Susskind NYT Book Review

There’s a review of Susskind’s book The Cosmic Landscape in this Sunday’s New York Times book review section. The reviewer does a reasonably good job of laying out what the Landscape controversy is about, characterizing Susskind’s attitude as “braggadocio” with “an air of smugness”, and noting that “He allows remarkably little doubt about string theory considering that it has, as yet, not a whit of observational support.”

This week’s Village Voice has a profile of Susskind.

Update: More about this over at Uncertain Principles.

This entry was posted in Uncategorized. Bookmark the permalink.

62 Responses to Susskind NYT Book Review

  1. Chris Oakley says:

    “As you know, there’s something of a war against science going on,” Susskind says.

    He didn’t say which side he’s on.

  2. Lubos Motl says:

    I also like the review.

  3. D R Lunsford says:

    What did it say, other than more mindless parroting of hyperbolic claims, followed by a “discussion of the issues” admixed with (anti)hero-worship? It’s Oprah-level dilettantisme. Can’t the Times find anyone with sense of history to put things in context?

    -drl

  4. Thomas says:

    Love the phrase from the review:

    “yields a gargantuan number of models: about 10^500, give or take a few trillion.”

  5. arnold says:

    I still have to read the book, but the impression is that the review is very fair and well-written.

    I fear people who believe to have the truth…without any evidence.

    What strikes me the most is that clearly the idea of Landscape is on the same level as the idea of Intelligent Design: there is no evidence for none of the two: you can believe them or not.
    But some people like to say that the former is scientific and the other is not.

  6. Thomas Larsson says:

    The next-to-previous comment was not by me.

  7. mathjunkie says:

    Have read the review.

    Is it true that ID=anthropic principle, and string theory is an alternative to ID?

  8. D R Lunsford says:

    I wonder if Einstein would wear a Susskind t-shirt?

    He would probably wear a Susskind t-shirt if on it, Susskind were wearing an Einstein t-shirt…

    -drl

  9. Juan R. says:

    New piece of nonsense.

    .

    Question: When stringers will take the next logical step; the multi-multiverse?

    Multi-multiverse: Existence unit formed by existence regions called multiverses. In ONE multiverse, string theory is correct in others multiverses string theory is just wrong. Since number of multiverses is greater that number of universes in a single multiverse, the number of universes where string theory is incorrect is far greater that other case.

    Therefore, the interest of string theory for understanding more deeply the ‘nature of reality’ is proportional to epsilon when epsilon tends to universe.

    Sorry! 🙂

    Juan R.

    Center for CANONICAL |SCIENCE)

  10. island says:

    arnold Says:

    I still have to read the book, but the impression is that the review is very fair and well-written.

    I fear people who believe to have the truth
without any evidence.

    What strikes me the most is that clearly the idea of Landscape is on the same level as the idea of Intelligent Design: there is no evidence for none of the two: you can believe them or not.
    But some people like to say that the former is scientific and the other is not.

    This all changes if by some miracle the landscape provides string theory with the means to the ToE, right?

    I mean that we wouldn’t need direct evidence nor falsifiability if this were proven to be the case.

    Right?

  11. john baez says:

    mathjunkie writes:

    Is it true that ID=anthropic principle, and string theory is an alternative to ID?

    No, it’s not true that intelligent design is the same as the anthropic principle!

    Intelligent design is the idea that our universe was designed by some intelligent being. The anthropic principle is the idea that some aspects of our universe can be understood starting from the fact that intelligent life exists here.

    (In fact there are many versions of intelligent design and the anthropic principle, but let’s not get into that here – follow the links if you want more details.)

    It’s possible for intelligent design and the anthropic principle to
    both be true, both be false, or one be true and the other be false.

    For example, God could have intelligently designed our universe by running a machine that builds all possible universes – this would be intelligent because then God could avoid bothering with the details. Then our universe would be one of these with intelligent life – and it’s possible we could understand some aspect of our universe starting from this fact.

    String theory is not an alternative to intelligent design. In fact, all 8 combinations of intelligent design, the anthropic principle and string theory being true or false are possible.

    However, the astute reader will note that this discussion resembles medieval theology more than physics.

  12. mathjunkie says:

    Thank you John for answering my question.

    It seems I know next to nothing about the differences among intelligent design, anthropic principle and string theory.

    I think you are the researcher who favours LGQ more than the string theory, right?

  13. Santo D'Agostino says:

    Dear mathjunkie,

    The nature of the intelligent design (ID) idea can be understood by analyzing the structure of their argument, without being distracted by all the details. It is:

    1. There are problems with the theory of evolution.

    2. Therefore God exists.

    ID is therefore seen NOT as a scientific theory, but rather as an argument for the existence of God, in the same spirit as the arguments for the existence of God proposed by certain medieval philosophers and theologians. ID is just as fallacious; one can prove a mathematical theorem, but proofs of the existence of God are not possible.

    Every scientific theory has problems; it is not the way of science to throw up our hands in the face of problems and say, “OK. I give up. God must be the only explanation.” The scientific approach is to analyze as clearly as possible what the problems are, and then to either revise the theory or construct a new one that is better.

