Tuesday, August 05, 2008

Tevatron falsifies Connes' model of physics

This is just way too funny and I can't resist.

Yesterday, Tevatron issued a
press release (see also Google News)
explaining that they combined the work of the two teams, CDF and D0 (a rare policy they have followed a decade ago to discover the top quark, too), and concluded that at the 95% confidence level, the Higgs mass cannot be 170 GeV.

Now, this sentence may sound very bizarre. What do they exactly mean by "170 GeV"? Do they mean the exact number? (See the picture above to appreciate the actual excluded interval: it is tiny, indeed.) A person would have to be extraordinarily hapless to predict the Higgs mass to be exactly 170 GeV, right?

It is very unlikely that someone would become the author of the first model that is falsified so explicitly. He or she may have been hit by an asteroid, too. 170 GeV is in the middle of an interval favored by various classes of models (although way above the ranges that are believed to be relevant by the newest high-precision data) but could someone be that unlucky? :-)
Update: In 2009, the Tevatron will extend this analysis and exclude the 160 GeV - 170 GeV window at the 95% confidence level.
(If you want to know, 170 GeV is not so random. It is the mass for which the quartic coupling in the Standard Model is exactly high enough for the Landau pole to appear relatively soon, namely near the GUT scale. In other words, the Higgs mass has to be below 170 GeV for the Standard Model to be well-behaved up to the GUT scale.)

Anyway, once again, could someone be so unlucky?

Take a random person who has claimed to be able to predict the masses, for example a great man and mathematician, Alain Connes. ;-)
Connes' predictions for the masses (click)
What is his figure for the Higgs mass? Well, in their 2006 paper, they "determined" the Higgs mass to be 170 GeV. This is what I call "bad luck". Alanis Morrissette should add Alain Connes to her "Ironic". :-)

It's great to build falsifiable theories (and to make guesses about the masses based on naive, random, and rationally unjustified formulae, which is what Connes et al. really did in this case) but you shouldn't forget that such falsifiable theories suffer from a kind of insecurity: they can actually be falsified. The more hopeless a wrong theory is, the shorter time it typically takes to falsify it.

And in some unhappy cases, that may occur much earlier than you could have thought at the beginning. In Lee Smolin's case, it occurs within 5 minutes after the first different physicist opens any paper written by Smolin: they go glub glub glub to the bottom of the sea. In Alain Connes' case, it can take two years.

But it's still damn too fast. Who could have thought that we won't need the LHC to settle such questions? Good bye, the standard model derived from a new kind of noncommutative geometry, an "alternative to string theory". ;-) We loved you so much - even though we never quite understood how you could ever possibly derive anything that cannot be derived from the normal standard model. ;-)

See also Resonaances explaining why it was unlikely that physics would learn anything about the SM and beyond from Connes' approach and why predictions such as 170 GeV were visually captivating but not serious even before they were falsified.

A noncommutative geometry blog reacts to the Fermilab news by a "profound discouragement" that they should have felt years ago after reading my explanations but they chose not to listen. That blog thinks that the event is really serious because it even translates wise verses by Lucretius to English:
Pleasant it is, when over a great sea the winds trouble the waters, to gaze from shore upon another's tribulation: not because any man's troubles are a delectable joy, but because to perceive from what ills you are free yourself is pleasant.
Oops. Only later, when I updated this text, I realized that the blog is the blog of no one else than Alain Connes himself. Who else than this big mind could refer to Lucretius! :-)

Lucretius had to be great but as a physicist, I would choose a completely different text for this nice opportunity - Feynman's "The Key to Science" was designed exactly for similar opportunities: :-)

(Incidentally, all Feynman fans should watch "Take the world from another point of view" from 1972, 4 parts per 10 minutes. There are fascinating stories about curiosity, brushing the teeth, quarks before QCD was fully settled, etc. Feynman also talks to a very stupid journalist who thinks that scientists can only talk to each other because they read the same magazines with the same gossip haha. Great to see that idiots have been on Earth at least for 35 years!)

