There’s a very interesting new paper on the arXiv by Joe Polchinski, a survey article for Studies in History and Philosophy of Modern Physics, entitled just Dualities. It’s an unusually lucid summary of the story of dualities in quantum field theory and string theory. This is a very complex subject which has been a central one in theoretical physics for the last few decades, but most expository writing on the subject has tended to be either superficial promotional material or mired in technical detail obscuring fundamental issues.

One reason for this is that, as Polchinski does an admirable job of making clear, in a very real sense we still do not understand at all the fundamental issues raised by these dualities. He notes that “we are still missing some big idea”, and points to the same comments from Nati Seiberg last month that I blogged about here. For most of the dualities at issue, our current standard technology for dealing with QFTs (the Lagrangian and the path integral over classical fields) is capable of capturing the two QFTs that are in some sense “dual”, but we lack a viable larger framework that would give the two QFTs in two different limits and explain the duality relationship.

For an example of the problem, probably the oldest and most well-studied case where we are missing something is Montonen-Olive duality, a non-abelian duality between electric and magnetic charges and fields. A currently popular idea is to find the explanation of this in “Theory X”, a 6d superconformal QFT, with duality coming from compactifying the theory on a torus (for more about this, see talks last week in Berkeley). The problem with this is that we don’t have a definition of the “Theory X”.

Polchinski places this problem in the context of a conjectural “M-theory” with various string theory limits. This has been the dominant idea in the subject for nearly 20 years now, but we seem no closer now to finding an actual realization of this conjectural picture than we were back in the mid-90s. Twenty years and thousands of papers have just given better understanding that various possible ideas about this don’t work.

One place where I think Polchinski’s survey is weak is in the treatment of this conjecture, where at times he takes as solid result something highly conjectural. For instance he starts off at one point with:

String-string dualities imply that there is a unique string/M-theory.

and moves on to the conjecture that

In this sense it may be that every QFT can be understood as a vacuum state of string/M-theory.

The problem here is that he’s built a speculative view of the unification of physics, constructed on an assumption about a “unique” theory, when we don’t know at all that such a thing exists. One basic lesson of mathematical research is that you need to keep very clear the distinction between what you really understand and what is speculation, because your speculation is often wrong and if so will lead you in the wrong direction. I think particle theory of recent decades likely suffers from people forgetting that some ideas are speculative, not firmly grounded, and may be pointing in the wrong direction.

One wrong direction this takes Polchinski is to the non-predictive, pseudo-scientific landscape of supposed string theory solutions and the multiverse, which he blithely invokes as our best fundamental explanation of physics. Tellingly, unlike the clear explanations of other topics, here he makes no attempt to describe these ideas other than to note that

they rest on multiple approximations and no exact theory.

In a final section, Polchinski addresses the question of what all this tells us about what is “fundamental” and what is the role of symmetries. This is the crucial question, and I’d argue that our lack of understanding of where these dualities come from likely is due to our missing some understanding of how symmetries are realized in QFT or string theory. This has been the lesson of history, with the Standard Model only coming into being when people better understood how symmetries, especially gauge symmetries, could act in QFT. Polchinski largely takes the opposite point of view, arguing that the fundamental theory maybe has no symmetries, local or global. He quotes Susskind as suggesting that symmetries have nothing to do with fundamental equations, are just calculational tools for finding solutions. I think this is completely misguided, that a strong case can be made (and I do it here) that “symmetry” (in the sense of the mathematics of groups and their representations) lies at the very foundation of quantum mechanics, and thus any quantum mechanical theory, even string/M-theory, whatever it might be.

Wondering whether there will be an arXiv trackback to this, and whether Polchinski has something to say about it…

Update: The arXiv Monday evening has a large collection of excellent review articles entitled “Exact results on N=2 supersymmetric gauge theories”, edited by J.Teschner (first is arXiv:1412.7118, last arXiv:1412.7145). Some of the results reviewed are based on deriving implications of the existence of the 6d (2,0) models discussed here and in the comment section.

Update: I’ve put this blog posting in the Multiverse Mania category, not because of the posting content, but because of comments in the comment section from Polchinski and Bousso.

This entry was posted in Multiverse Mania. Bookmark the permalink.

61 Responses to Dualities

  1. CIP says:

    Polchinski notes the analogy between S duality and the duality of descriptions in the Fourier transform. Such dualities of description have seemingly become ubiquitous in math, exposing deep connections between algebra, arithmetic, and geometry.

    Even the wave-particle duality that so confounded early quantum mechanics fits the pattern. Despite Susskind, these things are probably all quite deep.

  2. Giotis says:

    Well you don’t have to resort to 6d SCFT to give a geometric origin of S-duality; as it is described in the paper the Montonen-Olive duality is explained by IIB self-duality.

    Moreover IIB self-duality on a circle is geometrically realized by compactifying M-theory on a 2-Torus (or similarly from F-theory) via modular invariance (i.e. the diffeomorphism invariance of the Torus).