    Responding to island and arnold, it is interesting to compare and contrast the spirit of thought of the IDers with some of the popularizers of string theory. As argued by many people, both on this site and elsewhere, there is a clear difference between ID and string theory. The former clearly has no chance to ever be science, whereas although the latter is not yet a scientific theory, its proponents are striving to construct what one day might become a scientific theory. That is, a theory that makes specific predictions that can be confronted with observational or experimental data.

    All the best,
    Santo D’Agostino

  14. Steve Myers says:

    On anthropic principle: since any model has to be general to be useful there will be a large — possibly infinite — number of particular solutions. So can you work backwards from particulars to get a model? Not without some leap or guess or postulating a rule outside the set of particulars. You begin with the fact that Lotte has blue eyes and end with ” there is at least one such and such, such that it possess some property.” See Russell’s “Inquiry into Meaning and Truth.” Every math student has to learn that before he can understand what constitutes a proof. Look, you can always find some function (in fact, there are an infinite number) to fit any data. That’s why the curve you choose is based on data predicted. Of course, you run into the same problem again but that’s why science is ongoing, not a finished project. But isn’t all that obvious?

  15. Urs says:

    are striving to construct what one day might become a scientific theory. That is, a theory that makes specific predictions that can be confronted with observational or experimental data.

    I still don’t think this is the right way to put it. I’d say “people are striving to construct a phenomenologically successful theory”. There is nothing unscientific about string theory. It’s “just” phenomenologically unsuccessful. This may be reason enough to abandon it. But let’s keep the facts straight.

    Consider a cosmologist who finds himself living in a universe which on large scales is not homogenoeus but highly chaotic. By pure thought, he has come up with the theory of general relativity. Now he is trying to see if this theory is phenomenologically viable as a cosmological theory on large scales.

    So he searches for solutions of his theory and tries to match them with the observed data.

    First he tries FRW cosmologies. Most fail miserably. One, the mixmaster scenario comes closer in capturing the chaos he observes, but is still way too symmetric.

    So he constructs cosmological models which are inhomogenous, depending on one, two, three, then a dozen parameters. All of them being exact solutions of GR, all of them coming a little bit closer to the cosmos he actually observes, but none being quite right.

    His fellow bloggers challange him. They say “What an unscientific theory you have. Surely you must believe in alchemy and poltergeists if you also believe in your theory.”

    But his theory is GR. He just happens to live in a cosmos which is not described by a simple (highly symmetric, low-parameter) solution of GR, but of a highly complex one.

    So for him, GR as a cosmological theory is highly unsuccesful, phenomenologically. But it is not unscientific. He could make predictions (say of CMB spectra, of nucleosynthesis, etc.) if only he knew which solution to GR describes his cosmos. But he doesn’t know.

  16. michaeld says:

    Santo D’Agostino, exactly. It seems that many religious people seem to believe that the sum value of science and religion is a fixed constant, and so every failure of science is a success for religion and vice versa.

  17. island says:

    mathjunkie says:
    Is it true that ID=anthropic principle…

    Some interpretations of the anthropic principle indicate that we are not here by accident, and IDists erroneously latch onto this as evidence for god’s existence.

    The Privileged Planet… by Guillermo Gonzalez and Jay Richards, is once such derivation of this sort.

    While it is not correct to conclude that just because we’re not here by accident there must be an intelligent agent involved, this is exactly what IDists do, and unfortunately… skeptics are conditioned to buy straight into this hype, and so their automatic response is to try to find ways to lose the implied “specialness”… rather than to look for some good physical reason for why it might be true.

    string theory is an alternative to ID?

    Stringy theories that use multiverse rationale can lose any implied anthropic specialness in an infinite number of possible universes, since one of them had to be like ours, but there is no “illusion” of intelligent design without an unfounded leap of faith, regardless of what Leonard Susskind would have us believe.

    It’s not a matter of either/or, as Lenny claims… either we accept the multiverse scenario, or… “without any explanation of nature’s fine-tunings we will be hard pressed to answer the ID critics.” since “the appearance of intelligent design is undeniable”.

    There is no implication for intelligent design, because there is no inferrence of intelligent design from evidence for purpose in nature, without direct proof. Evidence for goal oriented design or structuring in nature does not constitute evidence for intelligent design if there is no landscape.

    Lenny also doesn’t appear to know what theories are valid and which are not when it comes to origins science, because “multiverse rationale” isn’t a valid argument against fine-tuning unless and until a multiverse is proven to exist, or if multiverse “reasoning” proves to be necessary to the one true theory of everything. I think… that very last part is correct… but nobody answered my question.

  18. island says:

    FYI: Santo D’Agostino and michaeld are not correct as the anthropic princple applies to this debate. ID doesn’t not require the faliure of science to abuse the anthropic principle as evidence for intelligent design.

  19. Santo D'Agostino says:

    Hi Urs,

    Thanks for your comments. Your analogy has given me a very helpful perspective on string theory.