The purpose of science is to look for correct answers to questions about Nature rather than for "any predictions" or even for "wrong predictions". Once we know that a theory is wrong, its inherent value for science is exactly zero. Falsification is the very end of the story. This is a very basic point about science that the Woit-like idiots will never understand.

Well, now everyone is liberated from this pseudo-noncommutative unnecessary superconstruction. Many people are still non-liberated from loop quantum gravity, doubly special relativity, variable speed of light etc. etc. but the brightest and fastest ones among them may get liberated in a few months or years, as smoothly as Alain Connes. But I actually don't believe that there are too many alternative physicists who are as fast and as honest as Alain Connes.

(Incidentally, when I mentioned the Landau pole appearing near the GUT scale for the 170 GeV mass, I am not 100% certain that Connes realizes this fact about the running couplings. Nevertheless, I assure you that the knowledge of this fact and similar facts is absolutely crucial for anyone who tries to make well-informed guesses about the Higgs mass.)

N.C.G. R.I.P.

Experiments are very healthy

I am looking forward to the new inflow of data from the LHC and elsewhere: experimental data are extremely healthy. Also, I want to see those 99% of model builders who have been showing their "courage" by writing down thousands of random, rationally unjustifiable guesses about things that couldn't possibly be predicted in this new context.

The overpopulated mutated bacteria of phenomenology - randomly engineered models and their mostly cheap modifications - will be mostly eradicated: 99% of existing hep-ph papers will become strictly irrelevant and higher standards will return to science. People will actually realize that making predictions about hard-to-predict questions - something that looked so easy and free at some moment (and certain people love prophets so much!) - actually costs something. And perhaps, most people will prefer to say "I don't know" about questions that they can't answer, instead of emitting "courageous" but random and rationally unsubstantiated guesses.

And maybe, scientists will again be allowed to say "I don't know, there are many possible answers" when they don't know and when there are many possible answers, a comment that they were brutally criticized for by the Woit-like scum just a year ago. It doesn't frighten me not to know certain things and it would be great not to be frightened by those who are frightened by ignorance. ;-)

In particle physics, it ultimately works in this way because solid arguments and observations can actually be collected and they play a very important role in determining which ideas are valuable and which ideas are not so valuable. I am afraid that e.g. climate science won't easily return to the situation in which the data actually matter. They have already crossed the Rubicon behind which it is more important to be a prophet than to be right.


  1. If only to fix ideas... Why do you think that Connes model is/could be a truncation of string theory? Do you have some conjecture to explain the lack of SUSY? Any idea about how the web of string dualities fits in the NCG scenario?

  2. Dear Alejandro, I was asked the question about NCG as a truncation of string theory in the fast comments, look there.

    I don't know what "lack of SUSY" you're talking about. I think it is not a settled question and the recent Tevatron data lead me to believe that SUSY at the LHC is more likely than 50%. I raised my guess to 60% and incidentally, I was surprised to see the exact same figure at a recent Cosmic Variance posting.

    Most stringy dualities are clearly invisible to a truncation of the theory, so if you interpret NCG as the truncation of the string compactifications to the zero modes, NCG will have nothing to say about dualities, except for trivial comments.

    For example, if the same NCG model would represent two descriptions that would be mirror-symmetric to one another, it would "have" mirror symmetry in it. But you can't really decompactify the finite-dimensional non-commutative manifolds so you can't map them to the normal Calabi-Yaus.

    You probably overestimate what I meant by the comment. What I really meant was to invite Alain Connes to learn string theory properly - the NCG picture has some relevant tools but it is of course a very small fraction of what is needed to make realistic model-building including quantum gravity.

  3. Dear physicist with an Arab name who tried to post here,

    sorry, I couldn't approve your message because it would be introducing emotions that have absolutely no room in science, especially not in this very technical discipline.

    With the current understanding, I don't think that there was anything "beautiful" about the theory, and even if it looked so, it would be completely irrelevant because science has different goals. Play the Feynman video to get a hint.

    I am always jubilant in similar situations because whenever the truth about a question becomes crisp and transparent, I am jubilant. That's how I am. I prefer the naked truth over irrational brainwashing and over collective celebration of beautiful emperor's clothes that don't really exist.

