The Landscape in Toronto

Most of the talks at Strings 2005 about the landscape have now taken place, although there’s at least one more this afternoon by Dine. Frederik Denef gave a survey talk entitled Constructions and distributions of string vacua. One amusing thing he does is note that even in toy models with these exponentially large numbers of states, counting the number of states with vacuum energy less than some bound is computationally an NP-hard problem. He describes a wide range of constructions that people have come up with to fix the moduli, concluding that you can “throw enough ingredients together to get sufficiently complicated potential, and this will fix moduli, at least at effective field theory level”, but that these constructions are “ugly”. He then goes on to survey various results about the statistical distributions of these states, and ends by announcing a workshop in Trieste next spring on “String Vacua and the Landscape.”

The talk on Is the number of string theory vacua finite? by Michael Douglas makes Denef’s survey of distributions of vacua kind of pointless. The number of such vacua is definitely infinite, which ruins ones ability to get a probability distribution by counting vacua. Douglas hopes that by putting in a cutoff on the diameter and volume of the compactification space, as well as the size of the vacuum energy, he can make the number of vacua finite. He explains this conjecture, for which the evidence is not very compelling.

Even if he gets the finiteness he hopes for after imposing these cutoffs, the problem then is that the distributions of vacua depend strongly on the cutoff and are peaked at the cutoff value. This is what happens in examples that Kachru talked about at the conference. Douglas is reduced to arguing that “it seems a priori plausible that cosmological selection could depend on the volume of the extra dimensions”, i.e., that somehow the Big Bang would get rid of the problem that his program is predicting large compactification spaces when he wants small ones. There seems to be no reason for this other than wishful thinking. One thing is clear though now: it makes no sense to spend time computing distributions of these vacua, since this gives a result you don’t want. In this game though, it’s not like you give up on your research program when it gives results that don’t look at all like the real world.

This entry was posted in Uncategorized. Bookmark the permalink.

19 Responses to The Landscape in Toronto

  1. Peter Woit says:

    Hi jkg,

    The Douglas program is based on the idea that we live in a randomly chosen (by not understood dynamics of the early universe) part of the multiverse, randomly chosen among those in which life is possible. If this is true we would expect to be in vacua corresponding to high probabilities, not low ones (consistent with anthropic principle). Douglas et al hope (hoped?) to use this idea to make some predictions. It is these hopes that have collapsed.

    Susskind has written a whole book about why this is a great idea, you can read it later this year. Personally I always thought this was just nuts.

  2. Torbjorn Larsson says:

    I’m not qualified to say much in this except that the difference between Peter and Bob (and me) is split by ‘among which String methods can not _yet_ distinguish’.

    The ‘not yet’ is wide open to be realised as ‘soon’ to ‘never’. I’m not sure how we would be certain of any of these extreme cases.

  3. jkg says:


    Sorry for asking naive question, but I do not quite get what could be the relevance of probability distribution of universes in multiverse (if you can calculate it, and I understand that you cannot in the landscape scenario.) Suppose in string theory I you calculate the probability of having our universe to be 99.9% and in another string theory II this probability is 10^{-10}%. So what? Should I say that the second theory is wrong and the first is right? Certainly not! Leibniz said that we live in the best of all possible worlds, which is perhaps right, but why am I to believe the we live in the most probable one (or most improbable, or whatever?)

  4. Peter says:


    If you think about it for a minute, you’ll see there’s an obvious problem with your analogy between QFT and string theory. You can use QFT to make an infinite number of precise experimental predictions, then go out and check them and they work. This is called doing science. Sure, QFTs don’t predict absolutely everything, but they come with a precise understanding of what they predict and don’t predict. There are lots of QFTs, but one of the simplest ones seems to work perfectly.

    The problems with string theory in general are complex, but the problem with the landscape is very simple: it’s not science. The one hope for getting predictions out of it (one for which there was never any evidence, just pure wishful thinking), Douglas’s statistical program, has now collapsed, although its practicioners seem to be having trouble clearly admitting this.

    The people investigating the infinite number of vacuum states that make up the landscape can’t tell us how they’re going to ever get any sort of prediction out of this. What physical quantity are you going to calculate, use to make a prediction such that it can be checked and if it is wrong, you’ll admit that the idea is wrong? It’s fine if you can’t do this calculation yet, but it’s not fine if you don’t have a plausible program for getting there.

    What string theorists are doing now is launching into an endless investigation of an infinitely complicated structure, with no scientific justification for doing this. The only justification is the idea that string theory must be right, and this is what string theory leads to. This is no longer science, it’s a cult.

  5. Bob McNees says:

    “that is, there were a whole bunch (one followed by 100 zeros) of quite different versions of nature—with all different basic features and constants—among which String methods could not distinguish!”

    This is a bit of a disingenuous comment. I know, it’s hard to believe that anyone around here would offer a gross oversimplification of the state of things in string theory, but it seems to have inadvertently happened.

