Into the Swampland

Last month Cumrun Vafa gave a talk at Stony Brook entitled The Swamp Surrounding the Landscape. Tonight he has a new paper on the arXiv entitled The String Landscape and the Swampland. Vafa appears to be suggesting that, faced with the huge landscape of possible string vacua and the attendant inability of string theory to predict anything about physics, the thing to do is not to abandon string theory, but to head off into the even larger “swampland” of effective field theories that may or may not correspond to string theory vacua. He gives various arguments for why certain effective field theories may not correspond to string theories, but most of these are just something like “the string theory constructions we have looked at so far can’t give this kind of effective field theory”. Since one still doesn’t know what string theory really is, one probably can’t do much better than this. He also assumes that the rank of the cohomology groups of Calabi-Yau threefolds is bounded, which is a conjecture that at least some algebraic geometers don’t believe in.

Throughout his article, Vafa assumes that string theory must be true, asking “how” it will connect to experiment, not “whether” it will. For more than twenty years, string theorists have led particle physics deep into a swamp. It seems peculiar in the extreme that Vafa is now suggesting that, instead of hiking back out of the swamp to dry land, particle theorists should push on deeper into the swampland.

Update: Lubos Motl has a posting about the Vafa paper. It includes the news that Andy Strominger
believes that the program has two basic flaws: the conjectures are trivially correct in every theory of quantum gravity independently of string theory; and moreover they are wrong.

Some of my commenters claim that what Vafa is doing is designed to make string theory falsifiable. I don’t see this, and Vafa doesn’t make this claim himself. This seems to me an example of a common phenomenon. People take a string theory paper that already is going way out on a limb with not very solid arguments, then make a wild extrapolation that goes far beyond what the author claims and use this to promote the importance of the paper in a completely unjustifiable fashion.

To falsify string theory along these lines, one would have to show that it can’t lead to the standard model as an effective low energy theory. Vafa doesn’t claim this is conceivable, and his arguments can’t possibly do this. Most of the examples he gives of effective theories that may not be low energy limits of a string theory are gauge theories of high rank. It’s certainly conceivable that one can argue for something like a bound on the rank of the effective field theory gauge group if it comes from string theory, but there’s absolutely no reason to believe that such an argument can rule out the rank 4 case we care about (SU(3)xSU(2)xU(1)). I suppose one can argue that, if say Vafa can show the rank must be less than 500, and the LHC discovers a new gauge theory sector with rank 501, string theory would be falsified. But that’s kind of like saying that string theory is falsifiable, because if dragons emerge from the LHC interaction regions, string theory would be wrong.

Update: Jacques Distler also has a posting about the Vafa paper. He says he’ll wager that it is “far, far from true” that “‘anything’ is realizable somewhere on the Landscape”, and that “we will learn much” if we investigate this swamp. He doesn’t explain why it’s a good idea for the particle theory community to enter this swamp to investigate it carefully.

This entry was posted in Uncategorized. Bookmark the permalink.

26 Responses to Into the Swampland

  1. A.J. says:

    What Vafa is suggesting is utterly reasonable. Even if you are string theory true believer, you can at least admit that studying models which are at least very similar to string theory is a good way to learn more about string theory.

  2. Anonymous says:

    Don’t you think you’re being unreasonable? You press for string theory to be good science, and argue that it isn’t falsifiable. Someone outlines a program that could *make* it falsifiable — granted, a difficult program to carry out — and you complain. Would you be satisfied by anything less than Vafa completely renouncing string theory?

  3. Dear Peter,

    the fact that some people don’t believe that the number of Calabi-Yau topologies is finite does not mean that the physicists are not allowed to investigate physics with the opposite assumption that is consistent with all known facts.

    If one felt pressure not to investigate a scientific theory (or conjecture) just because there exists someone who does not like it, we could also stop doing physics or string theory because some people at Columbia think that physics beyond the Standard Model is not even wrong.

    Such a thing will certainly play no role whatsoever for me. I leave this method of deciding about science to the numerous crackpots who like to visit your blog assuming that it is a serious reading.

    You should be pleased if people start to study swampland – it is a step towards the research of f**kland. I explain on my blog why this should really please you. 😉

    All the best

  4. Wolfgang says:

    It seems that the main argument of this paper is that the “landscape” of truly consistent vacua is of zero measure compared to the “swampland” of models only seemingly consistent as semiclassical effective theories. In this sense (far away from the Planck scale) M-theory is predictive and I am waiting for ref [8].

