Not Even Wrong

A few months back I saw a call for papers for a volume on “Establishing the philosophy of supersymmetry”. For a while I was thinking of writing something, since the general topic of supersymmetry is a complex and interesting one, about which there is a lot to say. Recently though it became clear to me that I should be writing up other more important things I’ve been working on. Also, taking a look back at the dozen or so pages I wrote about this 20 years or so ago for the book Not Even Wrong, there’s very little I would change (and I’ve written far too much since 2004 about this on the blog). What follows though are a few thoughts about what “supersymmetry” looks like now, maybe of interest to philosophers and others, maybe not…

First the good: “symmetry” is an absolutely central concept in quantum theory, in the mathematical form of Lie algebras and their representations. Most generally, “supersymmetry” means extending this to super Lie algebras and their representations, and there are wonderful examples of this structure. A central one for representation theory involves thinking of the Dirac operator as a supercharge: by extending a Lie algebra to a super Lie algebra, Casimirs have square roots, bringing in a whole new level of structure to familiar problems. In physics this is the phenomenon of Hamiltonians having square roots when you add fermionic variables, providing a “square root” of infinitesimal time translation.

Going from just a time dimension to more space-time dimensions, one finds supersymmetric quantum field theories with truly remarkable properties of deep mathematical significance. Example include 2d supersymmetric sigma models and mirror symmetry, 4d N=2 super Yang-Mills and four manifold invariants, 4d N=4 super Yang-Mills and geometric Langlands.

But then there’s the bad and the ugly: attempts to extend the Standard Model to a larger supersymmetric model. From the perspective of 2023, the story of this is one of increasingly pathological science. In 1971 Golfand and Likhtman first published an extension of the Poincaré Lie algebra to a super Lie algebra. This was pretty much ignored until the end of 1973 when Wess and Zumino rediscovered this from a different point of view and it became a hot topic among theorists. Very quickly it became clear what the problem was: the new generators one was adding took all known particle states to particle states with quantum numbers not corresponding to anything known. In other words, this supersymmetry acts trivially on known physics, telling you nothing new. It became commonplace to advertise supersymmetry as relating particles with different spin, without mentioning that no pairs of known particles were related this way. In all cases, a known particle was getting related to an unknown particle. Worse, for unbroken supersymmetry the unknown particle was of the same mass as the known one, something that doesn’t happen so the idea is falsified. One can try and save it by looking for a dynamical mechanism for spontaneous supersymmetry breaking and using this to push superpartners up to unobservable masses, but this typically makes an already pretty ugly theory far more so.

The seriousness of this problem was clear by the mid-late 1970s, when I was a student. The one hope was that maybe some extended supergravity theory with lots of extra degrees of freedom would dynamically break supersymmetry at a high scale, leaving the Standard Model as the low energy part of the spectrum. There wasn’t any convincing way to make this work, and it became clear that one couldn’t get chiral interactions like those of the electroweak theory this way. 1984 saw the advent of a different high scale model supposed to do this (superstring theory), but that’s another story.

Looking back from our present perspective, it’s very hard to understand why anyone saw supersymmetric extensions of the SM as plausible physics models that would be vindicated by observations at colliders. For example, Gross and Witten in 1996 published an article in the Wall Street Journal explaining that “There is a high probability that supersymmetry, if it plays the role physicists suspect, will be confirmed in the next decade.” Ten years later, when the Tevatron and LEP had seen nothing, the same argument was being made for the LHC. After over a decade of conclusive negative results from the LHC, one continues to hear prominent theorists assuring us that this is still the best idea out there and large conferences devoted to the topic. Long ago this became pathological science. In the call for papers, the issue is framed as:

recent debates on the prospects of low energy supersymmetry in light of its non-discovery at the LHC raise interesting epistemological questions.

From what I can see, the questions raised are not of an epistemological nature, but perhaps the philosophers will find a way to sort this out.

Update: There was a workshop on this last year, abstracts here.

Update: I happened to come across today this 2021 interview of Daniel Freedman by David Zierler. Zierler repeatedly asks Freedman why he has faith in SUSY despite the long history of no evidence. Near the end, Freedman gives this very defensive explanation:

Zierler:

What I hear in your remarks is an adherence to supersymmetry despite its immediate experimental prospects. Is that a belief or is it something more?

