The string theory hype machine will never die. This week we have
- Press releases from RPI and Northeastern tell us Scientists Use String Theory to Crack the Code of Natural Networks and How String Theory Helped Solve a Mystery of the Brain’s Architecture. Scientific American asks Does String Theory Explain the Wiring of the Brain?. This is all about this paper (arXiv version). Sabine Hossenfelder likes this.
- Quanta today tells us that String Theory Can Now Describe a Universe That Has Dark Energy. This is about yet another of the sort of “string vacua” that people have been constructing for decades and that look nothing like the real world. This latest one is five-dimensional, but we’re told “the work is expected to launch a new era in matching the mathematical elegance of string theory to the actual world we live in.” The paper in question is here.
I’ve been documenting this sort of ridiculous hype for more than twenty years now. It has done a huge amount of damage to the public understanding of science and to the credibility of scientists. It also hasn’t helped the perception of string theory by other physicists, with string theorists now virtually unemployable unless they can figure out how to rebrand as machine learning experts. String theorist Manki Kim reports here that “string theory is in a very fast contracting phase, maybe I was dumb enough to hold on to a dead horse for too long. Should’ve given up long time ago.”
There’s no point in going into more detail about this kind of hype and its continued existence. The world is passing it by, moving on to fresh, new horrors.
To be completely fair to the current string theory hype apparatus, after 20 years of hype, they can hardly now revert to more honest reporting along the lines of “hey look we made this small incremental progress today and that’s pretty solid.”
I think a constructive conversation would be for the young people of the next generation. If you are an aspiring young physicist seeking to understand the deepest mysteries of the universe, how should you navigate the current climate? How would you build your career in such a way that when one day you have influence over the field, you take it in a positive direction? “Learn from mathematicians” isn’t a good enough/detailed enough answer, our field is also driven by hype and prestige of influential people.
PS: the most outrageous thing I find about the quanta article is that when they cite a 3 author paper that inspired the current study, they only mentioned the name of Eva Silverstein, a famous person, and didn’t even mention the name of her 2 other collaborators. A prime example of rich get richer. I would have expected better from writings that are related to currency and prestige in academia.
This problem isn’t really specific to string theory right? University PR oversells papers is almost a law of nature. This should be addressed generally.
Secondly it would be helpful to have more clearly targeted and titrated criticism, because you’ve mentioned press releases, a sciam article, the paper, and Sabine all without differentiation.
I hope it’s clear enough the paper itself is legit and doesn’t seem to make any inappropriate claims. Beyond that, the PRs seem to be the real offenders here, the sciam article less so (could be argued that’s healthy popsci), and I’m not sure what comment you’re making about Sabine. The title of her video may be click baity but the content itself seems to appropriately demarcate string theory from the paper.
clueless_postdoc,
Unfortunately I don’t have much useful advice for a young person entering the field of fundamental physics. It’s in a dark age and getting darker quickly. Part of what is accelerating this is the refusal to admit there is a problem.
What’s always been striking to me about the hype is how those promoting it appear to simply not care whether what they are saying is true, just whether it sounds plausible and people will believe it (if you want to put a name to this, think Michio Kaku). Unfortunately GenAI has precisely this characteristic: what’s true is irrelevant, the goal of the algorithm is to say something that appears to be a convincing response to the prompt.
In principle math has an advantage, since it has a strong culture of distinguishing between what is true and what isn’t. Maybe some day all math papers will be run through Lean or something that checks for technical correctness. In the meantime, the advent of GenAI written math papers with embedded hallucinations is going to be a big challenge.
Minor fix: In the second point, “The paper in question is here.” links not to the arXiv but to the Quanta article for a second time. I can’t quite tell from the Quanta article which preprint you mean.
Lee Whitney,
This is the 149th post on this topic that I’ve written over the last 22 years. The posts and comments on them that I’ve written probably come out to over 500 pages of text. In much of this I’ve carefully explained exactly what the hype is about, and why it is not true but hype.
Most of that effort was a complete waste of my time. The people putting it out don’t care whether what they say is true or not, so anyone discussing their hype is wasting their breath. So, no, I’m not going to carefully explain what this nonsense is about.
To waste a bit of time and say a little bit more:
1. “String Theory Can Now Describe a Universe That Has Dark Energy” is complete bullshit. The subtitle “In an unprecedented step, researchers crafted a detailed model compatible with the universe’s accelerated expansion.” is also complete bullshit.