  5. Arun says:

    There is a point to this – it does open a way to falsification of string theory. If one understands the types of effective field theories string theory cannot produce, and shows that e.g., our universe is governed by an effective field theory that is not stringy, then we can move string theory entirely to the math. department.

  6. woit says:

    I added something to the posting about the falsifiability issue.

    A.J.: I don’t understand your argument why this is utterly reasonable. The kind of models Vafa is talking about are atrociously complicated, far more so than the ones in the landscape, (thus the “swamp”). Why is it reasonable to spend time looking at these things? They’re hideously ugly and have nothing to do with the Standard Model or the real world. The only justification for entering this yucky swamp is the belief that understanding just how yucky string theory can be is important, given that we already know that it can be very, very yucky. I don’t understand why anyone thinks this is a sensible way to spend one’s time.

    Anonymous: If string theorists like Vafa want to say that string theory is too poorly understood to believe in the landscape, so they intend to keep working on it, that is not unreasonable, and such work sometimes even leads to interesting things (e.g. work on topological strings). But once they accept that the landscape exists and string theory is a radically non-predictive framework, if they are honest scientists, they should admit that the idea should be abandoned.

    Lubos: people are welcome to investigate whatever assumptions they want, I was just pointing out that this particular assumption is not one where the experts on the subject (algebraic geometers) are agreed about what is likely to be true. And as to the crackpot level of readers of my blog, have you taken a look a look at your own comment section recently?

    Wolfgang: What’s a specific experimental prediction of M-theory that comes from Vafa’s work?

  7. Wolfgang says:


    as I understand it, Vafa’s paper outlines a research program on how to make string theory predictive far away from the Planck scale. It argues that there should be a large number of field theories which are not consistent with M-theory and thus should not be observed in nature. The “swamp” of such models should be very large compared to the “landscape” of consistent vacua.
    This should/could provide for testable predictions at some point.
    The paper does not yet provide examples of such models;
    I expect subsequent papers (e.g. ref[8] ) to be more specific.

  8. woit says:

    Hi Wolfgang,

    I don’t see anything in Vafa’s paper that comes even close to giving a plausible way of coming up with a legitmate prediction from M-theory. He’s assuming the existence of the landscape, and once you do this there seems to be no way to predict anything. The question of how big the swamp is seems irrelevant to making real predictions. I don’t see why if he had a serious idea about how to predict something, he wouldn’t put it in his first announcement, keeping it secret until the later paper.

  9. A.J. says:

    Hi Peter,

    All I was attempting to say — and I hope you’ll forgive the vagueness; I spent most of yesterday typing — was that I think studying a far wider class of models is a good idea. I don’t care whether people do it with the intent of learning more about string theory, or with the intent of showing that the Standard Model actually lies somewhere in the swamp. The bottom line is that (I’m speaking here as a mathematician) we don’t really understand that much about any of these theories. Having a broader selection of examples might teach us a few things.

  10. Brett says:

    I think it’s a perfectly reasonable idea that there should be vastly more field theories that cannot exist as string theories than those which actually could arise as string theory vacua, say. However, given the complexity of either kind of theory, I would expect that that boundary between the two in parameter space is probably not going to be anything simple; it will likely have a fractal struture, and it is my guess that any randomly selected theory, even if it exluded as a string theory, will nonetheless be “close” to an string theory version.

  11. Renormalized says:

    I sometimes wonder if Lubos Motl is working on any publishable material or has any original ideals about physics. His real talent is in making useless arguments about crackpots and those less intelligent than himself. This includes the whole world since he thinks of himself as the smartest of all. It is easy to insult others work and ideas when you have none of your own to be criticized. His pompous imperturbable egoism is as worthless as his contributions to science.

  12. Not a Nobel Laureate says:

    You’re all wrong.

    The Theory of Intelligent Falling is the Answer.

    Evangelical Scientists Refute Gravity With New ‘Intelligent Falling’ Theory

    August 17, 2005 | Issue 41•33

    “Closed-minded gravitists cannot find a way to make Einstein’s general relativity match up with the subatomic quantum world,” said Dr. Ellen Carson, a leading Intelligent Falling expert known for her work with the Kansan Youth Ministry. “They’ve been trying to do it for the better part of a century now, and despite all their empirical observation and carefully compiled data, they still don’t know how.”