    My inclination is to adopt fairly strict criteria for what we call a scientific theory. For instance, it should at least be internally consistent and it should make quite definite numerical predictions that can be tested by observations or experiments. Ideally it will predict new phenomena, but that is not strictly necessary.

    It seems that based on some such strict criteria, string theory is not yet a scientific theory, although I admit that my criteria may be too strict, based on your nice analogy.

    Another analogy: suppose we have a candidate theory of the hydrogen atom, let’s call it theory A. In theory A there are a million different ground states for the hydrogen atom, no way to choose among them, and there is no way to predict any of the excited states. So nothing can be calculated about the hydrogen atom, no transition rates, no frequencies of emitted or absorbed radiation, etc. I would not call theory A a scientific theory, but I would admit that the effort of trying to shape it into a theory may qualify as a scientific activity.

    Contrast this with another theory of the hydrogen atom, let’s call it theory B. In theory B we have millions of ground states too, and no way to choose among them, but this time no matter which ground state is chosen, the theory predicts the relative energies of the corresonding excited states. In theory B at least you can predict the frequencies of emitted or absorbed radiation, and so I would be willing to call theory B a scientific theory. (The frequencies are always the same, no matter which ground state is chosen, so we still have no way of knowing what the actual ground state is in theory B.)

    My impression is that string theory is more like theory A, and so I would hesitate to call it a scientific theory yet. If I am wrong, I would value any further clarifying comments that you might like to make. For instance, are there any specific numerical predictions that string theory makes that can be confronted with observations or experiments?

    Santo

  20. Santo D'Agostino says:

    Hi island,

    I’m not sure why you say

    “FYI: Santo D’Agostino and michaeld are not correct as the anthropic princple applies to this debate. ID doesn’t not require the faliure of science to abuse the anthropic principle as evidence for intelligent design.”

    since I said nothing about the anthropic principle.

    I agree with your criticism of Susskind’s reasoning; as I already mentioned, it is unscientific to argue for the existence of God based on one’s inability to cook up a satisfactory scientific theory.

    As for your question,

    “This all changes if by some miracle the landscape provides string theory with the means to the ToE, right?

    I mean that we wouldn’t need direct evidence nor falsifiability if this were proven to be the case.

    Right?”

    How would you know if such a proposed ToE were a good theory? You would have to compare its predictions to observed and experimental data; in other words, you would need evidence and falsifiability.

    All the best,
    Santo

  21. Urs says:

    Santo D’Agostino

    it is a pleasure to be able to have a reasonable discussion.

    I would suggest to slightly modify the analogy which you proposed.

    The hydrogen atom itself is a solution of a theory. Quantum mechanics itself does not predict anything about atoms. You need to feed in the correct initial data first.

    Once you specify that your electron sees precisely the Coulomb potential, for some value of the one parameter describing it, you can derive atomic spectra using QM+choice of potential. You have to choose the form of the potential (~ a 1/r) as well as pick one value of the parameter a.

    You fix this by fitting it to a small subset of experimental data. Then you use the fact that given that input, QM predicts the rest of the spectrum.

    So, if people had discoverd the formalism of QM before having figured out that the H-atom consists of an electron orbiting a nucleus, they could not have used it to predict anything about the spectrum of that atom.

  22. Juan R. says:

    I said,

    Therefore, the interest of string theory for understanding more deeply the ‘nature of reality’ is proportional to epsilon when epsilon tends to universe.

    It would say

    Therefore, the interest of string theory for understanding more deeply the ‘nature of reality’ is proportional to epsilon when epsilon tends to zero.

    –

    Juan R.

    Center for CANONICAL |SCIENCE)

  23. island says:

    The nature of the intelligent design (ID) idea can be understood by analyzing the structure of their argument, without being distracted by all the details. It is:

    1. There are problems with the theory of evolution.

    2. Therefore God exists.

    No, that’s not the nature of cosmological ID, which is what Lenny’s book is about.

    How would you know if such a proposed ToE were a good theory? You would have to compare its predictions to observed and experimental data; in other words, you would need evidence and falsifiability

    I was asking about falsifiability of the “Landscape”, not testable predictions of a ToE.

  24. Santo D'Agostino says:

    Hi Urs,

    Thanks for explaining to me how my hydrogen atom examples are like your cosmology example for the purposes of this discussion.

    I suppose that if the formalism of QM had been discovered before it could predict anything specific, I would also say that it were not yet a scientific theory, but a formalism striving to become one. But as you said earlier, this does not mean it is not worth working on.

    All the best,
    Santo

  25. Santo D'Agostino says:

    Hi island,

    I misunderstood you. I assumed you were talking about ID, not cosmological ID. I don’t know exactly what the latter means, since I have not read Susskind’s book.

    All the best,
    Santo

  26. Adrian H. says:

    Urs and Santo

    I think the thing that is missing between you is the concept of explanation, as against prediction. Urs, you have pressed the idea that ‘string theory is science’ very succesfully, using the idea that it *could* make predictions. You stress the idea that prediction is often a matter of finding correct solutions given some initial conditions. As in GR, put in the values for mass-energy distribution, get out a model of space-time structure.