2. A sizable number of the 149 postings are about hype of the genre “string theory finally does something useful, explaining X”, where X is the value of pi, heavy-ion collisions, or any number of things. In every case I’ve looked into, there actually wasn’t an explanation of X, or if there was something new about X, it wasn’t “string theory”, but something tangentially related to string theory. In this case I know nothing about X=neuroscience and not going to spend my time learning about it and working out exactly what they hype is here. The pointer to Sabine is to a different point of view than mine. To be honest, I just skimmed her video. Anyone interested can follow the links and spend their time on this. I’d advise doing something more intelligent with your life…
David Roberts,
Thanks! Fixed.
Lee Whitney,
Something else I should say. A large number of the 149 postings about hype involved news stories generated by university press releases about how their scientists have shown that “string theory does X”. I don’t know of any exception to the rule that such press releases are always bullshit. Seeing the hundredth one of these when the first 99 were nonsense does not make me think “Gee, maybe finally they have something”.
I finally got around to reading the SciAm article by Lee Billings. It’s pretty good, explaining reasons to be skeptical about these press releases. It ends with a quote from a string theory who has worked on biophysics “I’m not sure that this study marks a critical breakthrough in our understanding of physical networks, and many experts may find the claimed relationship to string theory unconvincing,”
or, in other words, it’s hype….
@Peter,
Maybe there’s structural reasons why it’s hard for the leaders to admit defeat that are not their own fault, and maybe the world would be better served to create an environment that makes it easy for the leaders (of any field for that matter) to admit defeat. But from what I’ve seen, every thing that matters in academia (in theoretical circles, academic reputation/influence far outweighs any monetary currency), the ability to think about things you like, the influence you have, moreover the young people you train who followed your footsteps and you hope will carry the flag once you’re gone – they all seem to depend on the hype (or “excitement”) you generate around your research program. Maybe if you admit defeat you are not hurting only yourself, but people closest to you and people who believed in you. (and open your clique to attacks by enemy cliques, who are made of far worse people than your own clique, I’m sure).
In math there is truth in principle. But if a branch falls in a forest, does it make a sound? So many papers are produced that each day that nobody can keep up, so to get the fields’ attention (and for young people like me who needs a job) you have to advertise to the right people, otherwise your work simply won’t be read. Period. So generating some hype is somehow a necessary evil of the process. I don’t know a good way out of it.
ct’d in the above.
Academia is in some sense also a political system of sorts. It doesn’t seem like a democracy, but I also don’t know what exactly it is. There does seem to be some kind of difficulty when it comes to transition of power, though…
clueless_postdoc,
“PS: the most outrageous thing I find about the quanta article is that when they cite a 3 author paper that inspired the current study, they only mentioned the name of Eva Silverstein, a famous person, and didn’t even mention the name of her 2 other collaborators. A prime example of rich get richer.”
This is the full author list for that paper. If Quanta had to choose only one name to mention, it would of course be the second author’s.
G. Bruno De Luca, Eva Silverstein, Gonzalo Torroba
From what I can see, the Nature paper has little to do with “standard” string theory and much more to do with membrane dynamics in soft condensed matter. Although the main classical Hamiltonian (call it the action, if you prefer) is proportional to the total area, these surfaces can’t intersect with impunity (for example, there should be self-avoidance, although I’m not sure the authors discuss that much).
Manki clarified that by dead horse he did not mean string theory but his career. He was talking about career opportunities in string theory.
Nirmalya Kajuri,
Yes, his career is a dead horse, but that’s because he tied it to string theory, which in his words “is in a very fast contracting phase” (i.e. dying).
I recommend taking a look at Manki Kim’ s website
https://sites.google.com/view/mankikim/curriculum-vitae
Lots there about his career, including an amusing “CV of Failure”
Taking a look at his research statement and projects he’s interested in, this is plausibly exactly what one should be doing if one wanted to make progress on this problem
https://xkcd.com/171/
But he’s right that, in the marketplace of ideas, this one “is in a very fast contracting phase” and, at least in the US, you’re not going to get a permanent job pursuing it.
Hi,
Thanks for interests in my bluesky posts and my website. Although this much attention is burdensome to me and leaving a comment here is probably going to worsen it, I wanted to leave a comment clarifying what i meant by my bluesky posts, as that is my personal account for which I don’t always write precise statements.
As Nirmalya mentioned, by the post, I meant that the string theory job market is contracting, which is no surprise to anyone who reads this website. I did not and do not mean any other things. I hope you would understand and show some empathy.
