Why String Theory is Still Not Even Wrong

John Horgan recently sent me some questions, and has put them and my answers up at his Scientific American site, under the title Why String Theory is Still Not Even Wrong. My thanks to him for the questions and for the opportunity to summarize my take on various issues.

This entry was posted in Fake Physics, Multiverse Mania. Bookmark the permalink.

37 Responses to Why String Theory is Still Not Even Wrong

  1. Well done. Although I can find no mention of it in a recent arxiv article on the CMB cold spot, it appears the multiverse has somehow crept into the journalistic discussion of it … again,

  2. Peter Woit says:

    Thomas Lee Elfritz,
    Hadn’t noticed that, but it’s textbook “Fake Physics”:

    The paper
    would normally get zero public attention, and includes nothing that could seriously be taken as evidence for other universes. The authors put out a press release


    in which they highlight the multiverse angle:
    ‘Perhaps the most exciting of these is that the Cold Spot was caused by a collision between our universe and another bubble universe. If further, more detailed, analysis of CMB data proves this to be the case then the Cold Spot might be taken as the first evidence for the multiverse — and billions of other universes may exist like our own.”

    Press stories then come out like this one


    With a blaring headline:
    “‘Cold Spot’ Anomaly Billions of Light Years Across –“Caused By a Collision With Another Universe” ”

    In deciding where to lay blame for this kind of nonsense and the damage it does to the public understanding of science, I think the bulk of it should go to the scientists involved. Look, if you put out a press release highlighting this, say things like this to reporters, you know very well what the press stories are going to look like.

  3. That sounds about right. I didn’t research it that deeply. I usually just read the paper itself. But what I do is find those papers occasionally through the press, as I don’t have a lot of time to pursue every lead on every search engine using every combination of keywords. It’s nice to see that nowadays they at least provide a link to the papers.

    When I run into a whole slew of journalistic failures like that though, that’s usually a sign that something went awry at a deeper more fundamental level and the press is just parroting that problem. Therein lies the problem, as discussed widely now here and elsewhere. What can be done about that, I don’t know, except to just keep slogging.

  4. G Phillip says:

    Peter, just reading your book. I was waiting to see if all this string “theory” stuff worked out. Since it didn’t, and now with the failure to find SUSY it looks like it never will, I’m digging deep into concepts I have heard of but never really taken the time to study. What a great book! I love all the HEP history. In truth I’ve spent as much or more time studying the concepts you mention like renormalizations and group theory on the side as I have reading, but I’m now 1/3rd through and I can say I’m impressed. My goal is to complete it with a decent understanding before your next book comes out. I found it very honorable of you to acknowledge the contributions of Lubos Motl to the book. In my opinion, everyone in the field should be so professional. I also found it enlightening that many theories along the way that were logical, self-consistent, and even elegant were disproved beyond any doubt by experimental evidence. It’s a lesson we must never forget. Best.

  5. I liked your interview with John Horgan. I would add that there is another critique of Bostrum et al and the Matrix theory, and that is one of infinite regress. If we are in a simulation, then there is no reason to think that the creatures controlling us in our simulation are not themselves simulations, etc. ad infinitum. Obviously, there has to be a stopping point – there has to be someone who is not a simulation running the simulation for the simulation to exist in the first place. However, given the formulation as one of computability in the Matrix theory, where any entity who can simulate can also be simulated, there is no logical reason why it can’t be ad infinitum – it’s turtles all the way up ! Therefore: the Matrix theory is impossible, and thinking about it is a colossal waste of time. Cheers!

  6. Magnema says:

    @Henry Warwick:
    Impossible? I think that’s an implausibly strong claim. The argument does not show that we are in a simulation (in which case, yes, if that were the case with 100% certainty then we would have a problem). Rather, it shows that we should believe we are in a simulation. You could come to the conclusion that any unsimulated world will come to the wrong conclusion in such a case; however, since the premise is that most worlds are probably simulated, that is, in a sense, an “acceptable risk” to those who accept the argument – because, after all, according to their premise, they are probably right, even if uncertain.