    But I think this analogy misses something important about scientific theories that is not equivalent to prediction: scientific theories are explanatory.

    Take a bad theory that is maximally predictive in your terms: one that is of the form A & -A for any A. A contradiction in other words. From that contradiction one can predict anything and everything, since any claim logically follows. But even though it will be predictively a TOE, it is obviously not a scientic theory, or if you wish to stretch the term that far, it is a scientific theory, it’s just a very bad one. A makes an infinite number of false predictions along with the true ones.

    Finding a solution to the equations of GR is predictive of the structure of space time (if we found we live in the Goedel Universe, that would be an interesting fact!) but it does not explain why the mass energy distribution is the way it is, or why the speed of light has just that value, etc. Finding solutions to equations is not all, or even the main part of, the scientific enterprise.

    I think the worry that is being expressed about string theory and the landscape is that—and this is the thing that pulls it closer to ID than is really comfortable—the theory is incredibly low in even potential explanatory power. “Why are things this way, rather than some other.” If the answer is that we can get our universe out (I’m being optimistic) as a solution to some set of equations, by putting in by hand a whole lot of free parameters then it is of diminished interest. Not no interest at all, just diminished.

    And on the subject of explanation vs prediction, note that ID is a perfectly predictive “theory”: it predicts the life forms we see around us. We just have to put in by hand a whole lot of free parameters that describe God’s intentions!

    But ID is obviously low in explanatory power. And String theory may be closer to ID than it is to GR. That’s the worry.

  27. Adrian H. says:

    I should add that I’m not saying that everything can be explained, just that explanatory power is something that we seek to maximize in scientific explanations.

    And perhaps we should really say that ID is retrodictive, rather than predictive. Likewise the Anthropic Principle looks like a case of retrodiction: People exist therefore this universe must be one where people can exist!

  28. Urs, you seem relieved. But I don’t think many people on this blog think string theory per se is not a scientific theory. In particular I’ve never seen Peter make such a claim. If I understood right Peter’s mind, what he says is : 1) anthropic attemps to go out of the landscape are unscientific 2) more and more string theorists argue that unfalsifiability is not such a bad thing after all, and this is a sign that string theory has failed to achieve its primitive goals and that some people are getting nervous about this 3) it’s time to try something else.
    With all this I fully agree.
    Peter, please correct me if I’m misinterpreting your mind.

  29. Urs says:

    Adrian,

    thanks for the comment.

    It is hard to give a precise meaning to the term “explanation”. Some people feel quantum mechanics explains the spectra of atoms, while others are so dissatisfied with the interpretational issues of QM that they would only acknowledge that it successfully describes these observations, with a “real explanation” still to be found.

    But besides personal feelings it does not make much of a difference which standpoint one takes.

    So I would not want to engage in a discussion whether some theory “explains” enough to be worth of our considerations. A good theory is one that yields a high number of (maybe approximately) correct descriptions from a low number of assumptions.

    Consider the following scenario.

    Assume perturbative string theory is indeed a good description of physics beyond the standard model. Assume there are heterotic strings which are weakly coupled and spacetime is really compactified on some Calabi-Yau.

    Say the currently available modles by Braun, Ovrut and others can be further refined such that one day they find a choice of Calabi-Yau, a choice of gauge bundle over that CY and a choice of “Wilson lines” (these are the sorts of parameters one has to choose) such that the low energy effective theory obtained by strings on such a compactification produces the standard model on the nose, without any extra moduli. Let me call the string vacuum thus defined SV1. Our first candidate solution of string theory which matches observed low-energy physics.

    If this happened, one would want to study the postulate that the world is indeed described by SV1 in weakly coupled heterotic string theory.

    This would yield quite a number of predictions. For one, it would predict that at energy scales which resolve the size of the compact CY accelerators would see signatures of these extra dimensions, and the precise topology of these extra dimensions would be predicted.

    Note that this prediction might turn out to be wrong. You may even feel that it has no chance of turning out to be right. But it is a prediction.

    If it should turn out to be correct, we’d have a nice theory of physics beyond the standard model. If not, we’d need to work harder. That’s how science works.

    Indeed, there are serious reasons to expect that this scenario has little chance of being true. One might suspect that a weakly coupled perturbative description of high energy physics is likely to miss essential physics. That a non-perturbative description will be necessary, one that does not need to expand about a fixed background.

    This may be true, and people are working on it. If anyone has a good idea, it will be worth considering. If it helps to reassure people distgusted by the state of affairs in string theory, I might point out that I know a couple of string theorists who took a very close look at the recent results claimed in CDT. I wish I would not have to emphasize this.

    Anyway, I would like to point out that even if we had a nonperturbative definition of a theory beyond the standard model, we’d still face the problem of identifying the correct solution of this theory which describes our world!

    There is no reason to expect that, once spin foam models, CDT, AdS/CFT-like dualities or any other nonperturbative approach is well enough understood that one can talk about its space of solutions, it will turn out to provide a unique one. I expect the opposite to be true. There will be infinitely many solutions.