Manki Kim,
My best wishes for whatever happens next for you. I do empathize with your situation. Back in 1988 I was in a similar one (postdoc running out, no prospects for a continuing job in the field). Like you, I was pursuing what seemed to me the most interesting ideas for making progress. It was the other end of the string theory boom, so the relation of string theory to job prospects was the opposite of yours.
The job market for permanent jobs in HEP theory has been awful now for about 55 years, something I tried to write about in my book 20 years ago (it’s gotten worse since). This is and has been bad for progress in the field, forcing people to stick to a small number of “hot” topics if they want continued employment.
Being in the situation you’re in sucks, but it’s not unusual. Almost always, after a difficult period, people do manage to find something satisfying to do, generally with conditions a lot better than continuing to work in HEP theory. Good luck!
It is amusing to see branching networks as an advertisement for string theory (or even string field theory!), given the 40-year-old Froehlich liquid membrane story about the branching polymer instability of the bosonic string, related (somehow) to the c_matter = 1 barrier for Liouville quantum gravity.
The latter refers to placing 2D conformal matter on a randomly triangulated/curved background geometry and summing over all such geometries with fixed genus. Geometrically meaningful strings have c_matter = D > 2 components, and thus are “beyond the barrier” where continuum 2D quantum gravity is supposed to make sense.
Of course here they just minimize classical surfaces, which is at least superficially neither string nor string field theory.
Matthew Foster,
As I said above, these surfaces should have some kind of self-avoidance interaction (or somethings similar) which is not local on the world-sheet. Yes, this superficially looks like branched polymers.
Hi Peter, it’s true that generative AI can hallucinate, and may give wrong answers. But you make it sound as it is a wanted feature, while I see it as a intrinsic byproduct of an otherwise enormously powerful and useful tool. I use CGPT5.2 every day for coding work and problem solving, and it cut my development time by a factor of 10.
We do hallucinate too, and this makes LLMs more like us. In fact, I believe our brains work a lot more like LLMs than it is generally accepted. But that’s a long discussion… To be had on a sofa on the next occasion!
Cheers,
T.
Hi Tommaso,
The comment about hallucinations was specifically about the problem it causes for math papers (effortlessly generating complex, apparently convincing but incorrect math proofs). I agree that humans think and behave much more like LLMs than it is comfortable to realize. That you can use them to do what you are doing 10 times faster is great if what you are doing is good scientific work, not so good if what you are doing is producing flawed proofs or constructing bogus “string vacua”.
I’m mostly trying to avoid thinking or commenting on this since it’s changing fast and I’m well aware it’s a topic that I’m pretty ignorant about. We’ll all see soon enough the good and the bad that are coming out of this. These are times I’m finding it hard to be an optimist about humans and their new 10 times more so creations…
It’s been fascinating watching how mathematicians interact with LLMs.
My favorite recent episode: https://www.erdosproblems.com/forum/thread/281
In a nutshell, someone provoked an LLM into coming up with a solution to one of Erdos’ problems (#281). People (including Terry Tao) verified the correctness of the proof. Then they discovered that:
* there was a proof already in the literature
* #281 is a trivial application of a theorem Erdos himself proved
* the LLM proof is different enough to count as modestly interesting.
Tao: “Now I am really puzzled, because Erdos would certainly have known both of these facts in 1980, especially after working for so many decades on covering congruences, and being a co-author of the latter fact. I wonder what happened”
Chan: “It might even be the possibility that someone just told erdos this solution at a cocktail party and nobody continued working on it.”
The impression I get is that people expect LLMs to digest a lot of the low-hanging fruit, while doing relatively little to help with solving the marquee problems, asking new interesting questions, or inventing new branches of math.
On human brains working like “LLM”s- I’m surprised no one here has brought it up yet, maybe it’s more common in math circles. But one of the most common advice we give to beginning grad students is go to conferences, go to talks, even if you don’t understand anything. Go to them enough, one day you will be able to string together half coherent sentences about mathematical objects without understanding what they mean. But this is good, this orients you and eventually get around to knowing what they mean.
it seems LLMs got the “string together coherent sentences” part right, but more work is needed for “understanding what they mean”.
A.J.,
My impression is similar. In the hands of the untrained, this new tool is not going to do much for math, could even do a lot of damage (creating a huge literature of supposed proofs of zero reliability). In the hands of Terry Tao, it may very well help him a lot to exploit his already impressive talents.