    Now, I would agree that “thinking about it is [mostly] a colossal waste of time.” However, I don’t think the philosophical argument could be refuted so easily – and I am, very generally, wary of any claim leading to “impossibility” (or “necessity”) a priori, given how often those claims have been proven wrong. (Implausibility or near-certain truth a priori, including the simulation argument, are more tolerable, if not perfect.) Non-productivity of discussion is entirely within the realm of reasonable a priori debate, though.

  7. Peter Woit says:

    Henry Warwick, Magnema,
    Since we seem to agree thinking about the simulation argument is a waste of time, let’s not further discuss it here, OK?

  8. Tim May says:

    When I was in high school, there were some speculative reports in places like “Science News” and even in some issues of “Scientific American” about things like tachyons. Even by 1970 some exciting stories about black holes and wormholes.

    Black holes were sort of confirmed when I was in college (Cygnus x-1) and are of course now mainstream.

    There were later tales of possible magnetic monopoles. Interesting stuff.

    But I never saw a “world view” based on these speculations of tachyons, wormholes, monopoles, etc. Hype was not bad at all in those days. A few speculative articles, even the famous Bryce DeWitt article/book on the “Many Worlds” interpretation of QM, but all within the basic realm of plausibility.

    Not until “string theory,” going on nearly 40 years ago.

    I thank some of the string theory skeptic sites, like this one, for putting a (perhaps slight) lid on the bubbling cauldron of hype.

    –Tim May

  9. Justin says:

    I’m curious if HEP is healthier outside the US. Does anybody know if there is a better culture promoting independent thinking in Europe or does it suffer from the same problems we have here in the US?

  10. Jeff M says:


    Enjoyed the piece with Horgan. The differences between physics and math are fascinating, I think fundamentally the issue is that in math you can show that something is correct, but in physics (and essentially everything else) you can only show that something is incorrect. Mathematics is the only discipline where “this is right” is a truly meaningful statement. I once argued with my father in law, who was an art professor, about which was harder to teach. He though art was, since you had no way to judge what was right. I said math was, since I had to figure out a way to get people to learn the right way to do things, I couldn’t let them do it any other way.

  11. new says:

    Has any string theorist like Edward Witten, Briane Greene Michio Kaku Gates et al, commented on the latest SUSY Moriond result? How does this impact string theory?

  12. Magnema says:

    @Peter Woit: Before I quit, I want to clarify that I meant “from the perspective of a scientist.” I think philosophers should consider it, because if it’s wrong, then why it’s wrong would be something interesting to study in that field. Like math, philosophy has different standards of argument and different paradigms for useful. OK, done talking about it now.

  13. Shantanu says:

    Peter, something else I would have mentioned. For many years string theorists claim to predict a dark matter candidate. Then recently Verlinde has argued that string theory dispenses with the need for dark matter. How can a theory predict two opposite hypothesis? String theorists should make up their mind about this.
    (Same with principle of equivalence and I have heard similar opposing claims)

  14. Bernhard says:


    This is own take but from experience, I do not think HEP is healthier in Europe.
    I can’t also not think of any fundamental differences in culture in regards to promoting independent thinking, not by a long shot. The US and Europe share, in this respect at least, the same culture and if you found yourself locked inside an US or and European institution, you would not be able to spot any differences, in my opinion. Also, the mechanisms which exist to stall independent thinking exist both in the US and Europe in the same way (like the way grant decisions are made, etc).

    The US and Europe collaborate extremely closely in HEP, just look at the conferences, who attends it, who works where, etc. The geographical distance is not translated to any significant difference in scientific culture. And the rest of the world sill looks at their example for reference (including China and Japan), since historically most of the major advances came from one place or the other.