    So overall the very existence of M-theory explains S-duality.

  3. Joe Polchinski says:

    Thank you for your review. Indeed, the landscape discussion was a bit of an aside and perhaps should have been omitted as a distraction. My remarks on symmetries are very much outside the orthodoxy and I was glad for the opportunity to present them.

    About trackbacks, you seem to think that am responsible. I once stated my opinion, and someone, I don’t know who, acted in response. I stand by my opinion, and believe that whoever acted did so correctly. It should be noted that your blog has no actual science content. There is science news, science opinion, science criticism, science sociology, occasional science invective, but no actual science. (Lubos, who also seems not to be tracked, does have science content but also has so much that is inappropriate that his signal/noise is miniscule). There is a primary literature of science, and a secondary literature of science criticism etc. It is great that the secondary literature exists and links the primary literature, but this is one-way, the primary literature should strive for a maximum signal/noise.

    p.s. Regarding “they rest on multiple approximations and no exact theory,” you might note that the preceding arXiv paper was directed at examining these approximations and making them more precise.

  4. Peter Woit says:

    Thanks for the comments and thanks for writing the thought-provoking article. Your comments about symmetries struck me as quite similar to ones I’ve seen from Arkani-Hamed, especially about gauge symmetry, I think they’re less outside current orthodoxy than you think. As for the symmetries “a calculational tool, not fundamental” point of view, Howard Georgi was warning students like me taking his Lie algebras class back in 1978 along much the same lines (it’s not original with Susskind)

    About your “no actual science” claim I’ll have to respectfully disagree. There are lots of different things on this blog, but it’s not unusual for it to contain more actual science than a typical arXiv paper on the landscape…

    In any case, “actual science” by your definition doesn’t seem to be the criterion that the arXiv is using to bar trackbacks to my blog, since they now allow trackbacks to a wide range of journalistic sources (checking at random those from yesterday, one is to the Atlantic magazine). Despite putting a lot of effort into this, I haven’t been able to get an answer from the arXiv to the question of what criterion they are using to bar this blog. If you happen to find out, please let me know. At this point the only information I have about this is that you’ve called for such a bar and that has been very influential.

  5. Flavio says:


    do not get worried about Joe’s remarks about your blog. How can you two get along? Each of you two tells the other, politely, that he is not doing science. And you both believe this assessment of the other to be true, with every cell in your body.

    There are only for possibilities: either of you can be right or wrong. But all four possibilities are of no help. But maybe I am wrong; maybe you see it differently.

    Progress can only come by getting back doing science, by returning to the search for a solution to fundamental physics.

  6. M says:

    Recognising promising and failed lines of research is a key part of science. “Not Even Wrong” should have been on arXiv

  7. Peter Woit says:

    I don’t at all think he’s not doing science. The great majority of the article was about important and interesting scientific work. Better understanding these dualities is a great scientific problem, with both physical and mathematical aspects. As for the string landscape, for lots of it I’m willing to admit it’s science, just really bad science…

  8. Mike Duff says:


    I agree with your remark “So overall the very existence of M-theory explains S-duality’’. S-duality in D=4 follows from string-string duality in D=6 (hep-th/9501030 ) which in turn follows from M-theory in D=11 ( hep-th/9506126 ). (Don’t usually respond to blogs, but here I made an exception since M-theory is taking a bashing and this work somehow got overlooked in Joe’s review.)

  9. Peter Woit says:

    Mike Duff/Giotis,
    The problem with the “M-theory explains S-duality” argument is that, as Polchinski is careful to point out, you don’t understand what M-theory is. The reason I mentioned the explanation in terms of the 6d superconformal QFT is that that seems to be something one has a chance of understanding independently of a full understanding of M-theory.
    I seriously do think people need to keep clear what it is that they have a solid understanding of, and what is conjecture. Kudos to Polchinski for doing this somewhat better than usual.

  10. Giotis says:

    Yes of course professor Duff, what you mention is another classic derivation of S-duality from M-theory that was not included in the paper.

    Peter not sure what you mean; we know enough to make that claim with confidence.

    The reason why people often focus on the 6d SCFT (e.g. for the Langlands correspondence) and not on the Stringy/M-theoretic explanation of S-duality mentioned by professor Duff, is explained in page 5 of the seminal paper by Witten “Some Comments On String Dynamics” (http://arxiv.org/abs/hep-th/9507121).

    Anyway the 6d SCFT is a prediction of IIB or M-theory.

  11. Peter Woit says:

    What I mean is that you can’t claim to have explained something by showing that it is an implication of something else that is just as poorly understood. You’re just relating one conjecture to another conjecture. If M-theory was much better understood than 4d qft that sort of relation might have explanatory power, but the opposite seems to be true.

  12. Mike Duff says:


    Re S-duality from M-theory: Thanks for pointing out page 5 of the seminal paper by Witten, “Some Comments On String Dynamics” (hep-th/9507121). On this topic, see also page 33 in another seminal paper by Witten, “String Theory Dynamics in Various Dimensions’’ (hep-th/9503124).

