Saturday, May 22, 2010

Why cranks think that there can't be a TOE

Two weeks ago, I wrote about a silly text published by Nude Socialist and written by a man named Marcelo Gleiser who has mixed some random ideas about symmetries and asymmetries to argue that there can't exist any unifying theory. Of course, each single step in his arguments was completely illogical.

Now, Ms Sabine Hossenfelder complains that Mr Gleiser needed 15 years to realize that there can't exist a unifying theory. She was able to "prove" that there can't be a TOE within seconds. And in fact, she gives us her "proof". Gleiser's text shouldn't have been published in Nude Socialist because it's her idea.

Her "proof" goes as follows:
Of course the notion of a final, fundamental, theory of all and everything is faulty. For the simple reason that even if we had a theory that explained everything we know, we could never be sure it would eternally remain the theory of everything we know. As Popper already realized about a century ago, one cannot verify a theory, it can only be falsified. Thus, theories we have are forever out for test, always on the risk that some new data does not fit in. That's exactly what makes a theory scientific. It's also one of the points I made in my FQXi essay. You see, I'm an even Newer Scientist.
Holy crap.

Sweden is a country that has just been eliminated from the ice-hockey championship's final match by my countrymates, much like Germany was eliminated by Russia a few hours later ;-). But do they deserve such incredible idiots to be the representatives of "theoretical physics in Sweden"? It's really pathetic.




A scientific hypothesis or theory must be formulated in such a way that it implies things that could a priori be both wrong and correct. A posteriori, the predictions are - or have been or will be - tested and the result of these tests influences - or has influenced or will influence - the fate of the hypothesis or theory.

If the hypothesis passes the test, it survives. If it fails, it's eliminated. If it partially passes, it may be modified for the improved version to have a better chance in the future tests, and so on.

But if you could prove that there exists a test that will actually disprove the hypothesis, then the hypothesis is not only falsifiable: it is falsified. You don't need to do the test if you can prove what the result of the test will be. The hypothesis is wrong and a scientist must instantly abandon it. Crackpots of the Hossenfelder-Smolin-Woit type follow exactly the opposite strategy: they focus on these excluded hypotheses. It shouldn't be surprising that everything they say about physics ends up being complete garbage.

So Hossenfelder's statement that a falsification must actually exist is turning the rules of science upside down. There can't exist any way to actually falsify a hypothesis if this hypothesis is supposed to remain a viable theory. The falsification must always be just in principle.

Because she uses this crackpot-friendly misinterpretation of the words of a popular but irrelevant and confused philosopher, it's not surprising that all of her conclusions are complete junk as well. Needless to say, a theory of all fundamental interactions and matter types can exist, and we probably know the "first sketch" or even "second sketch" what this unified theory is already today.

But even if you have been brainwashed by the junk physicists and by crooks in general and if you believe that the statement in the previous paragraph is controversial, one can demonstrate the fallacy of Hossenfelder's reasoning in less ambitious but qualitatively isomorphic contexts.

For example, contemporary physics knows the complete theory of interactions between electrons, positrons, and photons that applies to all electromagnetic phenomena with typical energies at least up to ... a few GeVs per particle, if you want me to pick a limitation. It's called Quantum Electrodynamics.

This theory has been constructed and refined since the late 1920s. The result is a well-defined set of laws and rules that agree with all the electromagnetic phenomena that have ever been seen. Can Quantum Electrodynamics be falsified? Well, it could have been falsified in the past. It could have failed in numerous ways.

But it hasn't failed.

Because of these facts, it has become extremely unlikely - and de facto impossible - that the theory will fail as a description of other phenomena within the same energy interval and the same limited (but impressive) accuracy that Quantum Electrodynamics claims to achieve. Because Quantum Electrodynamics has passed all these tests, it has become an established effective theory of certain phenomena. It's pretty much proven to be correct - which is nothing else than saying that the theory has become de facto unfalsifiable.

It was formulated in such a way that it could have been falsifiable - after all, it predicts billions of phenomena at the 10^{-12} relative accuracy or better - but once we learned about the results of the tests, the theory has become unfalsifiable in practice because we know that no further experiments involving electrons, photons, positrons, and electromagnetic fields of the same kind are going to falsify the theory.

Does this a posteriori unfalsifiability of Quantum Electrodynamics make it less scientific? Not at all. It just makes it more likely to be true, more important, and more well-established.

The case of string theory is analogous except that string theory can't be deformed by further deformations and can't break down at any energy scale. That's a basic mathematical feature of the theory. However, much like Quantum Electrodynamics, it has passed lots of nontrivial tests that could have failed but they haven't. The possibility that string theory could be entirely wrong has moved to the realm of science-fiction and to the crackpot blogosphere.

