20 Years Later

Almost exactly twenty years ago I started writing a short article about the problems with string theory. I had been thinking about doing this for quite a while, and the timing of entering the twenty-first century seemed appropriate for evaluating something that had long been advertised as “a piece of 21st-century physics that had fallen by accident into the 20th”. The piece was done in a week or two, after which I sent it around to a group of physicists to ask for comments. The reaction was mostly positive, although at least one well-known theorist told me that publicly challenging string theorists in this way would be counter-productive.

One person who wrote back was Phil Anderson, I’ve quoted some of what he wrote to me in this posting. He suggested I send it to Gloria Lubkin at Physics Today, and evidently talked to her about it. I did do this, and after not hearing anything back for a week or two, decided to go ahead and post the article to the arXiv, where it appeared as String Theory: An Evaluation.

Rereading that article today, there’s little I would change. Its argument is even more valid now than then. The problems of the theory and how it was pursued evolved over the next twenty years in ways far worse than what I could have imagined back then. In particular, the “multiverse” argument explaining away why string theory predicts nothing is something I could not have conceived of in 2001. The tribalistic sociology that has led to a large group of people calling themselves “string theorists” when what they do has nothing to do with string theory is also something I would have thought impossible.

In many ways, twenty years of further failure have had less than no effect. Lubos Motl is still arguing that string theory is the language in which God wrote the universe, and Michio Kaku has a new book about to appear, in which it looks like string field theory is described by the God Equation. Ignoring these extreme examples, string theory remains remarkably well-entrenched in mainstream physics: for example, my university regularly offers a course training undergraduates in string theory, and prestigious \$3 million prizes are routinely given for work on the subject. The usual mechanisms according to which a failed scientific idea is supposed to fall by the wayside for some reason have not had an effect.

While string theory’s failures have gotten a lot of popular press, the situation is rather different within the physics community. One reason I was interested in publishing the article in Physics Today was that discussion of this issue belongs there, in a place it could get serious attention from within the field. To this day, that has not happened. The story of my article was that I finally did hear back from Lubkin on 2/21/2001. She told me that she would talk to the Physics Today editor Stephen Benka about it. I heard from Benka on 5/6/2001, who told me they wouldn’t publish an article like that, but that I should rework it for publication as a shorter letter to the editor. I did this and sent a short letter version back to them, never heard anything back (a few months later I wrote to ask what had happened to my letter, was told they had decided not to publish it, but didn’t bother to let me know). In 2002 an editor from American Scientist contacted me about the article, and it ended up getting published there.

Looking back at how Physics Today has covered string theory and related speculation over the past 25 years, I did a search and here’s what I found:

The only thing I could find anywhere during those 25 years indicating to Physics Today readers that none of this speculation had worked out was a short opinion column by Burt Richter

It seems to me that those now in charge of Physics Today should be thinking about this history, their role in it, and what they might be able to do to make up for this heavily one-sided coverage of a controversial issue.

This entry was posted in Uncategorized. Bookmark the permalink.

60 Responses to 20 Years Later

  1. Alessandro Strumia says:

    In 2020 we have webinars. It would be interesting to have a debate between you and Lubos.

  2. André says:


    About your last sentence: some people get elected because they lie more and bark louder than others, and some opinions get coverage in the press because the people behind them lie more and bark louder than others. The sad consequence is that in both politics and theoretical particle physics, the US hast lost its claim for leadership. Hopefully, the change on politics will be followed by a similar change in the physics press.

  3. Peter Woit says:

    While there are some analogies between the way tribalism and reality-denying fake news sources have wrecked US democracy and the problems of theoretical physics, I don’t think debating that is useful (and I’ll suppress further comments trying to do so…).

    Physics Today plays a fairly unique role as the “physics press” in the US. There’s not much else in the way of media here aimed at covering physics news for professional physicists. The role it has played in the string theory controversy is rather peculiar. Some of the articles linked to above explicitly say their goal is to answer criticisms of string theory, but they don’t explain these criticisms or provide a link to them, and such criticisms have not been allowed to appear in the pages of Physics Today. There appears to have been an editorial decision to only host one side of the argument.

  4. Udi says:

    You complain that people are calling themselves “string theorists”, when in practice their work has nothing to do with string theory. But isn’t this a good thing? Doesn’t it demonstrate that “string theory” is actually not a mono-culture?

  5. Shantanu says:

    Peter : Did you try contacting Physics World (Brit equivalent of Physics today)?
    what kind of articles have appeared there?

  6. Peter Woit says:

    I don’t think training students in string theory and refusing to admit that it’s a failed research program, while at the same time moving on to something completely different is a healthy situation.

    The Physics World coverage of string theory has always been more even-handed, and in particular they’ve covered my criticisms of the theory, see for instance

  7. Physics Today last year (2019) published a piece by Gordon Kane in which he argues for a larger particle collider by once again shifting the predicted masses of supersymmetric particles up. Frankly I think it’s embarrassing that a scientific magazine would run pseudoscientific “predictions” like this on their pages. I wrote about this here


    I also contacted PT about it and said that it would be appropriate if they let someone argue the opposite side. Needless to say, I did not suggest myself, but said if they can’t find anyone else I’d be available. They wrote back to say they might do a collider pro-con, then I didn’t hear again from them. I asked a few months later what happened to this and got no reply.

    The bottom line is that they ran all the false assertions by Kane and never corrected them.