  13. Peter Woit says:

    Giotis/Mike Duff,

    I just took a look at the page recommended in hep-th/9507121, and I was amused to notice that Witten makes a version of the point I’ve been trying to make:
    “Thus, if one asks, “How can the S-duality of N= 4 Yang-Mills theory be made
    obvious?” one answer is that this can be done by embedding N=4 supersymmetric
    Yang-Mills theory in the heterotic string and then mapping to a Type IIA theory by using string-string duality. The weakness of this answer is that it embeds the gauge theory in a problem with many other features – such as gravity – that may not be material. One would like to “flow to the infrared,” eliminating as many degrees of freedom as possible, and obtaining the minimal theory in which the S-duality is still manifest. The self-dual string in six dimensions may be the answer to this question.
    The self-dual string in six dimensions does not look easier than the Type IIB model that we started with; certainly we understand it less. Nevertheless, it might be the right structure for understanding the four-dimensional field theory.”

    Witten talks here about the self-dual non-critical 6d string, these days it seems that attention is now on the simpler 6d superconformal QFT, presumably easier to understand. But the point is the same: embedding something you don’t understand in something much more complicated you don’t understand isn’t a great explanation. Much better is to embed in the simplest thing that has the properties needed to explain what you want by this mechanism. What you want here follows from the 6d superconformal QFT, which is hard enough to understand. Embedding this in string theory or M-theory just puts you in a more complicated context, while it’s the 6d properties that have some explanatory value.

  14. Mike Duff says:

    Peter Woit

    Re your view that “The reason I mentioned the explanation in terms of the 6d superconformal QFT is that that seems to be something one has a chance of understanding independently of a full understanding of M-theory. I seriously do think people need to keep clear what it is that they have a solid understanding of, and what is conjecture.”

    Here is what Greg Moore says (http://www.physics.rutgers.edu/~gmoore/Strings2011FinalPDF.pdf).

    ”Important characteristics of the six‐ dimensional (2,0) theories S[g]: These theories have not been constructed – even by physical standards ‐ but some characteristic properties of these hypothetical theories can be deduced from their relation to string theory and M‐theory.”

  15. David Ben-Zvi says:

    While it’s certainly true that there is no mathematical definition or construction of the 6d SCFT (or even physical, as Mike Duff quotes from Greg Moore), the kind of object it is is clear mathematically, and one can formulate precise conjectures/axiomatics for its properties. Of course the main evidence for any of these constructions comes from string theory and M-theory, but on the other hand there are mathematical theories very close to it (eg its dimensional reductions or its Coulomb branch deformations) which can be defined and understood using current mathematical techniques.

    For example in the workshop Peter links to Kevin Costello presented a mathematical definition of an approximation to theory X (the (2,0) theory) — roughly speaking it’s the part of the theory seen by its local operators and surface defects, but ignorant of the 4-dimensional defects, and not enough to explain S-duality. This part makes sense as a perturbative quantum field theory and agrees well with many predictions in physics.

    At least as far as mathematics goes, this is one of the great appeals of theory X – it’s extremely useful to have a conjectural structure with this powerful predictive power (eg Langlands duality) that’s close to things we understand or at least can formulate precisely; while we’d love to “go to 11” and see indescribably further through M-theory, this seems very far from the capabilities of contemporary math.

  16. Giotis says:

    Exactly professor Duff. Thanks for pointing out this excerpt.

    @Peter some remarks:

    As I have mentioned above 6d SCFT is a prediction/achievement of the second superstring revolution.

    The best way we have to describe the theory so far is via AdS/CFT. Indeed the theory is dual to M-theory on AdS7xS4.

    Although the theory sits in a conformal fixed point we should not forget that even in six dimensions it is the low energy limit of 6d LST (Little String Theory). Moreover the theory itself is known as the theory of tensionless strings in six dimensions (though of a solitonic nature).

    Finally the discussion on whether we have to resort to the full machinery of String/M-theory is far from closed. Witten just made an argument at that point in time without expressing any certainties. Check for example his remark in chapter 4.2(page 15) of his paper “Geometric Langlands From Six Dimensions”. This statement is similar (but stronger) to Moore’s comment referenced by professor Duff above. Moreover Meng-Chwan Tan reproduced (http://arxiv.org/abs/0807.1107) the field theoretic results of the aforementioned paper regarding Langlands correspondence using the dualities and machinery of M-theory.

    From another stringy point of view S-duality reduces to T-duality as it was demonstrated by Vafa in his classic paper “Geometric Origin of Montonen-Olive Duality” (http://arxiv.org/abs/hep-th/9707131).

    Finally we should not forget that S-duality and T-duality are just parts of a bigger U-duality group which can only be understood within the general context of M-theory.