  15. Avattoir says:

    I’m interested in knowing Peter Woit’s view on the opening point in Jeff M’s comment:
    “… math … can show … something is correct, … physics … only … that something is incorrect”.
    On Jeff M’s dw with his father in law, my impression is they somehow agreed to equate “harder” and “more uncertain”. Surely it’s not the case that teaching competence in a given field varies according to its participants’ tolerance for bullshit.

  16. Tim May says:

    Peter, this is not directly related to this thread, but..

    Tonight I saw the biopic “The Man Who Knew Infinity,” about Ramanujan and Hardy (and Littlewood in a minor role). Well done. Probably not comprehensible to anyone who has not looked at infinite series and convergences.

    One thing that struck me in the end credits was something along the lines of “His work is now helping us to understand black holes.”

    Say what? Surely he deserves even more credit for predicting the multiverse.

    I looked into and found a quote from the ever-speculative (though fun to read) “New Scientist”:

    regarding some of Ramajun’s work:

    “Devised by Ken Ono of Emory University in Atlanta, Georgia, the formula concerns a type of function called a mock modular form (see main story). These functions are now used to compute the entropy of black holes. This property is linked to the startling prediction by Stephen Hawking that black holes emit radiation.

    “If Ono has a really new way of characterising a mock modular form then surely it will have implications for our work,” says Atish Dabholkar, who studies black holes at the French National Centre for Scientific Research in Paris. “Mock modular forms will appear more and more in physics as our understanding improves.”

    My thinking is that Ramanujan’s work stands on its own. Co-opting him into the publicity brigade is akin to how Hardy is now being treated as a kind of father of secure credit card transactions. (And, yeah, I know about number theory and have worked in crypto since the 1980s.)

    I don’t dispute that some of the partition work may relate to entropy, etc. Just that claims about how Ramanujan has helped the world to understand black holes seems like journalistic hype. Probable intended for sex appeal, in the abstract sense.

  17. vmarko says:

    Tim May,

    If you want to calculate the statistical entropy of an ordinary Schwarzschild black hole in the context of loop quantum gravity (by counting all possible ways the horizon area can be constructed from pieces which belong to the spectrum of the area operator of LQG), then you need to use one of the Ramanujan’s formulas to evaluate the entropy.

    It has to do with the math problem (from number theory) of counting all possible ways a large integer can be partitioned into summands, and Ramanujan was the first one to solve that problem.

    HTH. 🙂

  18. Jesper says:

    @ Justin, Bernhard,

    Although my experience with the US is limited I do think that there could be a slight difference between Europe and the US with respect to theoretical physics. I think that there is more diversity in Europe, where there are communities with a strong tilt towards mathematics, for instance algebraic QFT and those involved in noncommutative geometry. Once could also mention those working on asymptotic safety around Martin Reuter and Jan Ambjørns Dynamical Triangulations as examples – there are more. In my experience String theory is powerful in Europe, but not as powerful as it appears to be in the US. Perhaps others can confirm this.

  19. Mitchell Porter says:

    Tim May said

    ‘I never saw a “world view” based on these speculations… Not until “string theory”’

    String theory isn’t a random speculation. It was discovered by Veneziano *before* the standard model, supersymmetry, grand unification…, in an attempt to describe hadrons. It turned out to contain fermions, gauge bosons, and gravitons, as well as being the natural endpoint of the quest to unify through symmetry. There is nothing else like it.

    The only serious strategic shift that string phenomenologists face so far, is a need to abandon the idea of weak-scale supersymmetry, and that was never an implication of string theory, it was a phenomenological assumption.

  20. NoGo says:

    I have also read the linked interview with Witten, where he says:
    “Witten: There are not any interesting competing suggestions. One reason[…] is that interesting competing ideas (twistor theory, noncommutative geometry, …) tend to be absorbed as part of a larger picture in string theory. The competing interesting ideas have been very fragmentary and have tended to gain power when absorbed in string theory.”
    I am curious, and will be grateful if someone in the know can comment on, were these theories really “absorbed” by string theory in some meaningful way, or it’s just another conjecture about properties M theory may have when it’s found, or something in between?