    I have since stopped reading Physics Today and have been happier with Physics World.

  8. lol says:

    I’ve been reading this blog fora while, and never commented. But I guess 20 years is indeed a good time to take stock of what has been accomplished. So lets briefly discuss what has happened in “string theory” (i.e. high energy theory, i.e. whatever theoretical physicists at top places work on) in the last 20 years.

    1. A wide range of conformal field theories have been solved numerically using the conformal bootstrap. This has resolved decades old questions about phase transitions in statistical physics and condensed matter, including many ongoing experiments.

    2. The AdS/CFT correspondence conjecture, which is the only known non-perturbative definition of quantum gravity, has been rigorously proven in several specific examples (i.e. a certain 2d example, as well as for vector model like CFTs). Also tons of extremely nontrivial evidence has been amassed in pretty much every other well defined example, such as precise derivations of black hole entropy.

    3. Dualities between quantum field theories, again many with immediate experimental relevance (such as electrodynamics in 2+1 dimensions and O(N) models) have been proposed and nontrivially tested. These non-supersymmetric, completely physical, dualities were inspired by earlier dualities that involved supersymmetry, and were inspired directly from string theory. This is one particularly beautiful example where idea from string theory and condensed matter have worked together to produce a profound result of great interest to people in both fields.

    4. Numerous extremely general theorems about QFT have been proven. One particularly famous example are C theorems (i.e. quantities that are monotonic under RG flow, which is basically the most fundamental aspect of QFT), originally shown in 2d by Zamolodchikov decades ago, which have now been proven rigorously in 3d and 4d.

    Note that in these examples, I have avoided any speculative topics (i.e. black hole information, the origins of dark matter, string pheno, etc), since those are more open to debate, and less universally accepted.

    I would say by any reasonable standard, this is a very impressive list of accomplishments, certainly more impressive than anything alternatives to string theory have offered (tho in fairness, almost no one works on such alternatives, which might come as a surprise to laymen who only read blogs), and certainly worth the extremely trivial amount of funding that theoretical physics gets (which altogether in the last 20 years probably amounts to less than a standard mid scale experiment).

    In sum, I would encourage readers of this blog to retain an open mind. Sure, the standard model has not been derived from string theory, and no one has observed a string, but string theory has continued to profoundly influence numerous fields ranging from pure math to experimental condensed matter physics. This is why everyone is still working on it… not bc of a sinister cabal of the Elders of Physics. Also, no string theorists read Lubos blog anymore, he went over the deep end years ago (and Gordy Kane is like 80 years old? give the guy a break… hes earned the right to say crazy things in his retirement). Unfortunately there are no blogs written by active competent string theorists these days (perhaps bc they are too busy producing results!).

  9. Peter Woit says:

    I wrote about that spectacularly bad Kane piece here

    It’s hard to argue with Physics Today’s decisions to publish what Witten has to say, but their decision to repeatedly publish articles by Kane containing claims that are outright wrong or that no one except him believes is pretty bizarre.

  10. Alessandro Strumia says:

    Dear LOL, what is very impressive in your list is that you consider it a “very impressive list”. The field needs smart young theorists, not those who can be attracted through hype. Why not telling that all predictions about new physics at LHC were wrong, that best hopes for discoveries are now gone, that theorists moved to reconsider a variety of old QFT problems and made some technical progress while waiting that something really very impressive will appear on the horizon?

  11. Peter Woit says:

    You demonstrate very well exactly the problem: instead of admitting that string theory has failed, you’ve redefined QFT, or more generally “high energy theory, i.e. whatever theoretical physicists at top places work on” as “string theory”. Your justification for this is a list of developments in QFT that have either zero to do with string theory or only a tenuous “inspired by” connection.

    Nati Seiberg predicted this back in 2005


    “”Most string theorists are very arrogant,” says Seiberg with a smile. “If there is something [beyond string theory], we will call it string theory.””

    He didn’t predict though that what string theorists would discover “beyond string theory” was just QFT.

    One problem with this redefinition of string theory as QFT is that it’s not done consistently. For instance, the Columbia undergrad course on string theory I believe follows Zwiebach and aims at the light-cone quantization of the string. If “string theory” now means “QFT”, why isn’t this course instead explaining how to quantize a field (for instance the free EM field, so students could understand what photons really are)? Not everyone has gotten the memo that “string theory” no longer means string theory, and that it now means QFT (or maybe something else, whatever is most convenient for those who don’t want to admit that what they used to call “string theory” has been a failure).

  12. lol,

    There are so many things wrong with your comment, I don’t even know to begin. Let me just highlight two of your most glaring mistakes. First, you seem to believe that AdS/CFT is the only “non-perturbative definition of quantum gravity”. Gross. You should stop commenting on things you clearly know nothing about. Also, have you heard that we don’t live in Anti de-Sitter space? Second, black hole entropy. Do you realize that no one actually knows what the entropy of a black hole is? So what the heck do you think this calculation shows? As Doyne Farmer put it so nicely, it’s “math piled on top of math”. Physics is not math. Physics is about describing nature.

    But, yeah, thanks for being here to document the problem everybody else is talking about. You have evidently never thought about what you believe. Time to start using your brain.