  17. Chris Austin says:

    On the issue of whether or not the 6d interacting superconformal field theories have Lagrangians: on pages 10 and 14-15 of 0712.0157, (cited by Giotis in the post on Weinberg and Seiberg), Witten seems to argue that an obstruction to these theories having Lagrangians in 6 dimensions is that when you compactify them on an S^1 of radius R to 5 dimensions, R occurs in the denominator of the resulting d = 5 Yang-Mills action, (so that the Yang-Mills coupling constant is proportional to \sqrt{R}), rather than in the numerator, as would normally be the case for a KK reduction. But it seems to me that a qualitative counter-example to this argument is provided by Witten’s own work on obtaining type IIA superstring theory by compactifying d = 11 supergravity on an S^1 of radius R, on page 4 et seq of hep-th/9503124: the resulting superstring coupling constant is proportional to R^(3/2). The reason for this is that the type IIA superstrings result from solitonic d = 11 2-branes that wrap the S^1, (the 2-branes are the classical solutions of d = 11 supergravity found by Duff and Stelle). Could it not be possible that the 6d (2,0) theories have Lagrangians that admit classical 1-brane solutions, and on compactification on an S^1, the d = 5 Yang-Mills fields arise from 1-brane configurations that wrap the S^1?

  18. To amplify again what Giotis mentions above: there is an actual definition of the 6d (2,0)-CFT is via AdS7/CFT6 holography and it translates to a good mathematical definition in a suitable limit.

    This deserve a reminder from time to time. In that same String2014 “vision talk” already mentioned above, “Physical Mathematics and the Future“, Greg Moore in section 5.1 lists some ideas of how to capture the 6d (2,0)-SCFT, only to conclude (on p. 28) that:

    As stressed to me by Edward Witten, thanks to the AdS/CFT correspondence, we do have an understanding of the complete solution of the theory in the large rank limit for su(N) theories

    In the limit in which the Chern-Simons terms in the dual 7d theory dominate, this even becomes essentially a mathematical precise definition, where the partition function of the 6d theory is obtained from the geometric quantization of the 7d Chern-Simons theory in direct higher analogy to how the 2d WZW model arises from 3d Chern-Simons theory (CS3/WZW2 correspondence). And this is the way that in the abelian case and for just the 2-form sector Witten has studied the 6d theory since

    – E. Witten, “Five-Brane Effective Action In M-Theory”, J. Geom. Phys.22:103-133,1997 (arXiv:hep-th/9610234)

    – E. Witten, “AdS/CFT Correspondence And Topological Field Theory” JHEP 9812:012,1998 (arXiv:hep-th/9812012)

  19. Peter Woit says:

    I’m a bit dubious about your argument that at large N, AdS7/CFT6 is nothing but the phenomenon we understand in terms of geometric quantization for AdS3/CFT2. Wouldn’t the same hold true for the well-known AdS5/CFT4, and is that really all there is to that in large N? I’ve never heard anyone make that claim before.

    In any case, the initial point remains that you’re trying to explain one conjecture (electro-magnetic duality) in terms of others that are equally if not more mysterious (AdS/CFT + M-theory).

  20. In the limit where the CS terms dominate. That’s the content of

    — E. Witten, “AdS/CFT Correspondence And Topological Field Theory” JHEP 9812:012,1998 (arXiv:hep-th/9812012):

    This flux [in 4d SYM] is encoded in the AdS/CFT correspondence in terms of a five-dimensional topological field theory with Chern-Simons action. A similar topological field theory in seven dimensions governs the space of “conformal blocks” of the six-dimensional (0,2) conformal field theory.

    For just the abelien self-dual 2-form theory in 6d, it’s very construction in

    — E. Witten, “Five-Brane Effective Action In M-Theory”, J. Geom. Phys.22:103-133,1997 (arXiv:hep-th/9610234)

    is via abelian 7d Chern-Simons theory in direct analogy to CS3/WZW2. This statement later became the main theorem (i.e. was made mathematically rigorous) in

    — M. Hopkins, I. Singer, “Quadratic Functions in Geometry, Topology, and M-Theory” J. Differential Geom. Volume 70, Number 3 (2005), 329-452. (arXiv:math.AT/0211216)

  21. Peter Woit says:

    You’re just talking about a topological limit or sector of the theory. I don’t see how that gives you electro-magnetic duality of the 4d theory, that seems to me to require behavior under conformal transformations that you can’t see in the TQFT limit, or am I missing something?

  22. The 7d CS theory is topological, not the 6d theory that it encodes holographically. Just as for CS3/WZW2. The conformal structure in 6d arises as the polarization structure for the geometric quantization of the 7d theory.

  23. Peter Woit says:

    OK, but based on what I do understand here (the subtleties of getting from 3d Chern-Simons to a 2d CFT), I’m suspicious that to do what I don’t understand here (get from 7d Chern-Simons to a 6d CFT) you need to make some conjectural assumptions at least as strong as electro-magnetic duality.

    In any case, what you seem to be claiming is that one just needs to understand 7d Chern-Simons, not the full M-theory, to get non-abelian duality.

  24. Peter Woit says:

    And I’m still wondering about AdS5/CFT4. 5d Chern-Simons is enough to give you 4d large N CFT??

