two dual characteristics, you just can't have equivalent pairs of theories with one theory with the first trait and an equivalent theory with the opposite trait!

Again, I surely do care about the difference between the interpretations. And all the pop-science 100 times repeated comments that two totally different theories are equivalent, indistinguishable, or that something is "untestable" are complete idiocies. Is the Many Worlds testable? You bet. Do you need the theory to predict probabilities? You bet, look at the interference patterns or any other experiment. Does MWI predict probabilities? Not at all. So Copenhagen predicts everything, MWI predicts nothing, and their difference is exactly as "testable" as the difference between the health of a healthy baby and a baby that has been cut to 500 equal pieces of flesh. Is the difference between these two babies also "untestable"? I don't think so. Quite generally, pretty much everyone who has claimed modern physics to be "untestable" is brain-dead, much like the people who kept on reading texts by these brain-dead writers for years.

The second, Choo Choo's world with dinosaur bones, is used because "the theory with many worlds isn't less simple" because simplicity isn't equivalent to the "low number of the worlds". I generally agree with this point, too. Astronomy, astrophysics, and cosmology have evolved towards the explanations of the Universe which asserts that the number of stars in the Universe is really huge – although we only observe a much smaller number of stars "directly". Distant galaxies look like stars, some white dots. But we "know" that they're composed of many stars.

Of course we can't prefer the theory in which the distant galaxies are said to be "single stars" just because the number of stars would be smaller according to this theory. This theory would be much less compatible with the data and if you adjusted it to make it compatible with the data, it would be artificial (basically because you would have to invent some interpolation between two different totally theories, one usable for the nearby Universe and one for the distant one) and the class of such generalized theories would be much less likely due to the large number of possible adjustments (analogy of the many ways in which God may be malicious). There are actual tangible reasons why many of our theories contain many stars in the Universe, many more than what we see, and (although these answers aren't quite settled by now) we may also prefer theories with many vacua in the landscape or many Universes within a multiverse – for reasons that are potentially as powerful as those that make us prefer theories with many stars in distant galaxies.

So although morals from Kiki's world and Choo Choo's world were wrong, the actual assertions in Sean Carroll's world that are based upon them are correct: We don't discard a theory as a redundant copy just because it's said to predict the same numbers in some way (because the statements "what these numbers experimentally mean" may be totally unequivalent). And we don't discard a theory because it has many things, like many stars or many worlds, because the assumption that "the number of things should be small" is just a weird variation of Occam's razor that has no justifiable reason to be right (basically because it can't be transformed to a low probability of the disfavored theory by Bayesian inference).

We discard a theory because it's wrong – or we discard a theoretical framework because the framework is extremely unlikely because it requires a huge number of unlikely adjustments and choices to be compatible with the experimental data – because with modest amounts of adjustment, the theory is clearly incompatible with the experimental data or almost certain to contradict them. The Many Worlds Interpretation of quantum mechanics is either unavoidably and sharply wrong, or incapable of producing predictions about the observable entities altogether, or insanely unlikely in this sense and that's why we discard it. MWI doesn't actually have axioms that would be sufficient to derive the statements "the probability of this or that observed or unobserved phenomenon is close to 100% or close to 0%". Some of the absolutely essential "Copenhagen" axioms of quantum mechanics – related to the collapse – have been removed to obtain the MWI axioms, basically for ideological reasons, and the castrated theory is as hopeless as a car without any engine.

That is the actual reason why we say that Sean Carroll doesn't have the slightest clue about the inner workings of quantum mechanics. Indeed, the incorrect claims about two theories that aren't equivalent don't affect us. Indeed, the large number of worlds in the many worlds isn't an automatically lethal vice. Only large or small numbers that may be translated to tiny probabilities (the inverse googolplex or so) are lethal for theories!

So while it was fun for Kiki and Choo Choo to visit all the foreign lands so that everyone danced (and there was no war anymore), the only moral of their story is that you should dance in the crazy way and buy a Škoda.

And that's the memo.

P.S.: I believe that as a bloc, the 236 comments under Alexander's post (current number) are just the enumeration of all possible mistakes, irrational and wrong evaluation of all the claims in Alexander's text, and random irrational connections to all adjacent topics that you could invent. Discussions of this kind may be fun and many of you are amused by my response, too, and by your creative additions that you will add, too. But they are not science. The rules of the scientific method may be summarized in a few paragraphs – and proven in a few more paragraphs – and the postulates of quantum mechanics may be written down on one page – and proven (I mean the weaker, probabilistic and empirically rooted, physics-like proof in both cases) on a few more pages.

Everyone who is filling hundreds of pages by gibberish while claiming that these pages are needed because he is solving some "problems" with quantum mechanics is a clueless crank. That includes the guy with the orange eyes who has even failed to disappeared in a puff of smoke so far – and Scott Alexander who has discussed quantum mechanics by combining orange eyes with acids, bases, supernovae, orange eyes, and dinosaur bones because this kind of idiotic mishmash is basically unavoidable in texts by anyone who fails to understand that the correct statements about the foundations of quantum mechanics (and their "proofs") are a matter of few pages, not hundreds of pages.

If someone can't get this simple point – that quantum mechanics is straightforward, well-defined, and its rules may be seen to be inevitable rather quickly – then he is destined to mix supernovae with acids, dinosaur bones, malicious God, and orange eyes of men who like the orange man. This inevitability is due to the GIGO dictum – garbage in, garbage out.

No comments:

Post a Comment