  13. Apostolos Syropoulos says:

    Someone who works at Columbia University posted on Twitter someting about an interview with Susskind the inventor of the super duper theory of string. I dared to reply that string theory is a failure and he blocked me and send me a direct message asking me if this was a joke or if I am as stupid as I look. I can only imagine what people have told you all these years…

    If you think this comment is interesting but do not wan to mention the university just say an American university. If you like I can send you a screen shot of his post

  14. AcademicLurker says:

    My impression, from this blog and elsewhere, is that the formalism of string theory requires a considerable investment of time and effort to understand. Do string theory-ish techniques have applications outside of string theory itself? If not, it seems like having students spend significant time learning it is like having them spend time learning to write calligraphy in elvish. Perhaps the time could be more profitably spent on other subjects.

  15. CFT gal says:

    Dear Sabine,

    Can you explain what you think is wrong with lol’s statement “AdS/CFT is the only “non-perturbative definition of quantum gravity”? To be honest, lol’s comment demonstrates way more expertise in the subject than you do.

  16. Peter Woit says:

    lol/Sabine Hossenfelder/CFT gal,

    I’m going to put a stop to this kind of argument about AdS/CFT, since I’m getting flashbacks to years of the String Wars, where endless arguments of this kind shed more heat than light. I will add my own final point, agreeing with Sabine that we don’t live in 5d AdS space-time: there is no such thing as a well-defined AdS/CFT-based theory that describes quantum gravity in a universe like ours.

    Another aspect of the flashback problem is that I’m reminded of something I really detested about the behavior of string theory partisans during the String Wars: posting anonymously from major physics research institutions, hiding behind anonymity to attack the competence of others and make really dubious arguments. Please don’t misuse anonymity to engage in unprofessional conduct.

  17. Peter Woit says:

    Apostolos Syropoulos,
    I’d suggest engaging in discussions about string theory on Twitter is even more of a waste of time than most discussions on Twitter, which is a pretty extreme situation.

    I noticed that this posting made it to Hacker News, so if you want an unmoderated computer nerd discussion, see

  18. lol says:

    Alessandro Strumia,

    You raise a good point, that indeed theoretical particle physics is struggling due to the lack of new interesting experimental data, the failed predictions, etc. The questions being considered today clearly are not as groundbreaking as the golden age of the standard model. But there are other interesting questions that can be answered, and I hope my list provided some of them. For instance, the numerical solution to the 3d Ising model phase transition from the conformal bootstrap is no trivial thing, some of the giants of physics (e.g. Onsager) most famous work was on this theory, and its taught us countless deep things about nature (to quote ICP: “****ing magnets, how do they work?”). It is in many ways, the simplest and most fundamental QFT, of interest far beyond its numerous realizations in nature. I would say preferring 4d SU(3)xSU(2)xU(1) gauge theories is a matter of taste, similar to the classic debate between micro and macro (Phil Anderson has a fun essay where he prefers the meso).

    Also, considering how incredibly difficult it is to write a consistent theory of quantum gravity, I’m happy with any version at this point (tho of course like everybody it would be great to find a version in 4d flat space without supersymmetry!). There has been lots of interesting work in the last few years about trying to get flat space holography from AdS… but the jury is still out there (its a hard problem!)


    You also raise a good point, that indeed many of the things called “string theory” today have nothing to do with string theory, and its more sociological that its called that (which is less arrogance of string theory than the fact that people are lazy with labels). But it’s worth noting that many of the big accomplishments I mentioned directly come from either the methods developed to understand string theory, or the theory itself. E.g. most intro string theory classes like the you mention spend half the semester teaching just 2d CFT methods. These are the exact same methods that are used to study some of the most cutting edge theories in condensed matter (like the work of Nicholas Read, who ironically works at a place that is no fan of string theory in general), but most of them were developed by people mostly motivated by string theory (e.g. sasha polyakov, one of the most underappreciated [by the public] and profound physicists of the latter 20th century). Also, the stuff about the 3d dualities, thats very trendy in condensed matter these days, was directly inspired by analogous supersymmteric 3d dualities derived from string theory (and found by a former string theorist now full-time condensed matter theorist Dam Son). Some people even claim you can derive the new condensed matter ones from the supersymmetric one.

    [Juvenile 3 letter response deleted]

  19. Peter Woit says:

    Academic Lurker,
    Yes, string theory is a highly technical subject for students to learn, much of this technology not very generally useful. I’d argue the most useful is the 2d conformal field theory part of a course on string theory, but this is quite a demanding subject, best learned on its own, without the extra baggage it comes with when used in string theory.

    The undergrad-level courses I’m aware of use Zwiebach and I believe don’t discuss conformal field theory, just try and quantize the string in light-cone gauge. This technology is pretty much useless for anything else.

  20. Peter Woit says:

    About flat space AdS/CFT: very smart people have been trying to do this not just in “the past few years”, but from 1997 on. After 23 years, I think the jury is in on that one.
    The 3d Ising problem is a great problem in mathematical physics, with theorists trying to solve it using strings back before my student’s parents were born. All in all, if high energy theorists want to give up on high energy theory and instead go work in condensed matter physics, good for them. Will be interesting to see how they do competing for grants, jobs, etc. in that field. And what the tolerance level among condensed matter experts for hype is…

  21. Naive Postdoc says:

    Dear Peter,

    I unfortunately think that the hype is already present (though to a lower degree) in theoretical condensed matter physics, particularly in the field of symmetry protected topological phases of matter (SPT) and I don’t think they have a problem with grant money.