  25. No. Check out the articles that I pointed to for the statement.

  26. Joe Polchinski says:

    I don’t speak for the arXiv, but clearly this is a judgement call, a question of signal/noise. Let me explain what I mean by `no science.’ Your most recent post on the landscape:
    “The latest news (arXiv:1410.7776) is that KKLT doesn’t actually work, that you can’t get stable string vacua that way. I don’t think though that this will have any effect on multiverse mania and its use as an excuse for the failure of string theory unification…”

    First, you seem to just assume that this paper is correct. An authority on the process of science might be expected to recognize that just because a paper appears on the arXiv, it cannot be assumed to be correct. Second, there is absolutely no discussion of the content of the paper, or what are the scientific issues involved. Rather, you just use it as an excuse to launch into your standard uncharitable assessment of a group of scientists, and to indulge in some of your favorite charged words (“mania”, “failure”). Third, you get both the science and the sociology wrong, see arXiv:1412.5702.

    I see zero signal here, and a lot of noise. And this is a very representative example. So no, I do not think that the arXiv should link to this.

    p.s. Thanks to Mike Duff and others for the discussion of references.

  27. Maurice Carid says:

    Hi Peter,
    with all due respect, but I think Dr. Polchinski has a valid point here.
    For a string-theory/landscape unrelated example: in your entry
    from Nov 3, 2014 “Last week’s hype” you slandered an arXiv entry
    as “university press-release-driven hype” and the press releases themselves as
    “nonsense spread around the media”. When Als and I – both after having read the paper – challenged you to explain what’s wrong with the paper and the press releases, all you contributed was “Maurice/Als,
    I’ve already wasted more time on this than is wise.”
    What useful information would a trackback to your blog entry have
    conveyed to the readers of that paper?
    cheers & no hard feelings!

  28. Peter Woit says:

    Maurice Carid,
    A statement is not slander if it is accurate. I think my characterization of that paper is pretty much the same as the way most physicists would describe it.

  29. Peter Woit says:

    You’re avoiding the question of what the arXiv policy is under which I am banned. What is it? If you don’t know what it is, don’t you think you really should ask someone before publicly defending it? I’m having a lot of trouble taking seriously your claims that you don’t have any responsibility for this or know who does. This disagrees with what I’ve been told, which is that people at the arXiv point to you to justify this decision.

    You seem to indicate that the arXiv policy is somehow based on a “signal/noise” evaluation, one that you have performed and found this blog wanting, largely because it regularly refers to the research program you are heavily invested in as a failure. The obvious problem is that others see what is signal and what is noise quite differently than you do. I suspect that my characterization of the anthropic string landscape program as pseudo-science driven by refusal to acknowledge failure is one that a sizable fraction of the scientific community would agree with (see the Ellis/Silk piece in Nature, although you likely dismiss that too as “noise”). You are very obviously using the “signal/noise” business as an excuse to ban a legitimate scientific point of view that makes you uncomfortable.

    As for the mention here of arXiv:1410.7776, the background there is that I was contacted by someone expert on that topic who felt that that paper and the research program it was part of posed a serious challenge to widespread assumptions among theorists about KKLT moduli stabilization, that this was not getting sufficient attention from other theorists, so they wanted me to make this issue more widely known. My initial reaction was not to get involved on the grounds that the string landscape business, as serious science, is what some would accurately characterize as noise, not signal. In the end I decided to briefly mention it as one of six short items. Rereading what I wrote, I still think it’s precisely on target as regards the sociology and the evaluation of that line of research.

    As for your claim that I assumed the paper was correct and that I’m too dimwitted to realize that papers on the arXiv are sometimes not correct, that’s just ridiculous. It’s interesting to note that my brief mention here did have the effect my correspondent wanted, causing extensive discussion at another blog, and perhaps this even had some influence on arXiv:1412.5702. If what I write here is noise, it’s noise that a lot of people in the field take seriously, and that’s the problem that seems to be bothering you.

  30. Raphael Bousso says:


    You write that you are not too “dimwitted” to recognize that an arXiv paper might be incorrect. We should accept that.

    There are two remaining possibilities. One is that you analyzed arXiv:1410.7776 carefully and convinced yourself that the technical issues it raises support your claim that “KKLT doesn’t actually work.” If so, I’m sure you will have an equally specific response to arXiv:1412.5702, which does not reach this conclusion.

    I would normally recommend that you post your analysis on the arXiv. On the other hand, by posting it here you could silence those who say that your blog contains no science. Anyway, you should post it somewhere, since the technical questions at issue are of scientific interest quite independently of any applications to KKLT and the landscape (which you may feel would not by themselves merit the effort of writing up your analysis).

    The only remaining possibility is that you knowingly choose to advertise as “news” any paper that you perceive as fitting your narrative, without having any idea whether it is correct. I suspect that this wouldn’t bother your clientele, so you might as well let us know right away whether a detailed analysis will be forthcoming from you.