    So even with a working nonperturbative theory beyond the standard model we’d face the the issue that predictions cannot be made before some parameters in its solution space are fixed. Fixed somehow.

  30. D R Lunsford says:

    Peter – you handled that very well over on the blog “Uncertain Principles”. (Is there a rule that blogs have to carry cute names?) The problem is, correct arguments seem to have lost their value. I was looking at the postings to a right-winger’s blog the other day – e.g. one guy offers a detailed, fact by fact, unemotional indictment of the Bush admin, and it just didn’t make any difference to the dittoheads in the audience – they simply refused to allow themselves to be tied to facts.

    Somehow the art of actually paying attention to an argument has gone missing

    -drl.

  31. Thomas Larsson says:

    Say the currently available modles by Braun, Ovrut and others can be further refined such that one day they find a choice of Calabi-Yau, a choice of gauge bundle over that CY and a choice of “Wilson lines” (these are the sorts of parameters one has to choose) such that the low energy effective theory obtained by strings on such a compactification produces the standard model on the nose, without any extra moduli.

    This could happen, of course, but I think everyone around here agrees that the odds are so small that they can safely be ignored. If so many bright people haven’t gotten any quantitative predictions out of string theory in 20 years, why should things change now? Haven’t people turned to the Landscape precisely because they have given up hope that any useful prediction will ever come out of string theory?

    Moreover, one must not forget that all natural predictions from string theory, like SUSY, extra-dimensions, 496 gauge bosons, new long-range forces, etc., are in apparent disagreement with experiments. A particularly impressive class of experiments seems to be the search for permanent electric dipole moment, which has been blogged about here, here, here, and here. It may seem strange to use experimental results to argue about theoretical physics, but I strongly feel that the by far simplest explanation for this apparent disagreement with observation is that it is due to a factual disagreement with observation.

    Besides, didn’t we agree that the correct theory of QG in 4D must allow for diff anomalies, as I understand section 6.1 of Nicolai et al.? If string theory does not allow for such anomalies, how could it possibly be right? Even if you dislike my treatment of anomalies, you cannot reasonably deny that the absense of 4D diff anomalies is fatal?

  32. woit says:

    Fabien,

    Yes, I’d basically agree with your summary.

    Fabien and Urs,

    I’m careful not to say things like “string theory is not science”. Some things string theorists are doing are science, but some are not. Very specifically what I claim is not science is when string theorists like Susskind say they accept that string theory leads to 10^500 vacua, but then are unable to come up with a remotely plausible idea of how they are going to use the theory to make falsifiable predictions in this case. The only kind of “predictions” he talks about are the anthropic ones that are not falsifiable and aren’t really scientific.

    Lots of string theorists now seem to be saying that, while they reject the anthropic arguments, even if string theory does have 10^500 vacua, that is no reason to give up on it. My point of view is that anybody who wants to claim this has to come up with a plausible scenario for how this is ever going to lead to predictions. From everything I’ve seen, no one can do this, so it appears to me they are just defending the indefensible and not doing science.

    The problem with Urs’s SV1 scenario is that it is exactly what people have been looking for for more than 20 years and not finding. Braun, Ovrut et. al. aren’t able to fix the moduli problem without doing something like KKLT which leads to the 10^500 possibilities, and likely complete loss of predictivity. Sure if this problem magically goes away, you’ll have something predictive and be doing science, but right now this looks like pure wishful thinking to me, and wishful thinking that deadly problems will just disappear is not how science is supposed to be done.

  33. Urs says:

    Peter,

    I tried to point out how a theory generically has infinitely many solutions. So it’s not clear to me why it should be a problem if some theory has only finitely many. And this holds for strings only after some conditions have been imposed. The space of all possible string vacua (not restricting to flux CY compactifications) is still infinite, as far as I am aware.

    I also tried to point out how, generically, from the space of all solutions one has to pick one by matching it to a subset of experimental data. Only then are predictions possible. I tried to give several well-understood examples for this.

    Some people are trying to go beyond what can usually be done. They don’t want to just find the right solution, but also understand “why” it is the one we observe. As with all “why” questions in science, trying to answer them tends to lead into non-scientific territory.

  34. woit says:

    Urs,

    I’m repeating myself, but again, here’s the main issue: if you can’t predict anything you aren’t doing science. If it is true that all these flux vacua really are legitimate string theory solutions, and one can get almost any value of the standard model parameters out of this, how are you going to predict anything? I’m still claiming that there is no plausible scenario in which this line of research leads to a testable scientific prediction. If you know of one, let’s hear it. Saying that we’ll be able to fix the Calabi-Yau, then make predictions isn’t plausible. There’s no reason to believe you will be able to fix the Calabi-Yau, to show that one works the (possibly infinite set of) others don’t. And “predictions” of Planck-scale stringy effects are dubious for both practical (you can’t ever do such experiments) and theoretical (perturbation theory isn’t good enough) reasons.