    There’s a reason why so many string theorists and string mathematicians such as Witten, Kapustin, Gaiotto, Freed, Hopkins, Johnson-Freyd among many others are working on this topic and have, in my naive opinion, brought a similar attitude to the one they have in HEP as a community. Let me not expound my accusations too much and only say that they use emergence as a wild card and the worrisome thing is that there are almost no SPT experiments, though much more likely to be in the near future than in HEP, but in the meantime speculation reigns.

  22. Alessandro Strumia says:

    Dear LOL, I agree that this work might indirectly lead to useful tools to understand physics of fundamental relevance. Still, computing the brachistochrone is just the optimal way of spending theorists time, while the main line of development stagnates. The difference between 4d SU(3) x SU(2) x U(1) and QFTs for magnets* is discussed in a footnote of Weinberg QFT book:

    «* This section lies somewhat out of the main line of development and may be omitted in a first reading».

  23. BootsTrap says:

    Lol, it is strange to see numerical bootstrap on the list and even at the top. These CFT’s were understood long ago and various perturbative schemes to compute are available, e.g. the famous 4-epsilon expansion. The recent attempts were about putting some old equations on a computer and present nothing new, just another numerical gadget to add few digits to the already known exponents. That it became somewhat popular is due to a low entrance fee – even a student could put these things on a computer and get some plots for a paper. The 3d non-SUSY dualities are exciting, as well as the a-theorem, and sphere free energy but have nothing to do with strings and are not even related to them.

  24. tomate says:

    I was then a bachelor student, but already interested in getting a picture of what contemporary physics was all about. I remember finding this paper in the stack of unclaimed papers on the side of the printer in the computer room. I read it with curiosity, happy that I would have some arguments I could hardly understand to give myself a posture in my chats with my senior pals. It contents stuck with me for quite some time, until I had to take decisions on my future, and this paper was a big part of my choices, in a physics department were if you were a HEP theorist there was no choice apart from strings and branes.

  25. Peter Shor says:

    @Naive Postdoc: Symmetry protected topological phases of matter is not an area which physicists have worked on for 40 years with very little to show for it.

    Theoretical physics tends to have fads, with lots of people working in the popular areas, and new areas tend to get overhyped. The worrying thing is that string theory is still overhyped after people have spent 40 years on it with very little return, and many of the researchers have given up on doing active research in it and are working in other fields.

  26. Naive Postdoc says:

    Dear Peter (Shor),

    I totally agree with you that the case in string theory is much worse. I was addressing a particular issue raised by Peter (Woit) that we would see how much of their hype was tolerated by condensed matter physicists and how much money they would get … My point was that perhaps condensed matter physicists are more tolerant than he thinks, SPT and the string theorist’s move towards it being the case and point. On the whole, the influence of string theory may have permeated other branches and their subcultures deeper than we’re aware.

  27. lol says:


    I’m not sure what you mean by “These CFT’s were understood long ago”. Most interesting CFTs are strongly coupled, which means that just writing down some Lagrangian isn’t very helpful (and of course many interesting CFTs don’t even have Lagrangians, especially in higher dimensions, such as the 6d (2,0) theory), bc Feynman diagrams teach you nothing about them. Things like epsilon expansions give some intuition, but they are totally nonrigorous and don’t always work (for the obvious reason that epsilon=1 is not a good expansion parameter). The big question in formal QFT for the last 30 years or so is how to find non-perturbative ways of understanding QFTs (eg CFTs), and most experts in both condensed matter and particle physics acknowledge the numerical bootstrap to be one of the biggest breakthroughs in that regard. The point is that the progress is not just computing a few more digits of accuracy, its conceptually having a way to define and compute things in theories for which no non-perturbative method was known before (except Monte Carlo methods in some of the simplest cases, but those don’t give very much physical intuition, and are limited to the most simple quantities in the theory). For instance, you can read this writeup in a condensed matter journal here (https://www.condmatjclub.org/uploads/2020/01/JCCM_January_2020_02.pdf), which discusses recent experiments (involving a spaceship!) that bootstrap results have influenced.

    Also, the reason i mentioned 3d non-SUSY dualities and the a-theorem, is precisely bc they were derived using cutting edge CFT and QFT techniques that were developed by string theorists. In particular, the dilaton effective action (which is how the a-theorem was proved) and mirror symmetry (which is how the supersymmetric version of the 3d dualities were proven, which strongly influenced the non-susy dualities. mirror symmetry of course has had a profound influence on pure mathematics as well).

    Finally, on a sociological note, it’s always hard to judge an extremely technical fields as an outsider (even a very smart one, say a prominent scientist in an adjacent field). I think a good rule of thumb is to see if the people working in that field have produced results that have influenced many other fields, in addition to things just of interest to their field. For instance, the giants of string theory (like Witten, Moore, Polyakov, Maldacena, Gross etc), even if they had done absolutely nothing for string theory, would still be famous due to their numerous profound contributions to cosmology, condensed matter, pure math, etc, many of which have had direct experimental repercussions (I can give you a long list for each of these people if you want, but you could prob just go on Inspire and see for yourself). This is why people take them seriously, not just bc quantum gravity is trendy, or silly hype. Conversely, if you come up with your own theory of quantum gravity, but haven’t already done anything profound in other less speculative fields, and if you’re new methods aren’t useful for anything outside what you’re doing, then you shouldn’t be insulted if people don’t take you that seriously (even if you write lots of popular books to convince laymen that theres a conspiracy against you). A good analogy might be CERN: you might know nothing about what particles are, but you’ll certainly appreciate the world wide web that was invented as a byproduct. The point is that good lines of research inevitably produce useful byproducts (since all correct things are eventually connected in theoretical physics), whereas bad lines of research don’t. After all, they are Not Even Wrong 😉

  28. yaron says:

    @Peter Shor: Indeed, physicists have not worked on topological phases for 40 years with little return, *yet*. But how long should one wait?

