Wednesday, January 02, 2008

Susskind in Wonderland

The following video is a good example of science when it encounters cargo cult science. Leonard Susskind attends an alternative talk about quantum gravity at the Perimeter Institute (12/06/2007).
Bianca Dittrich: Introduction to quantum gravity
Dittrich spends an equal time with causal dynamical triangulations as she does with loop quantum gravity and spinfoams. The latter approach is studied by 50 times higher number of papers than the former approach but this discrepancy is a purely sociological artifact. The actual scientific value of both approaches is equal, namely zero.

I would disagree with every single transparency - with her statements about the goal of the quantum gravity research, about the differences and common features of gravity on one side and other forces on the other side, as well as with her appraisal what works and what doesn't. But because her talk is 42-minute long, it would be too much stuff.

Her talk is a typical example of cargo cult science. They use words that sound scientific, they talk about experiments and about their excitement. Formally, everything looks like science. There is only one problem with their theoretical work: the airplanes don't land and the gravitons don't scatter. It is because they are unable to impartially evaluate facts, to distinguish facts from wishful thinking and results from assumptions, and to abandon hypotheses that have been falsified.

They play an "everything goes" game instead. In this game, nothing is ever abandoned because it would apparently be too cruel for them. They always treat "Yes" and "No" as equal, never answer any question, except for questions where their answers seem to be in consensus - and these answers are usually wrong.

During the talk, she is asked various questions - does it bother you that there are acausal influences? Does it bother you that you may get wrong values of the Bekenstein-Hawking entropy? And so on. Quite clearly, nothing bothers her. She doesn't really respond to any of these questions. A theorist in trouble could offer some handwaving and wishful thinking how to get rid of otherwise lethal problems. But these guys don't even seem to be bothered by any physical constraints.

Go to 42:20 of the video. Questions begin here.

The first guy says that at low energies, many quantities in quantized general relativity are divergent but the coefficients of the logarithmic divergences are independent of the UV theory. They must agree, as graduate students at good schools learn in basic courses. Can you check that these coefficients come out properly, he asks?

This question is clearly several categories above the knowledge and skills of the cargo cult quantum gravity theorists. Most of them don't believe there are any logarithmic divergences anywhere at all and they would surely set the coefficients to zero. Dittrich and others start to laugh, indicating that a sentence with a convoluted word such as "logarithmic" must surely be a joke.

Nevertheless, a curly-haired participant tries to answer the question - saying that the coefficients should be matched. The information value of his answer is manifestly zero. He doesn't say whether they match or what one needs to assume in order to make them match. These people are just completely incapable to solve any problems in physics because they are not even able to see them.

At 44:10, the curly-haired person is reminded that the question was about the logarithmic divergences so he honestly admits that he has no idea about these basic questions. He had only offered a general vague talking point about some divergences or some numbers.

At 44:18, Leonard Susskind asks the question how can one recover the Lorentz invariance which is very difficult in every kind of the Hamiltonian lattice gauge theory and their picture is an example of this framework. Incidentally, Leonard Susskind discovered the Hamiltonian lattice gauge theory. But his present role is different: he was paid a lot of money to listen to confused, dumb people who talk about a malfunctioning variation of his framework but who have no idea what they are doing.

Dittrich tries to answer and after several minutes of fog, she says that an anomaly-free evolution might end up with some kind of Lorentz invariance, either the exact one, or a deformed one, or a "slightly" broken one. Now, the word "slightly" is clearly a ludicrous and dishonest form of marketing because as Gell-Mann's principle and naturalness have taught us, there doesn't exist any reason whatsoever why the breaking should only be "slight" once it exists.

The word "slight" is just a convenient lie.

At 46:00, Susskind therefore immediately asks why it is "slight". Surprisingly, even Lee Smolin agrees that if the Lorentz invariance is "slightly" broken, the theory is dead. But Smolin tries to argue that if one tunes the theory's diffeomorphism symmetry to be anomaly-free, then the Lorentz invariance follows.

