Thursday, April 30, 2009

NYU about Hořava-Lifshitz gravity

Keith Chen (click) wrote a rather illuminating story about the fate of Hořava-Lifshitz gravity (click and read this text first!) at the NYU.

Journal clubs are apparently alive and well over there. I think that they're important for the health of science as seen from a broader scientific community's viewpoint. I found the "[postdoc] journal club" to be one of the most satisfactory formats for creative scientific confrontation. All ideas and criticisms by sensible and quasi-sensible people can be raised. They can be debunked, too.

Andrei Gruzinov, a professor of theoretical astrophysics, wanted this paper (or these papers) to be analyzed - because of those "cool properties" that have been said about it (including its being a theory of quantum gravity and a replacement for inflation), mostly in dozens of Asian preprints. Andrei Gruzinov asked Matt Kleban to present the paper in the journal club.

Their analysis focused on the question whether the heavily nonrelativistic "z=3" theory in the ultraviolet can flow to the relativistic "z=1" theory in the infrared. See also a later article,
Can Hořava gravity flow to Einstein gravity?
Massimo Porrati gave some arguments why it was not possible. Incidentally, although he hadn't read the whole paper(s), Porrati claimed that if the paper were right, he would switch his field to swine flu. ;-) At the end, the previously excited people considered the attempt to be a failed one and they have lost interest in it.

So did I. Let me summarize the main problems:
  • in the IR, normal GR should be reproduced; but this effective field theory needs all the spacetime diffeomorphisms to be gauged, in order to eliminate all unphysical polarizations of the waves; because the spacetime diffeomorphisms are broken, one can't get rid of all the unphysical states; after this blog article was written, a paper by Charmousis et al. was released; it explains the appearance of this unwanted (and strongly coupled) mode at long distances in detail (the second, more serious problem they find); the experiments indirectly observing gravitational waves etc. seem to rule out Hořava's theory
  • it is questionable whether the very qualitative flow from "z=3" to "z=1" is possible, starting from a small perturbative deformation in the ultraviolet
  • even if it were possible, one would need a huge amount of fine-tuning to restore the Lorentz symmetry in the IR (so strongly constrained by the observations), especially once all the matter fields are added
  • the Lorentz symmetry violation in GR leads to many likely inconsistencies, including the possibility to construct perpetuum mobile gadgets of the second kind, at least in principle
  • even the very scale invariance of the hypothetical UV theory doesn't seem to have been demonstrated beyond the classical level
  • in the fast comments, it is argued that the correct, ordinary black holes with an area-extensive entropy probably can't be created in this theory; local theories like this one probably always predict a volume-extensive entropy; near the would-be event horizons, the deviations of Hořava's theory from GR are probably huge (e.g. Lorentz violations get amplified) and prevent the normal black holes from creation
In my opinion, these are big problems and every single one of them would be enough to kill my interest in a theory. Their combination is even more lethal. If you have something really substantial to say about those technical problems, you're welcome.

Philosophy of science

Meanwhile, I want to say some important things about the philosophy and sociology. I find the "postmodern" opinions of many young people, including those whom I don't find personally unpleasant, extremely worrisome. These attitudes are doubly sad if they appear at NYU, a place that Alan Sokal transformed into a bastion of the war against the postmodern nonsense. For example, Keith writes:
So it was pretty funny that last week everybody was excited about this paper and this week everybody loses interest in it. We have to wait for another breakthrough.
Well, this is how science works. We can never be sure that we're right: we can only be sure if we're wrong. Theories are being abandoned once they disagree with some aspects of the observations. That's what has apparently happened with this attempt, too. The theories that matter are those that can survive some tests - for some time that can be either long or very long. At the end, more demanding experiments find some problems, too (with the final theory of everything being the only possible exception).

Indeed, it often takes a two-hour journal club to settle such things and kill a hypothesis. Two hours is not a microscopic amount of time. That's because Petr Hořava is a thoughtful physicist. Similarly ambitious, average papers written by less thoughtful physicists than Petr take about 5 minutes to be killed - or 4 minutes, ask Lee Glub Glub Glub Smolin at the bottom of the sea for the most up-to-date timing. :-)

Keith adds a few comments:
Nobody really knows how to come up a workable quantum theory of gravity.
This sentence reveals Keith's complete misunderstanding of the last 35 years (OK, 34.5) in theoretical high-energy physics. But let's focus on the philosophy here:
So anybody can take strong opinion in this regard. But anybody that asks the right question is much much more likely to succeed than others.
Except that the journal club has led to the conclusion that the right questions have not been asked in this paper. "Right questions" are not the same things as "any questions" - despite the fact that Keith, as a skillful propagandist, composed his sentence in such a way that it is almost impolite to point out an obvious fact such as the beginning of this sentence. ;-)

Well, I don't find it impolite. Quite on the contrary: I think it is essential for scientists not to be manipulated by rhetorical exercises of others into fear to point out the obvious.
But shouldn't it be obvious that anything that is obvious is not going to work. Anything that is going to work is not going to be obvious.
Well, more precisely, anything that is going to succeed in the future is going to be slightly or very different from the things that have been tried and that have failed in the past. The previous sentence is a tautology.