    From yours and Native postdoc’s comments it seems we share some discomfort from the situation in string theory, but the question is what lesson do we apply to other research programs. Since people made nice careers in strings (40 years is probably around average from graduate school to pension) it seems there is less risk in following the hype than trying something new, even after one sees the hype is false.

  29. Avals Tsetrot says:

    I can offer a reason why most high energy theorists comment on this blog anonymously. It’s because not many people are willing to admit that they, looking for a distraction and in a moment of weakness , come to this blog to waste some time reading this incompetent dirt. Here we go, let’s see if the blog’s owner will have the balls to let this comment through, I bet he won’t.

    [Attacks on others deleted]

  30. Avals Tsetrot says:

    Since the owner deleted a part of my comment which he qualified as an attack on the others, let me try to waste a bit more of my time and rewrite it in a less polemic form.

    First of all I do not believe it’s a job of string theorists to predict new physics at the LHC, rather it is the job of hep-ph community, some of whose most illustrious representatives have deigned to comment in the above thread. Hep-ph people, not string theorists, apply for and win Advanced ERC grants to predict signals of new physics at the LHC.

    Second, there is no shortage of funding for alternative theories of quantum gravity, as long as they have a sliver of hope to be successful. One example is the Horava theory of Lorentz violating gravity http://arxiv.org/abs/0901.3775 (BTW Horava is a string theorist), whose ramifications have been investigated vigorously. This activity was supported e.g. by the CERN theory department (which at some point hired two junior staff members working on it http://arxiv.org/abs/1410.2408). Higher-derivative gravity is also under active investigation in spite that it is not perturbatively unitary, and one has to jump through all sorts of hoops to pretend that perhaps in some pickwickian sense it is. This activity has been supported by at least one Advanced ERC grant http://arxiv.org/abs/1705.03896.

  31. Peter Woit says:

    I’m actually very sympathetic to the program of exploiting conformal symmetry in 3d QFT, but have a different perspective on it, and see motivating it by string theory as seriously misguided.

    You write:
    “The big question in formal QFT for the last 30 years or so is how to find non-perturbative ways of understanding QFTs (eg CFTs)”
    What you’re missing is that this began not 30 years ago, but 47 years ago, with the 1973 discovery of asymptotic freedom and realization that the correct theory of the strong interactions was QCD, strongly coupled in the infrared. By the time I got to grad school in 1979, for quite a few years everyone at Princeton had been focused on non-perturbative QFT, aiming at understanding QCD. A lot of effort was going into studying other QFTs as toy models and everyone was spending their time studying what the condensed matter people knew about such QFTs. My thesis work was Monte-Carlo computer simulations aimed at QCD, but I began with testing out methods on some stat-mech models such as the 2d XY model (the simplest model with interesting topological features).

    The current interest in strongly-coupled stat-mech models among ex-string theorists looks to me like not a development inspired by string theory, but a return to what everyone was working on in the early 1980s, before they abandoned it in 1984 for 10d superstrings and complex 3d algebraic geometry (because Witten and others convinced them that all one needed to unify physics was to find the right Calabi-Yau).

    Of course there has been a lot learned since 1984 about non-perturbative QFT. I share your fascination with the question of how to exploit conformal symmetry in QFTs above dimension 2. In particular, I’ve been spending a lot of time looking at what Rychkov and others have done by studying the very non-trivial relations between the Euclidean and Lorentzian signature situations in 3d. I differ though I suspect in believing that string theory is worse than useless here. While some theorists may have gotten to this problem by starting at string theory and AdS/CFT, trying to understand 3d QFT by compactifying 10d superstrings is likely to at best tell you something highly obscure about uninteresting models.

    Where my point of view is completely different is that I’m not ultimately interested in condensed matter, but in fundamental physical laws, so care about 4d, not 3d. One of the main motivations of the twistor ideas I’m excited about is that twistors provide a compelling way to get 4d conformal symmetry. What I see as new has to do with the relations between the Euclidean and Minkowski conformally invariant theories, but in 4d, not 3d.

    So, I’d be happy to see a campaign to promote the study of conformal symmetry in QFT as the guiding principle (or even fad) in the subject. But to get students and others interested in this, starting by teaching them about 10d superstrings and AdS/CFT is a mistake, even if this is convenient for those who don’t want to admit failure of their research programs.

  32. Peter Woit says:

    Avals Tsetrot,

    “I do not believe it’s a job of string theorists to predict new physics at the LHC”
    The problem is that string theorists have given up not only on predicting new physics at the LHC, but on predicting new physics anywhere at any energy, as well as giving up at explaining anything about fundamental physics.

    The current problem is not that research on string theory is still dominant after 40 years, it’s that leading string theorists have given up on fundamental physics, trashing the subject on the way out with crap like the landscape, and continuing to try and convince people that their failure was not actually a failure.