Freedman:

It’s a belief which stems from confidence in the powerful symmetry which underlies the subject. Some human beings indulge in beliefs which have no basis whatsoever. Some of those beliefs are destroying our society at the moment. My belief in a credible and interesting physical theory isn’t going to hurt anybody.

This week and next there’s an interesting summer school going on at the IAS, with topic Understanding Confinement. Videos of talks are available here or at the IAS video site.

Taking a look at some of the first talks brings back vividly my graduate student years, which were dominated by thinking about this topic. When I arrived in Princeton in 1979, the people there had been working for several years on trying to understand confinement semi-classically, in terms of instantons and other solutions to the Yang-Mills equations (e.g. merons). By 1979 it had become clear that such semi-classical calculations were not sufficient to understand confinement and people were looking for other ideas. There were quite a few around, including the idea that there was some sort of string theory dual to pure Yang-Mills theory, and I spent quite a lot of time reading up on efforts of Migdal, Polyakov and others to find a formulation of string theory that would provide the needed dual. I ended up writing my thesis on lattice gauge theory, an approach which had the great advantage that you could at least put the calculation on a computer and start trying to get a reliable result for pure Yang-Mills numerically. Some of the calculations I did were done at the IAS, with Nati Seiberg and others. The other thing I spent a lot of time thinking about was how to put spinor fields on the lattice, the beginning of my interest in the geometry of spinors.

I strongly recommend watching Witten’s talk on Some Milestones in the Study of Confinement. His career started a few years before mine, with the early part very much dominated by the problem of how to make sense of Yang-Mills theory non-perturbatively, and this has has always been a motivating problem behind much of his work. In his talk he explains clearly the approaches to the problem (lattice gauge theory, 1/N, dual Meissner) that appeared very soon after the advent of QCD in 1973. He emphasizes how each of these approaches shows indications of a possible string theory dual, while frustratingly not leading to a string model that has the right properties, summarizing (41:30) the situation with:

The string theory we want is probably quite unlike any that we actually know, as of now. We don’t know how to make a string theory with the short distance behavior of asymptotic freedom.

In his talk he discusses later developments, in particular the Seiberg-Witten solution to N=2 SYM and the AdS/CFT duality between a string theory and N=4 SYM, explaining how these advances still don’t provide a viable approach to the confinement problem in pure Yang-Mills.

I’m looking forward to seeing the rest of the talks, and finding out more about some things that have happened over the years since I was most actively paying attention to what was happening with the confinement problem.

A few days ago I read a fascinating article in New York magazine: Inside the AI Factory, which explained how the very large business of hiring humans to do tasks that generate training data for AI works. One reaction I had to this was “at least this means math is safe from AI, nobody is going to pay mathematicians to generate this sort of training data.” Yesterday though I ran across this tweet from Kevin Buzzard, which advertises work (see this link) that seems to be of this kind.

A company called Prolific is advertising work paying 20-25 pounds/hour doing tasks in Lean 3. This company is in exactly the business described in the NY mag articles, hiring people to do tasks as part of “studies”, which often are generating AI training data.

One unusual thing about this whole industry is that if you sign up for this work you often have no idea who your employer really is, or what your work will be used for, and you sign a non-disclosure agreement to not discuss what you do. In this case, a few things can be gleaned from discussions on the Lean Zulip server:

• “it’s 40 dollars/hr now actually”
• “I think everyone signed a consent form preventing them from disclosing any problems or even their participation.”
• One problem was formalizing a proof of sin(x)=1 implies x=pi/2 (mod 2pi).
• A representative of the company writes: “Some of our biggest partners are keen to work with Lean users because of its applicability as a theorem prover. They plan to launch numerous studies that require participants to have either a working or expert knowledge of Lean 3.” By “biggest partners”, presumably this is Microsoft/Google or something, not some small publishing organization.

If anyone knows anything about what’s up with this incipient possible math AI factory, please let us know. On the more general issues of math and AI, I’d rather not host a discussion here, partly because I’m pretty ignorant and not very interested in spending time changing that. The situation in mathematics is complicated, but as far as fundamental physics goes, theorem-proving is irrelevant, and applying “big data”/machine-learning/AI techniques to generate more meaningless calculations in a field drowning in them is pretty obviously unhelpful.

For a much better place to read about what is happening in this area, there’s an article in today’s New York Times by Siobhan Roberts: A.I. Is Coming for Mathematics, Too. At Silicon Reckoner, Michael Harris is in the middle of a series of posts about this past month’s workshop on “AI to Assist Mathematical Reasoning.”