  21. Jesper says:

    @ NoGo,

    let me comment on noncommutative geometry. The ‘non-commutative geometry’ aspect emerging from ST does, in my opinion, not compare to Chamseddine and Connes approach – so I would definitely not accept that NCG has simply been “absorbed” into ST.

    These two approaches to understand the standard model are in fact radically different: ST proposes via the landscape that the standard model is, essentially, a random occurrence whereas NCG proposes the exact opposite, that the standard model is unique.

    Generally speaking, I think that if you should have a framework, that is large enough to encompass all other existing ideas, then it will most likely be empty.

  22. Peter Woit says:

    Mitchell Porter,

    The problem with string theorist’s “strategic shift” of dismissing the failure to find SUSY or extra dimensions at the LHC as just a “phenomenological assumption” is that finding such things at the LHC was the main argument string theorists used for the last thirty years when asked “how can your theory be tested?”. If they make this new “strategic shift” they will be acknowledging that there are no longer any plausible prospects for experimentally testing string theory, putting the idea of string theory unification firmly outside conventional understanding of what is science.

  23. Peter Woit says:


    The question of the relation of twistor and non-commutative geometry to string theory is an extremely complicated one, with the first problem deciding what “string theory” is. As Jesper points out, I don’t think non-string theorists would agree that these ideas have been “absorbed” into string theory.

    Looking at active research topics pursued by “string theorists” in recent years (for instance by looking at talks at Strings 20XX) I think one finds “string theorists” moving into other fields. A very active area for instance is the study of scattering amplitudes, and while ideas coming from string theory have had some impact, a twistor theory expert might reasonably make the claim that ideas from twistor theory have been even more important there. If “twistor theory” was a term used in the way “string theory” is, a “twistor theorist” might have a good case that “string theory” has been absorbed by “twistor theory”.

    Of course, this level of discourse is just playing with ill-defined words, shall we say, “not even wrong”…

  24. martibal says:

    Last march there have been two colloquia at ICTP in Trieste, one day by Vafa, the next day by Connes. I do not know if the videos are available, but it is difficult to imagine two approaches to theoretical physics more different.

    The claim that noncommutative geometry [NCG] would have gained power once absorbed within string theory [ST] is pretty much ambiguous. If one intends “power of scientific explanation”, then this claim is a nonsense. If one intends “political or sociological power”, then it may be true that NCG has gained visibility within the theoretical physics community when Seiberg-Witten map was – for a while – the hot topic in ST. But this period of time is over. I do not know what ST has kept from NCG, what I do know is that research in theoretical physics inspired by NCG has kept going on, with motivations that have nothing to do with ST, and with many beautiful developments that have nothing to do with ST neither.
    That gaining visibility or “media power” is not the same thing as gaining “scientific power” is the whole raison d’´être of this blog, no ?

  25. A.J. says:

    Noncommutative Geometry is more than just Connes’ attempt to get the classical Standard Model Lagrangian by looking at the Dirac operator on a limited set of NCG spaces. It’s a very general way of thinking about ‘spaces’. I’d be willing to bet that when Witten said that NCG had been enriched by String theory, he was referring to the variety of nontrivial ways in which string theory realizes such noncommutative spaces. This includes the Myers Effect and various noncommutative gauge theory constructions, and also more general ideas like matrix theory. Such realizations are often powerful because they give you new ways of looking at familiar examples and suggest non-trivial relations to other concepts (e.g., the Seiberg-Witten map).

    That said, Connes’ specific program of connecting Standard Model dynamics to noncommutative geometry can be thought of as a special case of the superconformal QFT geometry dictionary used in string theory. Urs Schreiber has pointed this out here a few times. It’s quite peculiar that enthusiasm for this model instead of string theory has become a tribal marker.