  31. Dom says:

    As an interested lay person, I enjoy reading Peter’s blog, I have tried all of the blogs he links to and although some interest me, only this one I refer to on a daily basis. I’m not actually that concerned (for example) whether the string approach is correct or not, I prefer to read lots of views. I would say one thing though as an outsider, the more confident I am about my approach to the problems I have to solve professionally, the more I welcome sceptical/hostile enquiry. When people with presumably enormous intellects get so upset that people criticise their ideas I can only conclude that there is some underlying and considerable internal lack of confidence. Einstein’s reported reply to the book “100 Authors against Einstein” – namely “If I were wrong, one would be enough.” strikes me as how a confident person deals with it.

  32. Bill says:

    Took a look at recent trackbacks; certainly, this blog would not decrease signal/noise ratio. Was surprised to see how few trackback there are on an average day. Don’t see why anyone would make such a big deal out of (or even have a strong opinion about) this.

    @Raphael Bousso: You can insult Peter all you want, but why such animosity toward his “clientele”?

  33. Bernhard says:

    It’s pretty clear the not linking of the arXiv to NEW has absolutely nothing to do with science given other completely non scientific material they link to. This censorship that reminds us of inquisition times. If the policy of the arXiv was to link only to highly technical and (as best as possible), unbiased material, that is one thing. But since there is no such thing (Motl, really?), the sober conclusion is that the ban is pivoted by string theory critic. By the way, professor Polchinski, if you only “stated your opinion” to someone that happened to have veto powers on the arXiv, that is the same as making the ban yourself, as you are aware of your own influence and power on the field (that you conquered with talent and hard work, but I’m afraid still not justifiable).

  34. Doug McDonald says:

    As to signal/noise, in the blog business, especially blogs like this,
    its not a dichotomy, its a trichotomy:


    I see this blog as being intended as metasignal

  35. Physics postdoc says:

    As a physics postdoc who does GR but not string theory, I’ve learned a lot about dualities in string theory and their relation to the 6D theory from the discussion in this comment section and I’ve picked up some great references. This is typical of my experience with Peter’s blog, and has kept me reading along for about 6+ years. So I disagree with Joe and Raphael. My experience is that this blog has a very high signal-to-noise, and I think this very comment section is an example. Peter deserves all the credit for this, I don’t know anyone who does a better job or puts more energy into moderating great discussions on a physics blog.

  36. Peter Woit says:

    A few explanations of things you don’t seem to understand:

    About my “clientele”. A sizable part of it is string theorists, with the one who contacted me repeatedly to suggest I mention the line of work in question just one of many. When you insult people who find the information on this blog worthwhile, you’re insulting a large number of your colleagues.

    You have it completely backwards if you think that “KKLT doesn’t work” fits my narrative, quite the opposite. I’m a big fan of Bousso-Polchinski, KKLT, and Susskind’s argument for anthropics and the string theory landscape. If you, Joe, or others can convince your colleagues that string theory is a theoretical framework with essentially zero predictive power, that fits my narrative perfectly and I’m happy to support you. I’ve been saying this ever since the landscape business started, for an example from the earliest days of this blog, see
    I mentioned arXiv:1410.7776, despite it going against my favored “narrative” because I thought some people would find it interesting, and it seems they did (I’m still wondering whether it motivated Joe to write that paper…).

    About the technical details in question, of course I’m not going to devote time to them and anyone reading the “KKLT doesn’t actually work” clause of that sentence in context would recognize that I wasn’t vouching for it. I’ve ad nauseam made the case that this kind of construction is not science, but pseudo-science: it’s designed not to give a plausible way of reliably calculating something that can be compared to experiment, but instead to provide an elaborate excuse for not being able to do this.
    I don’t see any point to taking seriously the details of this elaborate excuse. My non-expert take on the details is that Joe likely has it right with his comment that these arguments “rest on multiple approximations and no exact theory.” Maybe you can find an obscure deep corner of parameter space with stabilized moduli and hopelessly complicated solutions, maybe you can’t, but in either case this has nothing to do with serious science (in Joe’s lingo, it’s just “noise”).

  37. Anonymous says:


    Since when is “you must not discuss a disputed preprint, unless you yourself have posted a preprint on the subject” how science is done? Your attitude seems to be that those outside of your club should keep their mouths shut about your field. Regardless of whether the preprint in question is correct, this is deeply deeply unscientific.

  38. Peter Woit says:


    People really should take a look at the posting that so upset Bousso and Polchinski, it’s

    The short sentence
    “The latest news is that KKLT doesn’t actually work, that you can’t get stable string vacua that way.”
    with “latest news” a link to arXiv:1410.777 is the full and complete discussion of the paper in the posting. I don’t see how it’s not obvious that that was nothing more than a pointer to something people might be interested in, just giving the gist of the paper’s claim.

    So, Bousso isn’t objecting to my discussion of a paper, he’s objecting to my mentioning a paper. This is really, really odd.

