For quite a few years now, I’ve been mystified about what is going on in string theory, as the subject has become dominated by AdS/CFT inspired work which has nothing to do with either strings or any visible idea about a possible route to a unified fundamental theory. This work is very much dependent on choosing a special background, in tension with the idea that, whatever string theory is, it’s supposed to be a unique theory that relates all possible backgrounds. This issue came up in a discussion session at Strings 2021, and it turns out that others are wondering about this too. There’s this today from Lubos Motl:

Aside from more amazing things, the AdS/CFT correspondence became just a recipe for people to do rather uninspiring copies of the same work, in some AdS

_{5}/CFT_{4}map, and what they were actually thinking was always a quantum field theory, typically in D=4 (and it was likely to be lower, not higher, if it were a different dimension!) whose final answers admit some interpretation organized as a calculation in AdS_{5}. But as Vafa correctly emphasized, this is just a tiny portion of the miracle of string/M-theory – and even the whole AdS/CFT correspondence is a tiny fraction of the string dualities.This superficial approach – in which people reduced their understanding of string theory and its amazing properties to some mundane, constantly repetitive ideas about AdS/CFT, especially those that are just small superconstructions added on top of 4D quantum field theories – got even worse in the recent decade when the “quantum information” began to be treated as a part of “our field”. Quantum information is a legitimate set of ideas and laws but I think that in general, this field adds nothing to the fundamental physics so far which would go beyond the basic postulates of quantum mechanics…

When Cumrun correctly mentioned that the real depth of string theory is really being abandoned, Harlow responded by saying that there were some links of quantum information to AdS/CFT, the latter was a duality, and that was important. But that is a completely idiotic way of thinking, as Vafa politely pointed out, because string theory (and even string duality) is so much more than the AdS/CFT. In fact, even AdS/CFT is much more than the repetitive rituals that most people are doing 99% of their time when they are combining the methods and buzzwords of “AdS/CFT” and “quantum information”. Many people are really not getting deeper under the surface; they are remaining on the surface and I would say that they are getting more superficial every day.

According to Lubos, he’s not the only one who feels this way, with an “anonymous Princeton big shot” agreeing with him (hard to think of anyone else this could be other than Nima Arkani-Hamed):

There is a sociological problem – coming from the terrifying ideological developments in the whole society – that is responsible for this evolution. I have been saying this for a decade or two as well – and now some key folks at Princeton and elsewhere told me that they agreed. The new generation that entered the field remains on the surface because it really lacks the desire to arrive with new, deep, stunning, revolutionary ideas that will show that everyone else was blind. Instead, the Millennials are a generation that prefers to hide in a herd of stupid sheep and remain at the surface that is increasingly superficial…

So most of the stuff that is done in “quantum information within quantum gravity” is just the work of mediocre people who want to keep their entitlements but who don’t really have any more profound ambitions. As the aforementioned anonymous Princeton big shot told me, their standards have simply dropped significantly. The toy models in the “quantum information” only display a very superficial resemblance to the theories describing Nature. That big shot correctly told me that in the early 1980s, Witten was ready to abandon string theory because it had some technical problems with getting chiral fermions and their interactions correctly.

Harlow says that many of the people – who may be speakers at the annual Strings conference and who may call themselves “string theorists” when they are asked – don’t really know even the basics of string theory. And they can get away with it. Just like there is the “grade inflation” and the “inflation of degrees”, there is “inflation in the usage of the term string theorist”. Tons of people are using it who just shouldn’t because they are not experts in the field at all. Harlow said that many of those don’t understand supersymmetry, string theory etc. but it’s worse. I think that many of them don’t really understand things like chiral fermions, either. It’s implicitly clear from the direction of the “quantum information in quantum gravity” papers and their progress, or the absence of this progress to be more precise. They just don’t think it’s important to get their models to a level that would be competitive with the previous candidates for a theory of everything – like the perturbative heterotic string theory, M-theory on G

_{2}manifolds, braneworlds, and a few more. They are OK with writing a toy model having “something that superficially resembles a spacetime” and they want to be satisfied with that forever.

I don’t want to start here an ad hominem discussion of Lubos and his often extreme and eccentric views. On the topic though of the devolution of string theory as a TOE to playing with toy models of AdS/CFT using quantum information, it seems quite plausible that not only the “anonymous Princeton big shot” but quite a few other theoretical physicists see the current situation as problematic.

Has not thinking about AdS/CFT also deteriorated, or at least diminished in applicability? It wasn’t that long ago there seemed to be loads of ambitious talk about AdS/QCD and stringy condensed matter physics. Now I assume it’s about relating the duality perhaps to real-world applications in quantum computing. A solution ever in search of a physically realized problem, seemingly.

To me it reads like someone past his prime, blaming the lack of success in the field on anyone he can, instead of reevaluating whether the ideas themselves lead anywhere fruitful.

Gavin,

I agree that much of Lubos’s problem is that the dream of his youth (a unique M-theory, based on a matrix model or whatnot) has died and can’t be revived. People don’t work on what he wants them to work on because the ideas have failed. There is something to his description though of the current sorry state of the field, and what’s really interesting here is his plausible claim that Arkani-Hamed and others share it (although, unlike him, are not willing to say so in public…).