  26. The close relation between Connes-style NCG and perturbative superstrings has been pointed out way back in 1993 by Fröhlich and Gawędzki , has meanwhile been substantiated by theorems by Roggenkamp and Wendland, following Kontsevich and Soibelman, and has been used by Soibelman to study aspects of the string landscape in terms of spectral triples. Review and comprehensive pointers to the literature are at PhysicsForums-Insights: Spectral Standard Model and String Compactifications. Apart from the technical details discussed there, one highlight is that both approaches are KK-models that do agree on the critical spacetime dimension to be 4+6 = 10 (modulo 8, because NCG only sees it modulo 8).

  27. Peter Woit says:

    AJ and Urs,

    Looking at the document Urs links to, I see all sorts of interesting mathematics and physics, but I don’t see how the perturbative superstring has all that much to do with it. As usual when I disagree with Urs, what seems to me a tenuous and not very important connection is for him a “close relation”. Sorry, but all this just seems to me to be turning interesting discussion about deep mathematical structures into a waste of time ideological discussion, and I’m not in the mood to participate, or even try and properly moderate such a discussion.

  28. piscator says:

    Europe is less faddish than the US. The tendency for everyone to jump on the same hot idea is less. This is good when the hot idea is evanescent nonsense and less good when the hot idea is a fundamental breakthrough.

    It also involves many different countries, which encourages a larger number of distinct scientific hierarchies as to styles of science and the question of what are the important problems one should ask.

    But still, things are more similar than different. Everyone reads the same arxiv.

  29. Jesper says:

    Hi A.J. and Urs Schreiber

    I don’t doubt that there is are connections between noncommutative geometry and string theory – and the one with 2d SCFT is obviously interesting – but the question was whether string theory had “absorbed” noncommutative geometry. Noncommutative geometry is a very large field, mostly in mathematics but clearly also in mathematical/theoretical physics and the notion that string theory somehow captures all that doesn’t make sense to me. But perhaps this is a question of semantics.

    Urs, in your very nice PhysicsForum piece you end with the sentence “A very interesting question to ask therefore is: which 2d SCFTs (if any) would lift the Connes-Lott-Chamseddine-Barrett spectral triple? These would be realistic string vacua.” Does that imply that it is not yet known whether the connection between the Connes-Lott-Chamseddine model and these 2d models is real?

    In my view, Chamseddine and Connes’ work (and others) on the standard model is an interesting observation that might point towards a unified theory of quantum gravity. The questions are where the almost commutative structure originates from and how quantum field theory/quantum gravity fits into the picture. The answer could be string theory, it could be something else – I believe that we needs as many ideas as possible.

  30. martibal says:

    Claiming that NCG is subsumed by ST (or gained more power thank to ST) because spectral triples might be useful to study the landscape, or because both view spacetime as a dimension 2 objects (like several other approaches to QG by the way) sounds to me a bit like claiming that Riemann integral has gained more power since it has been widely used in ST.
    It is not because something is used in ST that it is “subsumed” by ST. This tendency of ST to swallow everything and deny independent interest to any ideas outside ST is unbearable, and counterproductive: first it puts the discussion on a conflictual tone, that not everybody is interests to get along with; second it is a way not to study the motivations of the other ideas, that have nothing to do with ST.
    But I agree with Peter that this kind of discussion just repeats itself again and again, and is a waste of time.

  31. Giotis says:

    Witten obviously is referring to the early 2000 medium size Noncommutativity revolution in String theory sparked (mainly) by his by now classic paper written together with Seiberg (you need basically a strong B-field to make Noncommutativity manifesting itself)

    “String Theory and Noncommutative Geometry”:


  32. Anonyrat says:

    Linked from the above: how do physical theories generally make predictions anyway?

  33. RandomAnonymous says:

    ^ That is a depressing amount of effort spent to reparameterise failure…

  34. Urs, in your very nice PhysicsForum piece you end with the sentence “A very interesting question to ask therefore is: which 2d SCFTs (if any) would lift the Connes-Lott-Chamseddine-Barrett spectral triple? These would be realistic string vacua.” Does that imply that it is not yet known whether the connection between the Connes-Lott-Chamseddine model and these 2d models is real?