    First of all, the notion of “the landscape” has been met with more than a healthy amount of skepticism in the string community. That’s not to say that it’s entirely wrong, but it’s certainly not the universally accepted source of communal shame that Peter makes it out to be. But let’s suppose, for the sake of argument, that all of the arguments surrounding the landscape go through, or even that Douglas’ speculation about the number of vacua being infinite holds up. Claiming that this dooms string theory requires that we already have a near-complete technical mastery of the subject, so that the notion of large (or infinite) families of vacua becomes an irreparable flaw in the theory.

    That’s just not true.

    Consider the following analogy: write down every field theory you can think of. I’ll even spot you four dimensions…go ahead, start writing. How many do you have? Is it a few? A few dozen? 10^100? 10^500? An infinite number? What does this have to do with the landscape, you say? Pretend for a moment that you’re ignorant of every mechanism by which you might effectively reduce the number of theories you’ve just written down, to those which you feel qualify as “realistic”. How will we ever describe nature? My lord, with so many theories, how will we ever make progress? Field theory was supposed to unify quantum mechanics and special relativity, but all it has done is land us in this embarassing morass of theories consistent with quantum mechanics and special relativity. Let’s all go home…field theory is a bust.

    Of course, now I’m the one who’s being disingenuous. You know lots of physical principles that narrow down the number of realistic field theories. We’ve even singled one out as describing most of what we see. But based on your reasoning, why should you even start to narrow down the field? Such a large task, even if it was only made large because you began with an incomplete knowledge of what principles you should apply, is apparently reason to just give up.

    The point is, we have no reason to believe that string theory isn’t full of selection principles that deal precisely with Peter’s (and many string theorists’) objections to the landscape. Personally, I’m not sold on the landscape. But I am aware that, left unaddressed, the idea of lots and lots of vacua for string theory might pose a problem for predictability. Do I abandon the field? No. In somewhat novel fashion, I conclude that perhaps I should work on the problem. Quixotic, I know, but I’ve heard that actually working on a problem usually gets better results than complaining about the problem.

    “But one should always have hope.”

    Yes, one should. String theory is hard. We’re working on understanding it. We run into hard problems all the time, and what to make of the landscape is one of them. As discussed earlier in the post, progress on such problems is often made by working on them.

  6. Anonymous says:

    Torbjorn, at the time I was talking about, they could not distinguish the right one. I cannot speculate about future String methods and what they will be able or not able to do about the Landscape. You probably remember it was described as a needle in a haystack—or Lubos Motl compared Landscapeology to a drunken man looking for the keys to his house not in the light of a street-lamp but in the middle of the ocean, on the surface. But one should always have hope.

  7. Alejandro Rivero says:

    And, every time I look in there, full of inaccurate information.

    Hey, our thread on numerology tries to keep a high standard of 99.9% numerical accuracy and above! Perhaps you mean nonsensical or misleading. But we are accurate.

  8. Torbjorn Larsson says:

    Thanks for the physicsforums tip.

    Since we laymen already lay about, perhaps you can clarify why “…among which String methods could not distinguish” is necessarily true.

    The KKLT model is but a very special case of a huge class of more general compactifications. It is made, as I understand it, primarily to have 4 large dimensions and a small positive cosmological constant. (It also entails other good benefits.)

    Why do we discount other compactifications and methods to constrain vacua just because this particular case failed?

  9. Aaron Bergman says:

    did you already try with your question?
    it is more oriented toward layfolk (general reader-type) questions.

    And, every time I look in there, full of inaccurate information.

  10. Anonymous says:


    did you already try with your question?
    it is more oriented toward layfolk (general reader-type) questions.

    the gist of the “Landscape” business is that around January 2003 there was a massive setback in the String research program. Some people at Stanford (Kachru, Kallosh, Linde, Trivedi) came out with a paper showing that String doesnt predict just one version of physics—this is what had been hoped for: a theory that would explain why a lot of the basic constants in physics are what they are.

    In fact the KKLT paper showed that String doesnt even narrow it down to just 2 or a dozen or a thousand different versions. KKLT found that the number of possible String “ground states” or “vacuum states” was more like ten-to-the-hundred. that is, there were a whole bunch (one followed by 100 zeros) of quite different versions of nature—with all different basic features and constants—among which String methods could not distinguish!

    This became known as the “Landscape” of string versions of physics, or (to put it in a more technical language) the Landscape of String Vacua. (Vacua or vacuums, in this case, just means the possible ground states or basic blank conditions of nature.)

    Later researchers (after january 2003) determined that the number was more than ten-to-hundred, it was maybe more like ten-to-500, or maybe infinite. This year infinite seems to be emerging as a favorite estimate.

    This development has been interpreted by some as ending String hopes of being a useful scientific theory (able to predict unique outcomes and explain a lot of stuff). One possible response to the existence of the Landscape is to say that String is over and it is time to look for a different theory.