  13. The proposed theoretical activities in infinite-dimensional infinite-volumed swampland unavoidably create an association to monkey, typewriter, and Shakespeare’s sonnet. Another association is a story in Mathematical Intelligence about Baron Monty Carlow and his ingenioius method of measuring the area of lake by randomly bombing the county and counting the fraction of bombs hitting the lake.

    The pioneers of Monte Carlo method of TOE building
    deserve all encouragement in their challenging task,

    Matti Pitkanen

  14. Arun says:

    If our universe is indeed hidden somewhere in the landscape, what is the rational reason for abandoning string theory? Nature could be both subtle and malicious.

  15. Aaron Bergman says:

    Some of us continue to hope otherwise, though.

  16. woit says:


    Before you start believing in something completely absurd like the landscape, how about first asking for some experimental evidence? I can’t believe people are abandoning the most elementary aspects of the scientific method. This is something we are teaching to children in elementary school these days, why can’t professional physicists understand this?

    A less elementary aspect of the scientific method is that you can always try and make a wrong idea work by making it more and more complicated, desperately trying to avoid confrontation of the idea with experiment. It should be completely obvious by now that this is what is happening with string theory.

  17. Wolfgang says:


    Jacques Distler conjectures on his blog that only models with 3 fermion generations are consistent (in Vafa’s sense) but not 1 or 2.
    Would you consider this a valuable post-diction ?
    (It would have been really nice as a prediction, but sometimes the calculations just take a looong time.)

  18. rrtucci says:

    Peter, just had an idea for “the special-edition cover” for your book: a picture of Lubos’ head (wearing a harvuhd tie), the head superimposed on the body of a tiny ant, which is at the focal point of a large magnifying glass, which in turn is being illuminated by the sun. Maybe also a smarter ant, with Witten’s head, running away. And an ant with Vafa’s head caught in a swamp.

  19. woit says:


    No, this is not a prediction or post-diction, since you have no idea of what the masses of the particles are. Remember, experimentally all we know is the number of light generations (more specifically, the number of generations with a neutrino less than half the mass of the Z). It’s entirely possible that the number of generations is really >3. If we had only observed 1 or 2 generations, and string theory really needed at least 3 generations, we wouldn’t falsify string theory because more generations could just be at higher mass.

    In general, a “prediction” of the number of generations is operationally meaningless without also knowing at what mass scales they are going to occur.

    I’m no expert on these constructions, so I’m not sure which difficulties in constructing models with one or two generations Jacques is referring to. My guess is that it is the fact that in the standard way of using Calabi-Yaus, the number of generations is the Euler characteristic, and it is hard to construct Calabi-Yaus of Euler characteristic one or two. But I suspect there are many ways around this, and that an expert in these constructions could come up with one way or another of getting one or two generations if they really wanted to. Again, since you don’t know the masses, one or two light generations, the rest heavier than we can observe, is no problem.

  20. Arun says:


    Why don’t I ask for experimental evidence? Because I know there is none. But if we’re into dealing with reality, then it is a given that Vafa, Motl, Douglas, etc., are not going to abandon string theory any time soon, or resign from the physics department. Then, all I can ask for is that they pursue the approach that is most likely to produce an experimental signature for string theory. Is this approach the right one? I don’t know, but until the string theorists explore it a bit more, we won’t know.


  21. woit says:


    I think by now it’s completely clear that the things that are known about string theory are insufficient to make any contact with physics and predict anything. Under these circumstances, it’s not unreasonable to try and come up with some better understanding of string theory, e.g. to work on trying to figure out what M-theory is. It is unreasonable to deny that this is the situation and promote the idea that you can get physics out of the Landscape.

  22. Shantanu says:

    Peter have you read hep-th/0509157 ? any comments on that?
    It has been discussed at

  23. woit says:


    Yes, I saw that. I was thinking of putting together a posting very soon of various links including that one. I’m not sure what one can sensibly say about it though. Do you see anything in that paper that looks at all like a promising idea about how to connect string theory to real world particle physics? I didn’t.

  24. Pingback: Not Even Wrong » Blog Archive » Mucking About in the Swampland

  25. Not a Nobel Laureate says:

    Looks like Vafa et al have some serious competators . . .

    Two female gorillas have been photographed using sticks to get through the swampland . . .

  26. Pingback: Not Even Wrong » Blog Archive » Loops ‘05, Again

Comments are closed.