    It was Alain Connes who was recommended to learn string theory and I think it is an excellent idea. I didn't tell it to you, which would probably be a less good idea.

    Incidentally, 1982 is 26 years ago. You surely don't expect sensible people to judge the value of a recently formulated theory by ad hominem episodes that are moreover separated from the present era by 26 years, do you?

    All the best

  4. Lubos, I am looking for some scanned copy of Witten's talk in Shelter Island II. You know, the one which did the turn away from Kaluza Klein gauge groups and towards embracing other alternatives: SO(32), E8xE8 (and, lately, D-brane junctions). Do you happen to have a copy?

    THis is most of a private request, but I am pasting exactly in this thread because in some sense Connes "truncation" is a bypass of that theorem. Note that Alain deduces independently the need of 2 mod 8 (claimed in that paper, according Castellani book) but then it gets a chiral theory. Note that a precondition for it to happen was a non null index for the Dirac operator.

  5. I don't have any scanned copy but I am also sure that it is ludicrous to ask for a secret 1985 talk at some island because what you want to be explained is elementary material explained in almost every textbook of string theory and every textbook of representation theory of Lie groups.

    There is absolutely no problem to get a chiral theory by compactifying a 10+8k dimensional theory to 4D: it's exactly the right dimension where a chiral 4D theory is possible. Because in 10+8k Lorentzian dimension, one has Weyl Majorana (chiral real) spinors, 16, and they get decomposed to (2,4)+(2bar,4bar) under SO(3,1) x SO(6), which links the 4D chirality (2 vs 2bar) to the 6D chirality (4 vs 4bar).

    Note that it is essential for the compact manifold to be even-dimensional, otherwise there is no chirality there, and it must be 2+8k or 6+8k to get complex reps that can be linked to the complex reps of SL(2,C) = SO(3,1) (locally): the other even dimensions have real or pseudoreal spinors and couldn't lead to a chiral theory either.

    The other possibility to consider is from 6+8k dimensions to 4D. There might exist some more elementary explanation why it's not possible, but surely 2 hidden dimensions are too simple, and only torus (full SUSY) preserves some SUSY. 6+8 = 14 dimensions is already too much, not admitting a physical supersymmetric theory. Even without SUSY, you could have problems.

    Connes' dimensionality is exactly the same one as in string theory, mod 8.

  6. Well it happened to be not so secret at all; it is reprinted in a old Appelquist-Chodos-Freund volume on Kaluza Klein theories which a friend happened to have nearby.

    The problem is deeper than having chiral fermions: once you have zeroed in 2+4k, "the question is whether V-A gravity can reduce to V-A weak interactions in 4 dimensions"

    It seems that the pivotal point, at that time, was a non-go theorem from Atiyah-Hirzebruch: "that for any continuous symmetry group, abelian or non abelian, the Dirac zero modes form a real representation". Witten was able to expand this theorem to cover the Rarita-Shwinger operator, partly. Then sect V of the paper nails the coffin for Kaluza Klein and section VI opens the chest of elementary, no KK, gauge fields in the 4+n dimensional space... and the topological justification of family multiplicity: "The Dirac operator becomes the d and d* operator on differential forms, and the number of zero modes (weighted by chirality) is the Euler characteristic".

    Really an explanation from Castellani et al. book, in addition to your argument, in addition to my own generally known credulity, are enough to acknowledge what is happening. But I was interested on some perspective too. With that paper, Witten is told to have fulfilled the other role of a lamplighter: definitively lighting off the Kaluza Klein lamp. It proves that a SM-like gauge group can not appear from the isometries of a compact space.

    It seems that the 2 mod 8 argument was first exposed by Wetterich in NucPhysB 222, p 20 (1983).

    Now, what amazes me is that there seems to have no problem in the limit of infinite mass of W and Z, where the standard model becomes again vector-like, and where nine dimensions are enough to have a group SU(3)xU(1) as isometries of some 5 dimensional manifold. One could have taken some time trying to grow from this nonproblematic nine dimensional theory to some 10 or 11 dimensional one.