All. Please, no more AI comments for now. I’m neither competent nor in the mood to moderate a serious discussion of this. There surely are much better places for it.
Many of us have often cast blame on university PR offices for the hype. I have certainly done so. However, my perspective shifted over time, through experience. I think that the authors of the research bear much of the responsibility for ensuring that their research is not grossly misrepresented in such releases. Press offices lack the expertise to vet many of the claims, let alone appreciate the context, and lack a culture of skepticism. In all the cases that I have been involved in, the contents of the article in question, at minimum, are sent to me for review before publication at some stage. In recent years, I give close attention to the contents, and take pains to ensure that what goes out is justifiable based on the science, extensively editing – or even drafting – the contents. Obvious measures in retrospect, given the visibility that popularizations of science sometimes achieve. Of course, this blog documents many cases in which scientists are complicit in the hype. (I bought into the string theory excitement in the 1990s, but left the field before too long. Not something I dwell on any more, but I’ve appreciated Peter’s blog in the years since.)
You should not blame the university press offices for the propaganda on Quanta. They have really gone downhill – they used to go around the community and talk to various people to see what was going on. However, for the last several years, all they seem to want to do is to obtain the opinion of Princeton, Harvard and Stanford and see what work is viewed positively there. They then go around publicizing that work. In return, Quanta gets nominated by these groups for science communication awards. Given the stature of these institutions, Quanta wins these awards. It is a win-win situation for all the players – the only loser is the science.
The whole biological physics thing is just taking two dimensional Feynman diagrams and the Nambu–Goto action and applying it to networks. Wikipedia’s article on the Nambu-Goto action says that the Nambu-Goto action “is also used in other theories that investigate string-like objects”.
https://en.wikipedia.org/wiki/Nambu%E2%80%93Goto_action
Just because string theory was the first field to use higher dimensional Feynman diagrams and the Nambu–Goto action doesn’t make the application of such things to other fields a success of string theory. It’s like saying that the use of calculus in economics is a success for Newtonian mechanics, because calculus was first developed for usage in Newtonian mechanics and later was used in economics. No, it’s all just the successful application of mathematics to different fields.
Attendee,
The press release and Quanta stories are different (in this case, the first hype example was a press release story, the second a Quanta story).
The press release story is always the same. Some obscure piece of work manages to get published in Nature or PRL, those publications urge authors to issue press releases. The universities do this (with full cooperation of the physicists), turning an unimportant scientific result into a piece of hype with a misleading title. The press releases get picked up by various publications that monitor them through various services, an author there adds some more hype to the press release version and publishes (to be fair, sometimes science writers do much better, in this case the SciAm article does real reporting and show some skepticism).
Quanta doesn’t get their stories this way. You’re right that they largely use their sources at places like Harvard, Princeton, Stanford to decide what to report. The Simons Foundation in general has very much an elite-focused model, for better or worse. In math they do a pretty good job. In theoretical physics the results (like this one) are sometimes awful, corresponding to the fact that these places have very prominent people sometimes pushing awful research. The “quantum computers produce wormholes” fiasco was a high (or low…) point.
clueless_postdoc wrote: “the ability to think about things you like, the influence you have, moreover the young people you train who followed your footsteps and you hope will carry the flag once you’re gone – they all seem to depend on the hype (or “excitement”) you generate around your research program”.
There’s something to what you say here. But:
The ability to think about things you like can actually be damaged by hype, because 1) only certain kinds of research are easy to hype, so when you get into the hype business, you tend to avoid working on certain topics that you’d otherwise enjoy, 2) to build up a powerful load of hype you have to stick to working on the same sort of thing, which limits creativity and 3) it takes time and energy to hype things, which could have been spent doing actual work.
Peter and All:
Natalie Wolchover of Quanta kicks off a new series called Qualia with “Is Particle Physics Dead, Dying, or Just Hard?”:
https://www.quantamagazine.org/is-particle-physics-dead-dying-or-just-hard-20260126/
Peter is quoted.
Quanta Magazine has a new article about whether particle physics is dead or dying:
https://www.quantamagazine.org/is-particle-physics-dead-dying-or-just-hard-20260126/
The Quanta article has a quote from Jared Kaplan that “Brilliant people like Nima Arkani-Hamed or Ed Witten, AI will be generating papers that are as good as their papers pretty autonomously.” That might well be true, but maybe not in the way he intended it…
I’ll try and write something about this very soon in a new blog posting.