  39. Maurice Carid says:

    Hi Peter,
    Because you gave us no clue why you think that
    http://arxiv.org/abs/1402.6144 is “university press-release-driven hype”
    I cannot be sure. But in this case the suspicion of Dr. Bousso that you
    “advertise as “news” any paper that you perceive as fitting your narrative, without having any idea whether it is correct” is correct, suggests itself strongly (I’d say “even in case you have no idea…”, though.).
    The above paper is not about the landscape or string theory
    but strictly only about the many worlds of quantum mechanics. You lauded the work of David Wallace on this idea on your post from Dec 1, 2014. The above paper is a much more mathematical and concrete elaboration on this subject than Wallace’s more philosophical musings.
    It thus seems to be excluded that you dismiss the paper in principle because of
    its subject matter.
    There seem to be two remaining possibilities. Either Dr. Bousso’s quote above is correct for this case: u just took a glance at the title, thought it’s about the landscape or “string theory explaining QM” and slandered it.
    Or you actually read it and found some technical flaw in the paper. If the latter is correct I challenge u to briefly reveal the flaw. (B.t.w. I have no connection to the authors of the above paper, I just read it carefully.)

  40. Joe Polchinski says:

    Just for the record (and my memory is not perfect here, it’s been a long time), I believe that my original comment about trackbacks, which had whatever influence it had, was made publicly on this blog around 2006, and that it was motivated by a discussion, initiated by Peter, about whether Steven Weinberg was senile. It’s true that Peter concluded that he was not, but you chose to publish the original speculation. Note that Weinberg has written at least three books and thirty research papers since that time. Now you’ve since toned down `senile’ to `mania’, `pseudoscience’, but it’s still not scientific discourse.

    You know Peter, you repeat the same rants so often, if the arXiv did link to you they would have to include an arXiv warning “this post has significant text overlap with 20 previous posts by the same author.” 🙂

  41. AReader says:

    Hi Peter,

    I do like reading your blog, because I think it is important we’re looking for alternative theories …

    I was just confused about this bit from you: “I’m a big fan of Bousso-Polchinski, KKLT, and Susskind’s argument for anthropics and the string theory landscape”. I was always under the impression you felt that string theory and multiverse are “wrong” – or have I misunderstood you, and you think they are quite likely to be correct theories?


  42. Peter Woit says:


    I’d been told by a physicist who talked to you a few years ago that you were slandering me, telling him and other people that you didn’t read my blog because I had accused Weinberg of being senile. This physicist told me that he had to explain to you that what you were saying was not true. Your behavior then and now was and is completely unprofessional and you should be ashamed of yourself.

    Some facts, with documentation:

    1. I have never thought that Weinberg is senile or said anything like this, privately or publicly. Instead I have said exactly the opposite. The blog entry in question is this one
    in which I discussed in detail Weinberg’s Living in the Multiverse article, hep-th/0511037.

    The morning that I was writing this, a colleague came into my office and I showed him the Weinberg paper. He reacted with a comment that Weinberg must be senile if he believed this was science. I argued with him that no, unfortunately for physics that wasn’t it, that Weinberg was as sharp as ever, but somehow this was becoming a popular point of view among physicists.

    In the comment section of the blog that afternoon, responding to a comment from Thomas Larsson that “the heroes of your youth have become pathetic old men” I wrote, in full
    “Hi Thomas,

    One of my colleagues this morning, after being shown the Weinberg article, commented that Weinberg must just be senile. Unfortunately I don’t think that’s what’s going on. Weinberg wants to be part of whatever the hot topic in particle theory is, and the landscape is the hot topic these days. It’s being driven mainly by younger people, not by seventy-year-olds, and you can’t put their behavior down to senility.”
    Yes, one of the “younger people” I had specifically in mind was Polchinski.

    I think my meaning in that comment was perfectly clear: I was disagreeing with Thomas and arguing against any attempt to attribute senility to Weinberg.

    2. Polchinski’s argument that trackbacks should not be allowed to my blog was made at this blog posting:
    see this comment
    Neither he nor anyone else at the time mentioned my supposed accusations about Weinberg. He was personally outraged about this blog entry of mine
    which discussed in unflattering terms an article by him and Bousso in Scientific American promoting the multiverse, anthropics, etc. This was one of the first such articles in a major publication by well-known physicists, perhaps the first that made clear the extent of the problem in the physics community. When I saw his comment, I didn’t recognize the text he was complaining about as mine, realized this by googling it. My reaction was that perhaps that was over the top, so in a comment responding to him I apologized for it. Many people then contacted me to tell me they thought I was wrong to do so, that what I had written was perfectly justifiable. Given Polchinski’s later behavior, they were right, and I was wrong.

    One odd thing about this was that Polchinski’s example of what was wrong with my blog was not a discussion of a scientific paper on the arXiv, but of a popular promotional piece in a magazine. It was unclear to me then and remains unclear to me now why he thought that was a good argument against allowing links to discussions of arXiv preprints on my blog.