  33. BootsTrap says:

    Lol, imagine that on a cold morning in 1984 Belavin-Polyakov-Zamolodchkov publish a paper where they put the 2d bootstrap equations on an average computer available back then and observe a kink close to the expected position of the 2d Ising model. This does not sound very exciting as compared to what they actually wrote. This could have generated couple of hundreds papers more… One thing it would not had led to it is understanding of the beauty behind 2d CFT’s. Well, the same is with the 3d numerical bootstrap. It cannot even prove that these CFT’s exist as the technique gives you ‘exclusion plots’ and adding one more operator can kill the solution. It does not lead to any understanding that would be comparable to BPZ. Just another technique to get more digits. Yes, it was popular some time ago, just because it was very easy to write a paper and we all need papers. The SUSY dualities have not helped to prove the non-SUSY 3d dualities. 3d and 4d generalizations of the c-theorem do not require any knowledge of string theory, as is clear from the proofs.

  34. Slava Rychkov says:

    Hi BootsTrap,

    let me tell you a bit about the history of conformal bootstrap. When I gave the first talk at the IAS about the conformal bootstrap revival back in 2008, I was challenged by noone else but my PhD advisor Sasha Polyakov who did not believe at the time the equations I was using were applicable to higher dimensional CFTs. So much for the fact that conformal bootstrappers are using well-known equations, if one of the CFT classics did not outright agree with those equations 🙂

    Also I challenge you to write a few more papers like e.g. the recent bootstrap work http://arxiv.org/abs/2011.14647 if you believe that any grad student can nowadays write a valuable bootstrap paper. You have to be a pretty damn smart grad student to master all the theory which goes into the conformal bootstrap AND develop an efficient numerical algorithm to put all that theory to good use. Good luck.

    Finally, if you don’t like numerical work, I do encourage you to look for analytical understanding of the numerical bootstrap bounds (which are rigorous bounds, mind you, even if computed numerically). Then you will become an equal of BPZ. It should be easier to understand analytically something which you know exists, shouldn’t it? So here is a nice challenge for you and everyone else, a challenge which did not exist before the numerical bootstrap work. You may wish to familiarize yourself with the work of Dalimil Mazac who did find analytical proofs of some numerical bootstrap bounds.

    The point being, before the numerical bootstrap, most people were skeptical that 3d critical phenomena would one day be accessible to the CFT techniques, and now nobody can doubt that.

  35. Dalimil says:

    what BPZ understood in 1984 are extremely special and simple examples of CFTs, namely the unitary ones in d=2 with central charge between 0 and 1. This leaves out a vast landscape of CFTs, namely irrational 2d CFTs (with central charge above 1), and all CFTs in d>2. There is currently no explicit solution of any interacting example of either kind but it is clear that if there is, it would be orders of magnitude more complicated and interesting than what BPZ accomplished. What the modern bootstrap has shown is that the CFT axioms are powerful enough to plausibly single out such theories, giving us some hope that one day somebody smart will find a solution of either an irrational 2d CFT, or an interacting CFT in d>2. Furthermore, from this point of view, the current understanding of 2d CFTs is almost as rudimentary as that of d>2 CFTs.

  36. TS says:

    It is not a new thing that “phenomenology” is thought of as being outside of “theory” but to see this as blatantly as in lol’s comment and the following discussion is rare. All the amazing progress in precision calculations that accompanied the LHC era? Didn’t happen, or isn’t “theoretical physics”!

    Maybe people should consider whether they actually believe that physics is about the real world when they can’t even be bothered to acknowledge the progress that is being made in predicting experimental results in the world’s largest lab.

  37. Alessandro Strumia says:

    Dear “Avals”,

    you don’t believe that “it’s a job of string theorists to predict new physics at the LHC”, but in past decades lot of activity in string theory had been done assuming that there is one vacuum and two scales: the Planck/string scale, and the SUSY/weak -breaking scale (while the vacuum energy vanishes for reasons to be clarified in the future). Then one constructs string models with N=1 SUSY to be discovered at LHC, focusing on Calabi-Yau compactifications, etc, etc.

    Unfortunately, whoever wrote laws of nature had not good taste or was not as smart as string theorists. The vacuum energy was found to be positive, and no SUSY was found at LHC. In the string context, this might mean that there is a huge number of vacua with SUSY broken at the string scale.

    If now you Aesopically like to think that you never cared about LHC, you miss what made physics more interesting than abstract theory: the pleasure of being slapped by experimentalists, hoping that a kick in the back will point to a better theoretical direction than keeping dragging the aether or retracting to pure math.

    PS: Peter, don’t worry about attacks. To keep physics serious we need people like the P-dual of “avals”, who bluntly say what they think, rather than what is polite and convenient.

  38. André says:

    Peter, Avals,

    it might seem that the job on any theorist working on high energy physics is to explain why there are three particle generations, where the gauge groups come from, where the particle mass hierarchy comes from, and where the coupling constants come from.

    It might also seem that most proposed theories, including string theory, where not really that successful in this endeavor, despite the help of a lot of brainpower.

    Avals, are you working on these questions? If not: Are you honestly and really saying that string theory will solve these questions?