It has become virtually impossible for the theory to be falsified by similar, general enough tests that we can imagine because uncountable tests of many diverse but similar kinds have already been successfully done. So while string theory is an immensely predictive theory - and more constraining than Quantum Field Theory - that doesn't give you any freedom about any qualitative aspect of physics and all of them could have failed, so string theory is surely a priori falsifiable, or falsifiable in principle, the tests haven't failed.

So string theory, much like Quantum Electrodynamics, has become unfalsifiable in practice simply because it's been established as a valid theory for certain questions - at least for all general questions concerned with the interrelations of gravity, other interactions, matter species, number of spacetime dimensions, their transmutations and emergence, equivalences, black hole interactions, information loss questions, and dozens of other basic concepts.

It has simply become extremely unlikely that the tests have worked by an "accident".

At the same moment, many other issues - usually those linked to the precise vacuum selection, but also some other issues - remain unanswered. So one can formulate various hypotheses about these more detailed questions - e.g. to try to guess the right vacuum or formulate new bizarre selection paradigms such as the anthropic principle - and try to see whether they predict something nontrivial and whether the predictions are right.

The alternative hypotheses attacking the questions that have not yet been answered have to compete; the criterion is their ability to nontrivially pass tests which are directly or indirectly linked to the empirical data. But hypotheses that contradict tests that have already been made are true underdogs: they're de facto born as dead babies.

But the same dynamics will always work in science. As some hypotheses are being established as the winners that are much more likely because they are much more compatible with the real phenomena than the competing hypotheses, these winning hypotheses immediately become unfalsifiable in practice because every sane scientist knows that the tests won't fail simply because very analogous - and at least morally equivalent - tests haven't failed in the past.

I find it flabbergasting that someone who could have received a PhD in physics can misunderstand these simple points about the very meaning of the scientific method.

And that's the memo.

3 comments:

  1. From the quote and the original post at the link I don't see where Hossenfelder said that a theory must have an actual falsification. I'm no physicist or scientist, so maybe you understand her underlying sense better than I do, but I didn't see it.

    I doubt that our horizons must stop expanding, and in that expansive sense I'm doubtful of the word "everything" in "Theory of Everything" if "everything" includes all phenomena that people ever could learn about during an indefinitely long run of science.

    But the defeatism of Hossenfelder and Gleiser seems out of left field. Why should any currently known physical phenomena necessarily forever resist unificatory understanding? That seems like Comte saying that we'll never know the chemical composition of stars. Events soon refuted him. Maybe soon Hossenfelder and Gleiser, too.

    Anyway, I wish Popper had not glued himself to the word pair "falsifiable" and "verifiable." He meant "testable for falsity" and "testable for truth." In ordinary English, "verifiable" amounts to "true and testable for truth," and "falsifiable" amounts to - well actually to most people it means "fakable." These senses of the words extend to their cognates in related languages. It's almost as if Popper didn't want the general public to understand him.

    And those who do bother to understand find themselves thinking in terms of "potentially verifiable" and "actually verifiable" and so on. It's needlessly vague and complicated, an indirect way to say things that one could say directly, e.g., "testable for truth."

    A true theory is, like you say, not actually disprovable since, being true, it's not provably false, and like you say it has a de jure but not de facto falsifiability. It's easier to say that it is still actually testable for falsity.

    I guess Popper wanted a word, not a phrase. What could he have said instead? English can be tough for word coinage. I've thought that "contratestable" (contra + testable) could be a word for "testable for falsity," but then "testable for truth" would be "protestable" and that already has another meaning!

    Well, one could coin words "procheckable" and "contracheckable." They aren't attractive words but they'd be lesser botches than Popper's stringent misuses of "verifiable" and "falsifiable" which harm the clarity of language.

    I like your blog a lot, Keep up the good work.

    ReplyDelete
  2. Popper's ideas are rather primitive and not a good ground for doing science. Einstein did not think the wave approach alone was enough to shore up quantum physics. (see the correspondence between him and Popper.)

    But are you talking about bad science or crackpot philosophy? The question of a TOE is not framed right is all. If we had a better unified theory it would be a meaningless question- string theory would not have its problems either for a unique theory.

    If there is a newer scientist that suggests there is no newest scientist- but I must admit unlike Noether certain PhD ladies may make good nuder scientists.

    ReplyDelete
  3. For a TOE there would be two separate possible falsifications, the first would be that is does not correctly describe all the physics we currently know, that might destroy the theory. The second would be if that TOE was valid for all the physics we currently know, but there turned out to be another particle or field it didn't consider, that would not kill the TOE, but the TOE wouldn't be a theory of everything anymore.

    In any case what physics mean by a theory of everything, is a theory of quantum gravity which unifying the known interactions. A unknown interaction might existing the universe which would the TOE, just a theory of what we previously thought was everything. The TOE is falsified as a theory of everything, but not as a good physical description what it previously described well.

    They may never be a time, when we can be sure they is nothing no unknown extras in the universe.

    ReplyDelete