  43. Peter Woit says:

    No, I don’t think string theory and the multiverse are “wrong”, I think they are “not even wrong”: they don’t predict anything, so can’t be “wrong”. Interpreting the Bousso-Polchinski argument as “if string theory make sense and implies anything, it implies a landscape with just about every possibility”, that’s perfectly in agreement with my point of view. We just disagree about where that argument leads you. I (and I think most physicists) believe it leads you outside the bounds of science, into another sort of activity.

  44. Hamburger says:

    I have been reading Peter’s blog for about a decade now without much active participation in the comments. I am now chiming in in support of him.

    I find the scientific discussions in this blog very valuable with a high signal to noise ratio. Other blogs that I reed are Tommaso Dorigo, Scott Aaronson, Sean Carroll, and Tim Gowers. I feel that all these blogs have similar signal to noise ratio. (But I do not know if these arXiv trackbacks to these blogs.) In any case, exclusion of Peter looks like at best a childish behavior and at worst a corrupt way of behaving at people one disagrees with.

  45. Curious Mayhem says:

    If non-scientific sites like The Atlantic are tracked-back on arXiv, then Not Even Wrong should certainly be tracked-back on arXiv as well.

    The trichotomy of signal/metasignal/noise is germane here. Not Even Wrong is metasignal. It isn’t and doesn’t pretend to be publishing original research. It does follow and comment on original research, in a scientifically informed way that is aware of multiple papers, ideas, lines of research, etc., at once. The idea that it’s noise is nonsense.

  46. Joe Polchinski says:

    Peter, Yes, you repeated an anonymous suggestion that Weinberg is senile and disagreed with it, so you got to put the idea out there without anyone taking responsibility for it. And even the rest, your ridiculous attempts to psychoanalyze Steve Weinberg: the idea that he was just following a hot topic, when he had actually created that topic twenty years earlier, it was the young people who were following him. It’s fine if you want to discuss this on your blog, but I still don’t think it’s arXiv material. The arXiv is for science.

  47. Yatima says:

    The arXiv is for Science.

    No disrespect intended, but I can only read that in the voice of GlaDOS, continually pursuing Science.

    This day is off to a good start.

  48. M says:

    While inventing the multiverse, Weinberg had enough free time for writing a few papers about the physics of our little universe. What a pity that string theory is so deep that younger generations no longer have enough free time.

  49. anon says:

    In case readers of this blog want to see the other side of this argument, let me give some support to Joe in this.

    Though I don’t have any first hand knowledge of trackback policy, I think it’s pretty obvious to everyone that in Peter’s case there was no policy. Someone just decided not to allow trackbacks to Peter’s blog.

    This might seem unfair, but to be honest there is no constitutional right to have your blog linked to the arXiv. Though the arXiv is not quite a peer-reviewed journal, it does by necessity exclude certain kinds of material which might be judged to be unsuitable.

    Peter’s blog is unsuitable for two reasons:

    1) Peter has a controversial and highly negative view of a significant fraction of current theoretical research. Therefore many (though perhaps not all) scientists would be upset to have Peter’s commentary linked to their arXiv preprints. Since the arXiv serves the scientists who upload their papers, I don’t think administrators really had a choice but to exclude links to Peter’s blog.

    2) One might still argue that relevant scientific criticism of a paper should be easily accessible through the paper’s arXiv preprint, whether the authors like it or not. This is where Joe’s point is crucial: Peter’s criticism is usually not relevant to the scientific issues under consideration. A relevant scientific critique must provide new knowledge, in particular one of the following: a) an argument that the analysis of a paper is erroneous or in contradiction with accepted facts, or b), a suggestion of a concrete alternative approach which is arguably more promising. But Peter’s critiques are far more fundamental than this. He just doesn’t think that the technical issues people are grappling with are worthwhile. This is Peter’s scientific judgement and is ultimately his opinion, it is not a technical criticism relevant to any specific paper on the arXiv. The thing is, all scientists come to their own conclusions on the relative merit of various lines of scientific research, and find much of it wanting. No scientist really needs Peter’s opinion on these matters, especially since his views are pretty simple and not extremely deep. I should emphasize that Peter has no scientific argument that string theory does not describe our universe. There are, however, scientific (albeit distressingly theoretical) reasons for believing that it does. However, I think Peter does make a convincing argument that we will probably never know for sure whether string theory is right. I hope this is not the case.

    I like Peter’s blog a lot. It’s fun to read and disagree with. He runs a tight ship, and links a lot of interesting stuff. But I think it plays a role quite distinct from papers that appear on the arXiv, and see no need to connect them.

  50. Peter Woit says:

    Anyone who follows the link I provided can read the 2006 discussion about arXiv trackbacks for themselves, and see that it was not, as you claimed here, about Weinberg. It was about your thin-skinned reaction to my accurate characterization of your Scientific American article as pseudo-science.

    I note that you don’t deny doing what I have been told you have been doing, lying to people in private for years about this issue, for the purpose of attacking my reputation. This is not just sleazy and unprofessional, but actually illegal.

Comments are closed.