  39. BootsTrap says:

    Dear Lol, Slava and Dalimil, I hope that most people agree that an analytical understanding is much better than a (yet another) numerical approach (even if it is the most computationally efficient one at present). There can be two versions of BPZ: economy class: with the help of a computer BPZ draw a plot with a kink in the neighborhood of 2d Ising model. First class: what they actually done. It does not solve all 2d theories, but solves the important ones, and it advances greatly our understanding. I guess we know what is the 3d economy class version of that. Instead of yet another paper that improves the numerics I would recommend to read some of the recent analytic bootstrap papers (for example, Alday and pals who might give us the first class version of the 2d story in the future). Perhaps, one should create hep-th-numerics for people who gave up. No offense, the better numerics for 3d Ising are useful, but I would not put it on the list at all, which is where this debate has started. Dalimil, I disagree that the numerical bootstrap has shown anything apart from adding more digits and it is nice that you moved to analytical methods: the CFT existed before, the kinks that finally turned into the islands are due to 3d Ising-type theories having actions, equations of motions etc, which singles them out of the continuum of ‘theories’ – they are disconnected from the continuum and we knew that.

  40. Slava Rychkov says:

    Hi BootsTrap,

    what a biased perspective on the conformal bootstrap past and present history! So many fertile exchanges exist in the conformal bootstrap community between the numerical and analytical methods! The numerical bootstrap people did not give up, on the contrary, by 2008, it’s the analytical people who long gave up the hope to access strongly coupled non-supersymmetric CFTs via the conformal bootstrap equations. Paraphrasing you, they could not afford a first class ticket and so they refused to fly instead of trying out the economy. The economy salon was pretty cushy, and totally empty, when we tried it. 🙂

    Thanks to the numerical bootstrap, this hope is now back and, to my joy, it stimulates a huge amount of analytical thinking which would not otherwise exist. The numerical work showed a window of opportunity, and afterwards analytically-minded, intellectually curious, and very smart, people decided to take a look at what is going on, not the other way around.

    As to the fact that numerical bootstrap “just adds more digits”, I do like to stress that it does so with rigorous error bars, unlike resummed perturbation theory and Monte Carlo simulations, where error estimation is often an art rather than science. This is pretty unique: in fact there is hardly any other method in quantum field theory which has this property. It gives bootstrap results the status of a final verdict in cases when one has to decide the fate of borderline cases, and more conventional techniques disagree. By now we have several experimentally relevant applications of this unique role of the bootstrap.

    But OK, it’s not like we need everyone’s approval. In fact your opinion is not uncommon, my ENS colleague Volodya Kazakov likes to bug me with very similar remarks, while I bug him back saying that it’s his job as an expert in integrable models, not mine, to figure out if the 3d Ising CFT is integrable and solve it. 🙂

    There are many real physicists out there struggling to understand concrete condensed matter systems. They know the value of a practical method when they see it, and it’s them who I consider the primary bootstrap audience.

    Good buy!

  41. Dalimil says:


    Your comparison of the modern bootstrap with BPZ is substance-free because solving CFTs in d>2 (or irrational 2d CFTs) is a very different (and apparently much more difficult) mathematical problem than solving rational 2d CFTs. So your dismissal of the recent advances is a bit like if a mathematician were dismissing the importance of partial results on the location of zeros of the Riemann zeta function by arguing that we know the location of roots of quadratic polynomials since many years ago.

    Of course, an exact analytical solution of a d>2 CFT would be optimal, but in its absence, if I were to decide whether it’s better to have an analytic answer to 5 orders in perturbation theory (with unknown errors) or a numerical answer with rigorous error bars of size 10^{-5}, I’d probably choose the latter.

    Perhaps the most important point though is that the modern bootstrap advances have made interacting conformal field theory in general dimension into totally rigorous fields of mathematics. It’s true that axiomatic approaches to QFT exist since the 1950s, but it was understood only from 2008 onwards that the CFT axioms are powerful enough to determine dynamical data. In principle, you can now explain these axioms to a mathematician who can then start proving rigorous theorems about interacting CFTs and thus also nonperturbative quantum gravity in AdS — quite remarkable if you ask me.

  42. Peter Woit says:

    Alessandro Strumia,
    I agree that we need more professional theorists who say what they think, even if it’s unpleasant. What I object to and don’t want to tolerate here is abuse of anonymity.

    Trying to think about what Physics Today should be doing (as opposed to what it has done), what’s needed is not so much articles by people like me or Sabine Hossenfelder answering the kind of articles they have been publishing. Instead, what’s needed is a different kind of article, with an honest discussion of what the problems are and what experts think about them. Instead we get these highly political one-sided arguments (which don’t even acknowledge explicitly the other side). A good start might be for Physics Today to go to the authors of the articles I linked to, along with others who over the years have argued for heavily promoted speculative ideas that have now been disconfirmed, and ask them to honestly address the current situation.

  43. Peter Woit says:

    In general I don’t think that theorists need to be addressing directly experimental results or fundamental open issues in the theory, since most of the time they won’t have anything useful to contribute if they try and do that. Better to work on possibly more obscure and technical things that are not understood, where progress is possible and might someday inspire a new idea that can help with fundamental issues. But fundamental open issues are not going to ever get solved if people misleadingly tell themselves and others that they have alread been solved (or have been shown to be insoluble).