    I actually think it is only the people asking “why” questions who are still doing science, since it is not inconceivable that if you could answer the question of “why” a certain vacuum state, then you could make predictions. The people I have a problem with are those like Susskind, who claim that even if you never understand “why” a certain vacuum, you can still do science in this particular context.

  35. Urs says:

    I’m repeating myself

    True. And I would have to, too, if I tried to reply. So I’ll leave it at that.

  36. Who says:

    Urs 18 January 4:13 AM
    “There is no reason to expect that, once spin foam models…is well enough understood that one can talk about its space of solutions, it will turn out to provide a unique one. I expect the opposite to be true. There will be infinitely many solutions.”

    Having a space of solutions which is controlled by a few fundamental constants like G and h-bar is qualitatively different from having a huge out of control space of solutions, so we ought to make the difference clear.

    Urs, I disagree with what you say as applied to the particular case of the spinfoam model of http://arxiv.org/hep-th/0512113
    In that paper Freidel and Livine work out a model of 3D spacetime and matter which has QFT as the zero-gravity limit (as G -> 0) and General Relativity with some quantum corrections as the (semi)classical limit (as h-bar -> 0).

    In that paper I do not see other constants which can vary to give a large space of solutions. However they still have to extend their results to 4D spacetime and matter.

    Still I think there IS reason to expect that when and if the Freidel Livine result is extended to 4D it WILL turn out to provide a small space of solutions controlled by a small set of parameters—like a few fundamental constants.

    Perhaps I am just more of an optimist than you, Urs 🙂

  37. Tony Smith says:

    Urs said “… Say the … modles … can be … refined such that … the low energy effective theory … produces the standard model on the nose, without any extra moduli. Let me call the string vacuum thus defined SV1. …”.

    Peter said “… The problem with Urs’s SV1 scenario is that it is exactly what people have been looking for for more than 20 years and not finding. Braun, Ovrut et. al. aren’t able to fix the moduli problem without doing something like KKLT which leads to the 10^500 possibilities, and likely complete loss of predictivity. …”.

    Urs then said “… from the space of all solutions one has to pick one by matching it to a subset of experimental data. Only then are predictions possible …”.

    It seems to me that Urs’s reply does not deal with what I see as Peter’s main assertion:
    Even with many very smart people working for over 20 years, nobody has yet produced a concrete conventional superstring model that
    1 – has no obvious fundamental flaws (moduli problem etc)
    and
    2 – agrees quantitatively with a significant subset of the Standard Model parameters (say, for example, ANY ONE of the following: the fine structure constant; or the weak boson mass scale; or some fermion masses; or the relative strength of the color force; etc).

    Only such a concrete conventional superstring model could qualify for Urs’s SV1.
    IF it were to be found, THEN Urs could say that it matched “a subset of experimental data” and therefore use it to make “predictions” a la SV1.

    However, since there 20 years of work by many very smart people has failed to find such an SV1, it seems reaonable to me to conclude:
    that it is unlikely that further work along the same lines (conventional superstring theory) will ever find such an SV1,
    and
    that other approaches should be tried.

    As to what kind of “other approaches”, consider the remark of Pierre Ramond in hep-th/0112261 “… Nature relishes unique mathematical structures. …”.
    Maybe it might be reasonable to work on the possibility that the “why” question:
    whether the observed set of Standard Model parameters might be unique
    could be answered by “because that set is determined by a unique exceptional mathematical structure”.

    Although I disagree with some other positions taken by Ramond, I do agree that exceptional structures might be a basis for construction of unique and testable physics models. Two examples of approaches using exceptional structures are (as I have mentioned in other comments):

    hep-th/0104050 in which Lee Smolin said: “A new matrix model is described, based on the exceptional Jordan algebra, J3(O). 
 There are 27 matrix degrees of freedom, which under Spin(8) transform as the vector, spinor and conjugate spinor, plus three singlets, which represent the two longitudinal coordinates plus an eleventh coordinate. Supersymmetry appears to be related to triality of the representations of Spin(8).”.
    and
    hep-th/0012037 in which Horowitz and Susskind also deal with 27-dim structure that may be related to the exceptional Jordan algebra J3(O), saying: “We 
 try to interpret bosonic string theory as a compactification of a 27 dimensional theory. We will refer to this theory as bosonic M theory.”.

    Personally, I am disappointed that now, 5 years after those papers, Smolin and Susskind seem to be working in other directions.

    I am particularly disappointed that Susskind describes his current direction in his 2005 book The Cosmic Landscape as:
    “… String Theory will almost certainly have Rube Goldberg complexity and redundancy …
    I don’t know an equation to describe it, only a slogan: “A Landscape of possibilities populated by a megaverse of actualities.” …
    … [nothing] … in this book diminish[es] the likelihood that an intelligent agent created the universe for some purpose. …”.