But whether the new things will be obvious depends on a subjective judgement. For example, I think that the Hořava-Lifshitz Lagrangian - or any similar Lorentz-violating attempt to circumvent the failures of Lorentz-invariant local field theories of gravity - is obvious. The only problem is that it seems to be wrong, too.

Whether or not things look "obvious" to you depends on your definition of "obvious" and your emotions, education, and experience. The discoverers usually find their ideas more "obvious" than others. Moreover, the more experienced the other people will become, the more "obvious" those theories will look to them. But that's not what science is all about. Science is about hypotheses' being right or wrong, not about hypotheses' being "obvious". The latter should be left to the educators who decide how much time they should spend with XY at schools - or to the journalists who try to hype some mind-boggling results. It has nothing to do with the scientific research.

But Keith's ideas get much worse:
The reason that those smart people fail to come up with a sensible theory of quantum gravity maybe because they are too strongly biased. I defend this theory does not mean that I believe that this the right theory.
Again, the statement that we don't know the right theory of quantum gravity is a misunderstanding of the last 35 years in theoretical high-energy physics. But the philosophical background behind the two sentences is even sicker.

In science, it is simply not possible to defend a hypothesis or a statement without rational arguments. You may feel compassionate when hypotheses are being killed just like cars that are being made in Henry Ford's flow production (or even in Chrysler that is just sadly filing Chapter 11). But you're not acting as a scientist if you allow these emotions to influence your decisions because the falsification of hypotheses is the most basic and most paramount among all tools that science uses and has to use.

In science, one shouldn't really defend theories against falsification. And one shouldn't defend theories that he believes to be incorrect (because of rational reasons), either. The closest thing one can do is to show that the would-be falsification is actually incorrect (or at least less complete or less certain than previously thought). For example, chiral superstring theories were believed to be dead in the early 1980s because they were thought to be spoiled by anomalies.

Green and Schwarz believed otherwise. But they actually had to make very extensive, detailed calculations showing that the "incomplete" proof of the nonzero anomalies (accepted by most physicists in the field) was actually incorrect and that all the anomalies canceled when calculated exactly. Without this result, the first superstring revolution of 1984 couldn't have taken place. And indeed, it would have been profoundly irrational to switch to a theory that was believed, for pretty good (but not quite good) reasons, to be anomalous. You know, it doesn't matter how beautiful a theory is or what the name of the author is.

The calculation of the anomaly cancellation was no detail. It was one of the key results deciding whether the theory was right or wrong. And by all these criteria (and according to hundreds of other tests that have been successfully made between 1984 and 2009), superstring theory turned out to be right, at the end. But if you have a "similar" hypothesis without such an anomaly cancellation - an argument showing that the "falsification" is actually incorrect - you simply can't assume that the cancellation is a "formality". It's not. A hypothesis is dead unless and until a mistake is found in the falsification.

In our postmodern reality, some people are even bothered by the fact that "ideas of a certain type" are constantly being falsified and abandoned. This type of compassion is even worse because what they consider "sad injustice and discrimination" is called "general results and general insights" in science. And they are actually a very good thing!

Scientists have made many mistakes in the past. While many of them failed to appreciate the principles and methods they should have learned, many others have been way too dogmatic. They didn't want to listen to a certain new type of evidence. They often needed a much longer time to accept a new, demonstrably better theory than the time that the ideal scientists would have needed.

Geologists needed decades to accept Wegener's crisp arguments in favor of the continental drift. Some physicists have needed (and, indeed, will need) decades and thousands of failed "alternative" attempts to understand why string theory is inevitable in the context of quantum gravity. But despite all the scientists' imperfections, it must still be rational arguments - sometimes more extensive and accurate arguments than the ideal scientists would find necessary - that can change the ideal scientists' opinions as well as the real scientists' opinions.

If a scientist needs too much time to understand a new theory - and why it is better than the old one - he is an imperfect scientist. But if he is generally deciding according to emotional rather than rational criteria, or according to quotas rather than calculations and observations, he is not a scientist at all. That's much worse. Imperfections slow science down but they are largely inevitable given the finite intelligence and limited integrity of the humans. However, dominant emotional criteria kill science.

Everyone who wants to fill science with emotional arguments and criteria that should have the same impact as technical arguments - but these technical arguments actually don't exist - is killing science and replacing it with politics and superstitions. Just like the real market imperfections can't be "cured" by communism, the imperfections of the "market of ideas" can't be cured by the postmodern feminist pseudoscience. The only thing that these "cures" can do is to kill the prosperity and/or science.