@Peter Yes, I agree. The problem is that no one in the field has any idea how to make progress with the truly interesting questions. One approach is to work on topics that produce papers, even if they aren’t truly interesting (this isn’t really a criticism; there are always papers written making incremental progress even in healthy fields). Another approach, apparently, is to blame the lack of progress on the younger generation. I wish there was more discussion about re-evaluating the direction of the field to establish where progress realistically can be made, rather than pointing fingers or holding onto old dreams that are unattainable in the foreseeable future.

In condensed matter, the SYK model has generated a lot of excitement as a toy solvable model of strongly correlated matter. This is of perennial interest, because we still want to understand things like high-Tc superconductivity in the cuprates, but lack good analytical tools for tackling even the simplest models for these materials.

SYK rather remarkably links “toy” Liouville gravity (Liouville quantum mechanics) and conformal invariance at short times, and random matrix theory at large time scales. It has a holographic description in terms of JT gravity, although I’m far from an expert on that part.

2D Liouville field theory (as in the original Polyakov path integral, important for non-critical strings) can be formally used to describe properties of certain critical 2D wave functions in topological quantum materials. (The connection is via so-called log-correlated random energy models in statistical physics). These are related to real experiments, notably the quantum Hall plateau transition. The underlying structures seem to be logarithmic conformal field theories akin to percolation, but still very poorly understood. These are now resurfacing in yet another context: measurement-induced transitions in random quantum circuits. These toy models are not dissimilar to actual quantum hardware at Google.

So there are indeed rich and fruitful developments that may yet profit from holography, Liouville field theory, and other stringy byproducts. I can’t speak for what use any of this is to elementary or “real” quantum gravity physics, but some of us on the condensed matter side are indeed excited.

Matthew Foster,

There’s lot of interesting work of the sort you mention, but it’s about quantum mechanical models or maybe 2d QFTs, which have little connection to string theory (other than that people who worked on string theory turned to these topics as “toy models” to play with when their work on string theory failed). There’s zero connection of any of this to the claimed motivation for string theory research as giving “our best hope” for a unified theory.

Peter, I think the part of Lubos’ comments that do resonate a bit with me, although probably not in the way he intended, is that the string community seems to have moved far from its origins.

As a condensed matter person, SUSY is the first part I’m happy to ignore (although there is a simpler version of SUSY that actually plays a useful technical role in many of the applications I described above, for quantum systems with randomness).

But one can ask the very basic question of what goes wrong with the bosonic string?

We know that in principle it can be defined in 4 dimensions (coupled to Liouville CFT), and it seems like a perfectly sensible starting point for an _effective_ theory of a gluon stretched between two quarks. I know real QCD is much more complicated than that, but nevertheless it is a well-defined problem.

I find it very surprising that none of the “modern” string textbooks mention the work of discrete matrix methods that demonstrated the instability of summing over membranes to branching polymers. It’s a quite basic and old result now, but it points out how hard it is to generalize path integrals to higher-dimensional membranes, owing simply to the entropic favorability of contracting the (closed, say) string to arbitrarily small radius and branching infinitely. Maybe this picture applies only in imaginary time, but it seems like the kind of result that should be more widely known and appreciated.

On that note, if one forgets about SUSY, what about a purely fermionic string, i.e. with fermion bilinears playing the role of spacetime coordinates, instead of the free boson? It’s likely strongly coupled, but has anyone tried this? One might hope that Pauli exclusion would help prevent the branching polymer collapse of the bosonic analog.

Matthew Foster,

Understanding string theory in four dimensions is a very old problem that has received (and continues to receive) a lot of attention from theorists. The main hope has always been that this will give you a calculational method for strongly coupled gauge theories like QCD (as opposed to a unified theory with gravity). The problem is simply that no one has managed to have more than very partial success with this.

This kind of work goes back more than forty years, to around 1980, when it was being pursued intensively by Polyakov, Migdal and others. During my graduate student years (1979-1984), trying to solve QCD was the center of attention, and I spent a lot of time trying to understand these string theory based ideas. Post-1984, what got most attention was the 10d critical superstring and unification, but for many people working on string theory, the 4d string was always a focus of their work.

So, sure, this remains an idea worth pursuing, but one should be aware that it’s an idea the best in the field have spent over forty years banging their heads against unsuccessfully, not something that hasn’t gotten attention for no good reason.

Hi Peter,

“So, sure, this remains an idea worth pursuing, but one should be aware that it’s an idea the best in the field have spent over forty years banging their heads against unsuccessfully, not something that hasn’t gotten attention for no good reason.”

Of course. I just meant that it is surprising to me that the intrinsic difficulty of solving the 4D bosonic string is glossed over. Ivan Kostov told me that in his opinion, the bosonic string _is solved_: it has the branching polymer instability seen in the discrete approach to 2D quantum gravity (Cates, Froehlich, Ambjoern, and others), and that’s it. Seems like a major omission that this wasn’t included early in, e.g., Polchinski’s volume 1. (And that’s not a dig at Polchinksi, since I did my Ph.D. at UCSB. He was the friendliest, most approachable string practitioner there, at least as far as the students were concerned).

Given this, without SUSY the case for strings seems doomed indeed. Another comment is that strongly fluctuating, say multifractal behavior, might make perfect sense in the context of quantum gravity. This is something we do know how to handle in certain contexts on the condensed matter side. The branching polymer picture, and what emerges generally from studies of 2D quantum gravity (dominated by extremal statistics of rare curvature fluctuations) is very different from the cartoon of smoothly joining strings in textbooks.

Would still like to know about the pure fermionic version, though…too busy with course preparation now to dig.