This is of course complete nonsense. Lorentz invariance doesn't follow and cannot follow from diffeomorphisms only. Let me just tell you four very different examples of theories that are diffeomorphism-invariant but not Lorentz-invariant to show you how incredibly sloppy their reasoning is.

(1) For example, you can have general relativity but the metric tensor may be positively definite. The spacetime is going to be Euclidean and the Lorentz symmetry will be replaced by the four-dimensional rotational symmetry. In fact, the Euclidean symmetry is much more likely to occur than the Lorentz symmetry in the context of a spinfoam that is, much like any form of luminiferous aether, incompatible with Lorentz symmetry as it always picks a preferred reference frame.

(2) But even such a Euclidean symmetry (a cousin of the Lorentz symmetry) can be completely broken in a diffeomorphism-invariant theory. Take a Lorentz-invariant theory and add a vector field (or any other tensor field, for that matter) with a potential that is minimized if the vector has a particular length-squared (or another appropriate invariant). Clearly, the vacuum will be Lorentz-breaking (spontaneously) but the theory can still be diffeomorphism-invariant.

(3) But I don't need these extra symmetry-breaking fields. In fact, a diffeomorphism-invariant theory doesn't have to include the metric tensor at all. Instead, it can have a tensor with four indices - Chris Hull would like these theories. For most choices of such tensors, it will be completely impossible to choose a local inertial frame. Alternatively, it can be topological.

(4) One more example. One can rewrite a classical theory with a Galilean symmetry in a diffeomorphism-invariant way, by adding auxiliary fields. And it will surely have no Lorentz symmetry.

At any rate, the statement that the diffeomorphism symmetry is enough to get the Lorentz symmetry is obviously and massively incorrect. Some people just don't appreciate how huge constraint the Lorentz symmetry actually is. The diffeomorphisms may reparameterize the coordinates in the same way as a global Lorentz transformation but if the vacuum is Lorentz-invariant, it must also be true that once you deal with the vacuum and go to a local inertial frame with the help of an appropriate diffeomorphism, all degrees of freedom take on Lorentz-invariant values. That's a highly non-trivial constraint. The vacuum must be empty - all kinds of objects and matter that might be predicted by your theory must disappear, in the relativistic sense.

For example, count how many additional terms you can add to the Standard Model if you drop the requirement of the Lorentz symmetry. Very many, even if you insist on renormalizability (counted according to the relativistic dimensional analysis) and the same field content. Otherwise, the number of all Lorentz-breaking terms is much more infinite than the number of Lorentz-preserving terms: the Lorentz-invariant terms are very special.

But even if you forget about this wrong implication, the assumption of the absence of anomalies is also very controversial and almost certainly wrong. In fact, it is one of the very goals of the master constraint program to deny that there can possibly be any anomalies in this framework. They like to "quantize" things in such a way that anomalies can't exist even a priori.

Around 47:00, Susskind asks a simple question whether they actually have a theory that is anomaly-free and Lorentz-invariant. They are simply not able to answer this simple question. They are afraid to say a clear "No" answer. Physicists - those without the "cargo cult" adjective - may disagree what is a natural extrapolation of the known laws to the case of the multiverse.

But they just cannot disagree whether a theory is Lorentz-invariant or not because it is a completely sharp technical question that always has a clear answer. Either one has a theory, or she has no theory. In the latter case, for example if she only has an undefinable hope that might be or might not be lost somewhere in an infinite-dimensional continuum of opportunities but no one knows how the place could be located, she shouldn't be talking about it at all.

Once she has a theory, there must exist a method to show the highly non-trivial statement that it is Lorentz-invariant. Incidentally, in all known cases, the right method is to show that the theory is equivalent to another theory that is manifestly Lorentz-invariant. If there is no such method, the theory is almost certainly breaking the Lorentz invariance. The opposite assumption amounts to a religious belief in the right value of infinitely many coefficients - an agreement that should occur for no good reason. It's just not a rational approach to make such assumptions. This rudimentary statement already seems too difficult or controversial for them.