So what Keith calls "bias" is actually nothing else than the scientific integrity itself. Keith adds:
OK, Hořava did not lay out the model in details.
Well, I actually think that he did. He has made something that is close to the best possible case for this idea. That's also the main reason why Petr's original papers are more widely cited than any of their followups.
But I think he has set up a proposal that is interesting enough to pursue it further.
This is exactly what the NYU professors did, isn't it? This is why they ordered Matt Kleban to present the paper in the journal club. Keith Chen may dislike the outcome. But in science, what matters are the correct outcomes, not outcomes that someone likes.

And one simply cannot write papers with pre-determined "signs" of the results. More precisely, Keith (or the authors in Beijing) may be motivated (and is free) to write "revolting" papers of a certain kind - but whether or not such a "revolt" has actually led anywhere must be decided by the scientific criteria again, not by unfairly pre-determined labels of "bias" and "worthy research" reserved for results with particular "signs". And whoever cares about "revolts" more than he cares about rational arguments is likely to produce random gibberish, at least in the long run.
One of the reason that so many people have jumped into this theory because they think that it is interesting and worth trying.
This sentence shows Keith's inherent inconsistency and bias. When it comes to the NYU people, he thinks that their opinions are "biased" and something is wrong with the sociology: people can't be trusted. But when it comes to some Asian authors of not-quite-penetrating followup papers, Keith decides to praise them and to repeat some uncritical hype. Where does the different attitude come from, Keith?
So my philosophy to these kind of difficult open questions is that we should keep mind opened.
Well, the disadvantage of replacing science with an "open-mind philosophy" is that you won't be able to do any science - because every scientific result and every event in science that matters is closing a small hole in the mind (and, sometimes, opening new holes in a piece of mind that was previously unknown). Every result is filling a hole in our Swiss cheese of knowledge and ignorance. Being constantly open-minded about everything is exactly equivalent to the ban on any progress in science, Keith. Opening one's mind entirely also makes it possible for the mind to completely evaporate. ;-)

"Fortunately" for Keith, it's not what he's doing. He's keeping his mind about many key issues - such as our knowledge of a consistent theory of quantum gravity - hermetically closed: he has only been closing the holes according to very different criteria than the rational, technical arguments that his senior colleagues at NYU prefer and that science is based upon.

And that's the memo.


  1. hello... hapi blogging... have a nice day! just visiting here....

  2. Luboš thanks a lot for your enlihtening blog about cutting edge physics take care man.

  3. Could you possibly elaborate a bit on:

    "I think that the Hořava-Lifshitz Lagrangian - or any similar Lorentz-violating attempt to circumvent the failures of Lorentz-invariant local field theories of gravity - is obvious. The only problem is that it seems to be wrong, too."

    Before you try answering please: add extra dimensions into the game, break the diffeomorphism only in the extra dimension, leave intact the ordinary dimensions, fix the lagrangian in UN such that it is free of the problems associated to the fourth order derivative action in the UV point, fine tune the extra dimensions and ordinary dimensions such that there exists the same numbers of degrees freedom in UV and IR point of your construction, make Newton's constant in UV dimensionless. All these can be done. They also resolve most of the problem that you have pinpointed to at the beginning of your writing

  4. Dear Led Zeppelin, thanks for your interest and kind words.

    Dear Qasem, thanks for your question but it's not quite clear whether you really want me to "elaborate" or whether you just want me to confirm that your framework is physically consistent.

    If it is the latter, then I will disappoint you. What you write cannot produce a consistent theory of gravity.

    Diffeomorphism symmetry - in all dimensions - is paramount for getting rid of (all) the unphysical polarization of the graviton. It's a gauge symmetry so it can't be broken. Whenever there is anything such as the metric, and a locally Lorentz-invariant action, it must be possible to find the diffeomorphism-invariant description for the wrong modes to go away.

    Lorentz invariance is really a "global" subgroup of the diffeomorphism symmetry at Minkowski backgrounds - so it must hold, too. But if you wanted something "more general", it's still true that it has been shown by direct experimental tests that the Lorentz symmetry holds up to the Planck scale - or, to say the least, generic first-order corrections that you get from any generic theory breaking the Lorentz symmetry at the Planck scale have to have coefficients that are unnatural.

    You: "They also resolve most of the problem that you have pinpointed to at the beginning of your writing."

    Well, I don't know how you measure "most", but whatever the method is, it's not good enough. Violating a single physical consistency criterion is enough for a hypothesis to die an instant death. So things that you write down can perhaps be "done", whatever you mean by "doing", but the result is not a viable theory of physics.


  5. Dear Qasem, there's a new preprint (click) today that studies the so-called "healthy" versions of Hořava's theory.

    They Stuckelbergize the GR and study the decoupled Stuckelberg mode.

    The conclusion is that the strong coupled problem plagues even the "healthy" theory, suggesting non-renormalizability. Moreover, the theory violates the equivalence principle and suffers from the lack of the Vainshtein effect.

    But I am not sure whether you're interested in these lethal flaws. You may only be interested in the "majority" of the optimistic hints and ignore anything else, in order to preserve your belief that it's a "majority".

    I am personally interested in science where most hypotheses are wrong, and I want to know which of them are right and which of them are wrong. This one is wrong.