    Another possible “other approach” might be the emergent vacuum picture advocated by Laughlin and Chapline (popularly described in Laughlin’s book A Different Universe (Reinventing Physics from the Bottom Down)). Here is Susskind’s criticism of Laughlin’s approach:
    “… Superfluid helium is an example of a material with special “emergent” properties … In a lot of ways, superfluids are similar to the Higgs fluid that fills space and gives particles their properties. Roughly speaking Laughlin’s view can be summarized by saying that we live in such a space-filling material. He might even say … space IS such an emergent material! Moreover, he believes that gravity is an emergent phenomenon. …
    There are two serious reasons to doubt that the laws of nature are similar to the laws of emergent materials. …
    The first … Laughlin himself …[argues]… that black holes (in his theory) cannot have properties, such as Hawking radiation, that practically everyone else believes them to have …
    [second]… insensitivity to the microscopic starting point is the thing that condensed-matter physicists like best about emergent systems. But the probability that … there should be one … endpoint … with the incredibly fine-tuned properties of our anthropic world is negligible. …”.

    It is sad to me that Susskind’s first criticism of Laughlin assumes that “practically everyone” believes that black holes have Hawking radiation, when, in fact, even Hawking himself has repudiated ( see his Dublin 2004 abstract at http://www.dcu.ie/~nolanb/gr17_plenary.htm#hawking ) information loss by Hawking radiation.

    It is also sad that Susskind’s second criticism of Laughlin assumes that the process of emergence CANNOT produce an “endpoint” with precisely the properties that our experiments observe.
    Perhaps a spin foam with J3(O) nodes might produce our observed universe,
    similar to the emergence of superfluid helium from helium atoms.
    Unless and until the physics community encourages research work along such lines, how will we know the answer?

    Tony Smith
    http://www.valdostamuseum.org/hamsmith/

  38. dan says:

    Dear Urs

    “I tried to point out how a theory generically has infinitely many solutions. So it’s not clear to me why it should be a problem if some theory has only finitely many.”

    Is it possible to work backwards, from the standard model, and particles like neutrinos, electrons, quarks, and show how they are solutions to string theory? knowing what we know about electrons, is it possible to show their properties such as charge, spin, mass, etc cetera, be explained in reference to a string theory as one solution among infinitely many?

    to use your other examples, while Newton’s laws of motion and General Relativity do not predict what distance the sun’s planets should be, they include solutions such as acceleration and force, within the framework of newton/Gr. Can the same be done with string theory?

  39. Urs says:

    IF it were to be found, THEN Urs could say that it matched “a subset of experimental data” and therefore use it to make “predictions” a la SV1.

    Yes, and that’s what I said. If such a solution were found, it could be used to make predictions. As in: If you have identified which FRW solution matches large scale cosmology, then you can use it to make further predictions.

  40. Urs says:

    Having a space of solutions which is controlled by a few fundamental constants like G and h-bar is qualitatively different from having a huge out of control space of solutions, so we ought to make the difference clear.

    The space of solutions that we are talking about is essentially the space of solutions to the effective field theory equations of string theory, which look like GR+YM+ higher order corrections+this and that. This space is large in precisely the same sense as the space of solutions to the equations of pure GR is large. There are many possible configurations of the system. (Infinitely many, usually.)

    This has nothing to do with the freedom in choosing dimensionful constants like G or hbar. And it is the generic behaviour of non-perturbative theories, too. There is no reason to exepect that they allow precisely one single configuration.

  41. woit says:

    The space of solutions that we are talking about is essentially the space of solutions to the effective field theory equations of string theory, which look like GR+YM+ higher order corrections+this and that. This space is large in precisely the same sense as the space of solutions to the equations of pure GR is large.

    The size of the space of solutions is thoroughly irrelevant. What is relevant is their structure: can you use some experimental data to fix the solution, then use it to make falsifiable predictions for the rest of the data or for future data? In GR you have no trouble doing this, in the string theory landscape you can’t. The first is a science, the second isn’t.

  42. Urs says:

    The size of the space of solutions is thoroughly irrelevant.

    Yup. Because it’s generically very large, in any case.

    What is relevant is their structure: can you use some experimental data to fix the solution, then use it to make falsifiable predictions for the rest of the data or for future data?

    Exactly. That’s what I am saying all along.

    In GR you have no trouble doing this

    Yes, GR is much easier to handle.

    in the string theory landscape you can’t

    I don’t know of a theorem that one cannot. But certainly it hasn’t been accomplished yet.

    The first is a science, the second isn’t.

    Are you saying string theory is a science but studying its space of solutions is not?

    I think what you really mean is that the particular way this space of solutions is currently “investigated” by Susskind and others is not science.

  43. woit says:

    Are you saying string theory is a science but studying its space of solutions is not?

    “String theory” is lots of different things, some scientific, some not. Studying the “solutions to string theory” is scientific until the point at which you show that there are so many, capable of describing almost everything, that you can’t come up with a plausible way you’ll ever get predictions. If you keep working on this after that point, you’re not doing science, you’re just hiding your head in the sand and refusing to admit your theory has failed.

    The anthropic nonsense Susskind has come up with to try and evade the implications of not being able to predict anything is one form of non-science. But I also don’t think spending your time working out the details of these solutions is science if your only argument for doing this is wishful thinking: “maybe some miracle will happen and this stuff will allow predictions, even though I can’t now plausibly see how this could happen”.