    Every 2d superconformal field theory yields a spectral triple as its “point particle limit”. (This is a mathematical formalization of the familiar idea that we may view the quantum superstring from far away and see only its center of mass point motion, with the quantum oscillations about it appearing as different species and different internal degrees of freedom of the resulting effective spinning particle.)

    But not every spectral triple necessarily needs to arise this way. Those that do would be called those that have a stringy UV-completion to a theory of perturbative quantum gravity coupled to gauge bosons and fermions.

  35. Witten obviously is referring to the early 2000 medium size Noncommutativity revolution in String theory sparked (mainly) by his by now classic paper written together with Seiberg (you need basically a strong B-field to make Noncommutativity manifesting itself)

    That’s quite certainly what the quote was referring to, but since above the discussion quickly shifted to Connes-style NCG, it is worth recalling that this “spectral geometry”, as it might possibly better be called, is a good bit richer than just some Moyal-noncommutativity of spacetime coordinates, and was understood to arise as the point particle limit of perturbative superstrings already in FröhlichGawędzki 93 .

    One could argue that the main point of Connes-style NCG (namely: spectral triples) is not so much the non-commutativity of the algebra that is part of the triple (that’s a nice side effect, that non-commutative algebras may be accomodated, too) but the main point is the Dirac-like operator in the triple (and the super-Hilbert space that, necessarily, comes with it, for it to be a linear operator on anything).

    Instead, Connes-style NCG is about extracting an effective target spacetime geometry (possibly non-commutative, sure) from the energy spectrum of a super-particle (hence a spinning particle), hence to extract an effective target spacetime geometry as seen from quantum super-particles that roam in it.

    This is noteworthy, since this is exactly foundational approach of perturbative superstring theory, only that there the worldline theory of that spinning particle is promoted to a worldsheet theory, since that turns out to be beneficial for the behaviour of the theory, as it provides a consistent means to incorporate interactions and counter-terms to arbitrary order (the higher string oscillation modes).

    Moreover, as has been shown, this gives a richer supply of non-commutative backgrounds than Seiberg-Witten, in fact it gives a general stringy idea of how to think of the non-commutativity in a Connes-style non-commutative geometry: The non-commutativity is the remnant of the higher superstring oscillations as we pass to the superstring’s point particle limit.

  36. lars says:

    Horgan: Do you still think string theory is “not even wrong”?

    Woit: Yes. My book on the subject was written in 2003-04 and I think that it’s point of view about string theory had been vindicated by what has happened since then. Experimental results from the Large Hadron Collider show no evidence of the extra dimensions or supersymmetry that string theorists had argued for as “predictions” of string theory. The internal problems of the theory are even more serious after another decade of research. These include the complexity, ugliness and lack of explanatory power of the models designed to connect string theory with known phenomena, as well as the continuing failure to come up with a consistent formulation of the theory” — Peter Woit from linked article

    You once said that “I use ‘not even wrong’ to refer to things that are so speculative that there would be no way ever to know whether they are right or wrong”

    If that was the definition for “not even wrong” that you used when you wrote your book, I can see how an increase in the severity of the internal problems you refer to might vindicate your initial claim that string theory is “not even wrong”, but it is not clear how results from the LHC could have vindicated such a claim.

    “No way ever to know whether they are right or wrong” would seem to imply “no experiment”.

    Please note that I am not challenging your claim about the internal problems, about which I am not qualified to weigh in one way or another.

  37. Peter Woit says:


    I put “predictions” in quotation marks for a reason. Despite what they sometimes told the press, string theorists never had real predictions for the LHC from string theory. They did however have things they could point to which, if observed, would provide encouragement for some of the research directions that led to string theory (e.g. SUSY). It is these that have not worked out.

Comments are closed.