At 48:10, Susskind also asks whether the cosmological constant is adjustable. Obviously, this question is ill-defined in the present context because the concept of a "cosmological constant" only makes sense within theories whose effective action can be organized as the Einstein-Hilbert action plus the rest (matter). But because the theories that are being discussed don't and can't reproduce general relativity, the concept of a cosmological constant is ill-defined. One can't even try to start to solve this problem.

Dittrich answers that the cosmological constant "depends on the framework" which is a euphemism for saying that it is impossible to give any meaningful answer. And she mentions a "framework" in which the cosmological constant is quantized; she won't tell you but in this framework, one uses a quantum deformation of the SU(2) group from loop quantum gravity (q is related to the cosmological constant) which is interesting but the whole context probably cannot work.

Susskind would like to know what the constant is and whether one may obtain the tiny observed value - i.e. whether these approaches could shed new light at the cosmological constant problem - but the reaction makes it obvious that none of them has ever thought about these basic things and furthermore, this fact doesn't seem to bother them at all. They don't seem to care about the cosmological constant problem, the value of the cosmological constant observed in 1998, or any other observations or scientific arguments. Their work is always motivated by an older confusion and (usually incorrect) philosophical preconceptions, never by actual problems in observable science or something that is connected to it. They live in a different world. They are just too dense.

But if an institute has a lot of money, it can employ a top physicist to listen to dense pseudoscientists. Note that I have politely used the word "employ" and not "corrupt". For progress in physics, it would surely be a better idea to employ all these Perimeter Institute "quantum gravity" people as additional secretaries of Susskind - or cooks in the restaurants where Susskind may spend his holidays - but unfortunately we don't live in a perfect world.

And that's the memo.

1 comment:

  1. I believe that the cosmological constant problem is a problem directly related to the reheating phase of inflationary cosmology. On the supercooled side there is a saturation production of micro black holes that produces a cosmological constant on the order of 10^123 (planck units) whereby all dimensions are curled up. On the downhill energy side of reheat the micro black hole production stops (but not before QFT states that the cosmological constant should be on the order of 10^120 for our vacuum density which is obviously wrong). Three copies of the Monster Group on a 22 dimensional manifold are required from initial conditions. One copy of the Monster symmetry is not protected by a Hilbert space while the other two copies are covered by a tensored N = 4 Super Yang-Mills represented as the number of vacua fields 8.0738*10^56: 196883^3 *22--> 196883*22^3 [8.0738*10^56 x 8.0738*10^56] = 1.36657*10^123 vacuum density (planck units). This gives us the idea why things are so curled up before reheat occurs. After reheat 4 dimensions become predominate. At the time of reheat the vacuum density calculates out: 196883*22^3/196883^1/3 *22 [8.0738*10^56 x 8.0738*10^56] = 1.06776*10^120 (planck units). Because each N = 4 Super Yang-Mills theory contains a copy of the Monster (196883 is lowest irreducible representation) we end up with two copies of the Monster possibly as N = 8 Supergravity theory operating within the Higgs hierarchal sector: 196883^2/3 * 196883 x 196883. Where 196883^2/3 is no longer a faithful representation of the Monster. We calculate close to the observed value (based on supernovae survey) in the following way. Close to initial conditions consider that the Planck density is the valid vacuum density: pp = Mp/lp3 = 5.1574 *10^93 g/cm^3.
    Before reheat and symmetry breaking the cosmological constant value is: Mp/lp3(1.36657*10^123/1.06776*10^120) = 6.007*10^96 g/cm^3
    After reheat: lambda = Mp/lp3 (1/8pi(1.06776*10^120)) = 1.922*10-28 g/cm^3

    As you can see the discrepancies are:
    1.06776*10^120 (what QFT says it should be but it is not)/1.36657*10^123 (curled up 22d + 4d) = 1/196883^1/3 *22
    The larger number 1.36657*10^123 ends up in the denominator due to a probability- inversion of graviton interaction at Reheat (aka GUT) so that probabilities do not become greater than one before the Planck energy is reached. We live in a “friendly giant” vacuum or friendly vacuum.