  44. It seems as if the “string theory” were now an alternative no to GUT, but to Quantum Mechanics.

  45. Urs says:

    even though I can’t now plausibly see how this could happen

    The progress from a large number of moduli to a mere 13 was obtained by understanding the space of possibilities in more detail. Constructing these compactifications requires quite some insight into the geometry and topology of the CY appearing there. There are still simplifying assumptions in these constructions (for instance that the CY is elliptically fibred) which are made only because otherwise current technology would not allow to compute the properties of these beasts (and hence compute the predictions these solutions make).

    So it’s a matter of understanding the space of solutions, which is quite complicated. Currently, claims that there is a phenomenologically viable solution in that space are just as plausible or implausible as claims that there cannot be any.

  46. Ethan says:

    I have a problem with Urs’s analogy to GR, and I think it touches on
    my difficulties with seeing the landscape as an interesting step to understanding the physics of the universe.

    No one ever evaluated GR by comparing cosmological models. What
    actually happened was that people either constructed satisfactory
    cosmological models using GR, or using their favorite competing
    theory of gravity. If an alternative theory was ever discarded on the
    basis of cosmological observations, I’m not aware of it. GR is
    validated by relatively local tests, including solar system tests, and that
    lends credibility to our attempts to do cosmology with it, not vice
    versa. Our hypothetical cosmologist in a chaotic universe would believe
    in GR for the same reasons we do, it just wouldn’t be a very useful
    cosmological tool for him/her/it. Understanding the universe in this
    case would be a lot like trying to understand the weather.

    Supposing that some version of string theory is correct, and that
    our current understanding of the landscape is correct, i.e. that it
    gives us a finite, but absurdly large number of possible low
    energy limits. In this case it seems to me that we learn something
    interesting about our universe only by ignoring the landscape and
    exploring less ambitious phenomenological models. I can understand
    pursuing string theory because one hopes to narrow the range of
    possible low energy states to a usefully small number of choices.
    What I can’t understand is embracing the landscape in its current
    form as a useful kind of physical theory.

  47. Chris W. says:

    Peter,

    On the topic of wishful thinking, I think it’s worth quoting your reminiscence about an encounter you had with Philip Anderson when you were at Princeton:

    Phil Anderson has always been somewhat of an intellectual hero for me. He’s really the person who discovered the Higgs mechanism, among many other things. Despite a reputation for being a curmudgeon, at one point he was quite kind to me. At some sort of social event at Princeton to mark students passing their generals, he came up to me and told me that he had graded my solid state physics exam. He complimented me on one problem in particular, one I had got wrong. I had realized something was wrong with my solution of that problem, noting on my exam that the result I was getting couldn’t be right and explaining why. He told me that this had impressed him, that one should always know what the result of a calculation should look like before attempting it. [emphasis added]

    This was something that I recall being emphasized very strongly in my undergraduate education (as a physics major). It would be interesting to discuss where and when this guideline should be applied.

    I’m sure some people would argue that it is based on an unjustified faith in physical (or mathematical) intuition, in domains where such intuition can’t be relied upon. I would argue that if one’s intuition is unreliable, then one should improve it by testing it and carefully studying those instances where it fails, ie, by identifying and elucidating the hidden assumptions that led one astray. In other words, instead of looking for a better oracle, learn to think critically—learn to make effective use of trial and error at every level you can.

    (Einstein once expressed his exasperation with a certain theoretician by saying that “the man can calculate, but he can’t think”.)

  48. Urs says:

    I think the analogy to GR applied to cosmological scales is quite good. There, too, we pick a classical solution to some equations of motion, which, once fixed, predicts properties of quantum fluctuations about this solution, e.g. CMB spectra.

    The reason that we can test GR locally and perturbative ST not is a matter of energy scales, not of principle.

  49. woit says:

    The reason that we can test GR locally and perturbative ST not is a matter of energy scales, not of principle.

    No, the reason we can test GR is that we have a consistent model of GR low energy behavior that
    1. isn’t completely incompatible with observations
    2. once you choose a few parameters it is unique and makes real predictions.

    In the case of perturbative string theory, if you don’t turn on fluxes and play the other sorts of tricks KKLT does, you have unstabilized moduli and your theory is incompatible with observations. That’s why you can’t use the work of the people Urs is quoting to predict things, not the problem of the energy scales. If you do something like KKLT, again you can’t predict anything, now because it can lead to just about any low energy physics (and, again, I’m dubious that the theory is under sufficient control at high energies to do reliable calculations).

    This problem with the moduli has been around since the beginning, and all attempts I’ve seen to get around it have had huge problems of one sort or another. One can believe that with more work along the same lines someone will make it go away, but everything I’ve seen in following this more than 20 year history and looking at what people are doing now leads me to strongly believe that this is just wishful thinking.

  50. Urs says:

    Peter,

    the problem you mention is the problem of finding a phenomenologically viable background. The reason why we don’t check perturbative ST “locally” (simply by producing strings in the accelerator) is one of energy scales.

Comments are closed.