  44. Lee Smolin says:

    Dear LOL,

    I agree that your four points are an impressive list of accomplishments, of whom the many who contributed should be rightly proud. But in the generations who entered physics in the 1940s to 1970s there are a number of individuals who each had a lasting impact on science greater than your whole list, while addressing both theoretical and experimental issues: Polyakov, Wilson, Feynman, Gell-Mann, Weinburg. Anderson, ’t Hooft, Dyson, And then think that they knew Einstein, Bohr, Born, Heisenberg, Schrodinger, Dirac…

    May I gently suggest that those of us who aspire to argue about the future of physics might, after 20 years, include humbleness on our list of New Years resolutions?

    Best wishes for the new year,


  45. Alessandro Strumia says:

    Physics so far kept attracting best people, so it’s possible that LOL & co are as smart as Polyakov & co, and that the real difference is that old generations did the easier important jobs, leaving the harder less important jobs to current generations. This might have now contributed to avoiding open discussions about difficulties and to hyping partial achievements. But the real end of physics will come if it will stop attracting smart “un-humble” persons.

  46. lun says:

    Alessandro, your comment is pretty telling on many levels.
    The “top talent” drift from theoretical physics to various incarnations of data science and computer science has been going on for decades. The fudamental problems theoretical physicsts are struggling with now, far from being “less important”, are are precisely the ones the likes of Einstein, Maxwell and their ilk struggled with: unification of all forces, quantization of gravity, microscopic understanding of entropy and so on. Its just that the current “elite” has been making a lot less progress in their resolution while feeling much more self-important about it.
    Then again, one big problem with “meritocracy” is that self-selected elites who believe they are the best and only people like them are the best is that intellectual incest and group-think, to the point of blatant blindness, follow naturally. We see this all over theoretical physics today, it is part of the problem with it.

  47. lun, I’d like to concur with you using the words of Roger Penrose : I once heard him say that there are two types of mathematicians, smart ones and dumb ones, and that he fell in the second category. I think the problem of today’s theoretical physics is that there are too many smart physicists.

  48. Peter Woit says:

    Some thoughts on comments above:

    I continue to see very smart young people wanting to go into fundamental theoretical physics. The best go to get Ph.Ds at the same sort of elite programs where I was educated. What I don’t see is papers with impressive new ideas from young theorists. This is not their fault: their elders are not coming up with anything either, and the way in the past you often got breakthroughs was young people working not in a vacuum, but set in a promising direction by their advisors or others around them.

    The combination of few new ideas, tribalism and a celebrity culture has led to more examples of huge rewards and attention for not so great (or even wrong) work, with people labeled as “geniuses” and collecting $3 million on very dubious grounds. A screwed-up rewards structure carries serious implications for the ability of a field to make real progress.

    As for “humility”, that’s definitely needed and missing in terms of people not admitting that an idea has failed, or has produced only very modest results. On the other hand, a fair amount of arrogance is needed to believe that one can make a breakthrough and be willing to bet one’s time and energy on it. So, we need more of both humility and arrogance…

  49. lol says:

    Dear Lee, Peter, etc,

    Happy new year!

    I think we actually agree on a lot more than than the stereotypes would assume. None of us think string theory has solved the deepest fundamental issues, none of us like overhyped results (especially when the scientists are to blame, not just the science journalists), we all seem to appreciate some nice results from the last 20 years (even if they aren’t as groundbreaking as those from the golden age of physics), and that progress in technical nonsexy things (like computing Ising model critical exponents to a few more digits) is still a worthwhile and honorable thing for scientists to do, even if it wont explain the hierarchy problem or solve global warming.

    One last thing I would try to emphasize tho, is that while a few famous string theorists at well funded institutions might well give off the impression of great arrogance in their accomplishments, for most of us, we’re working for very low pay, for many many years, and certainly not basking in the glow of media attention. Speaking personally, I have chosen anonymity on this forum not bc I’m secretly Ed Witten in disguise (he seems too busy tweeting about Israel these days lol), but bc i’m just a lowly postdoc at a non-American institution with very poor job prospects, making probably the equivalent of minimum wage in the US, and I don’t want any inadvertently foolish thing that I say online to hurt my already dismal prospects. For people like me (who are the vast majority of string theorists, high energy theorists, formal theorists, whatever), we do what we do out of love for science, fully aware of the huge experimental and theoretical problems with our field, out of a perhaps quixotic dream that perhaps we will stumble upon a solution, or at the very least make substantial, if not groundbreaking, progress on worthy technical and sometimes conceptual issues.

    The main reason I spoke out was simply to convince laymen (or non-string theory scientists) reading this blog that despite what they may have been told, we are making steady progress in some directions (maybe not the sexiest ones), and that we would appreciate to continue receiving our paltry salaries, which seem to decrease by the year. My main concern is that excessive negative press will defund not just those doing extremely speculative work (which I do think also deserves some funding, tho Im not the right one to defend them), but also the 90% of us doing humble meaningful work, and contributing to the results I listed.

    Perhaps going into the new year, in addition to decrying excessive hype, we could vow to also promote humble progress?

  50. Alessandro Strumia says:

    Dear lun,

    bibliometric data show that fundamental theory was concentrated in a few main centers up to ≈30 years ago. Next, internet made research diffused. Access to internet become enough to find articles and books and tools, and follow courses and webinars. A very strong young physicist can now get a safe position somewhere even outside main centers. Actually, IAS seems to me the last concentration of excellence. Recently we also got a diffusion in the topics that are being explored. This should be enough to bypass sociological problems.

    I fear that stagnation on big issues mostly happened because these are difficult problems.

Comments are closed.