    Another possible reaction which some String theorists have shown is to try to “save” the program by abandoning the expectation that a physical theory should be testable by making unique predictions. This has lead to talk about “Anthropic Principle” or as some say, the “Anthropic Lack of Principles”.

    We heard a lot of talk about that in 2003 and 2004 but the word has fallen out of favor. The idea was that the world is just the way it is and we dont have to explain it, because if it were very different then life wouldnt have evolved and we wouldnt be here talking about it. You wouldnt have Conscious Life-forms Like Us (“anthropic” is just a technicalsounding word for “Like Us” in a very broad sense, here).

    Now String theorists seem embarrassed by the word “Anthropic” and they are using the word “Environmental”. they sometimes say that some basic key physical constants don’t have to be explained because they just are what they happened to be.

    This is a way of saving String by excusing it from having to predict or explain basic features of nature. It is an abdication from the traditional scientific quest. String theory is all right as long as we decide we dont expect a fundamental theory to predict unique values of the constants of nature. Like, why, on the macroscopic level that we can see, does spacetime have 4 dimensions? Well one answer is let’s not try to explain that, it just is. It is environmental which here is a code-word for “how it happened to turn out”.

    Maybe the number 4 dimensions is a bad example and i should be taling about something more technical like the cosmological constant or the fine structure constant, but maybe you get what I’m trying to say without that.

  11. pseudo string fan says:

    Hi guys,

    I often meet the following terms in string theory’s discussion:landscape and anthropic principle

    I think landscape in string theory is quite different from that found in a dictionary. I also can’t find the explanation of anthropic in my dictionary. Could someone tell me what they mean in string theory? Thank you in advance.

  12. Alejandro Rivero says:

    What is depressing, is the following slide;large#slideloc

    The author seems to think that his audience is not going to be interested in anything related to experimental input, and he uses a full slide to justify the interest on QCD. Pitier if one thinks that string QCD is the main silver bridge for young graduates to return to reality.

  13. Alejandro Rivero says:

    Not rare, if you have sometime seen these panels, which are rather formal. Panelists either took the opportunity to keep selling his own product, or try to come when an enthusiastic, politically correct, speech to please the public. I think it should be funnier, and more interesting, if some place were left to think aloud, instead of reading aloud.

  14. Anonymous says:

    In terms of the physics, the panelists gave their ideas for about an hour. Then there were some comments and ideas from the audience. There was some discussion of the anthropic stuff, but it didn’t come close to dominating the two hours. The moderator ended with the vote and that was it.

    Thanks for the report, Aaron.

    Impressive lineup:

    Raphael Bousso (UC Berkeley)
    Shamit Kachru (SLAC & Stanford)
    Ashok Sen (Harish-Chandra Research Institute)
    Juan Maldacena (IAS, Princeton)

    Andrew Strominger (Harvard)
    Joseph Polchinski (KITP & UC Santa Barbara)
    Steve Shenker (Stanford), moderator.
    Eva Silverstein (SLAC & Stanford)
    Nathan Seiberg (IAS, Princeton)

    The organizers’ announced purpose of the session was “to explore, in an informal and interactive way, possible directions that may lead to major new progress in our field.”

    Judging from your account not much worth telling about except a show of hands 4:1 in favor of the cosmological constant being “physical” as opposed to “environomental”.

    If you remember anything else that struck you as interesting, later on, hope you will share it with us. Thanks again for the reportage.

  15. Torbjorn Larsson says:

    So no ideas worth mentioning and a vote not worth explaining?

  16. Aaron says:

    In terms of the physics, the panelists gave their ideas for about an hour. Then there were some comments and ideas from the audience. There was some discussion of the anthropic stuff, but it didn’t come close to dominating the two hours. The moderator ended with the vote and that was it.

  17. Anonymous says:

    Aaron:There’s just not all that much to say about it, physicswise.

    In that case, if it’s no big deal, then there’s no reason not to report your impressions and what you got of the sense of the discussion, is there Aaron? So please report.

  18. Aaron says:

    I’d be careful, if I were you, about ascribing beliefs on the anthropic principle without talking to the actual people. (One of the advantages of a conference like this is that you actually get to do that. I don’t feel it’s appropriate for me to talk for other people, however, so I’ll just leave it there.)

    I also wouldn’t get too excited over the panel discussion, either. It wasn’t some great string theory Waterloo or anything. There’s just not all that much to say about it, physicswise.

  19. Anonymous says:

    It’s hard to know where string theorists are on the Landscape issue collectively (if there is a common viewpoint) because of the failure of anyone to report Tuesday’s panel discussion

    In the comments on your blog #218, I remember Aaron did say there was a show of hands at the end of the session:

    “I won’t say much about what else happened, but at the end, there was a vote of hands on whether the cosmological constant was ‘environmental’ or ‘physical’. The results: 4:1 or so against.

    Posted by: Aaron at July 13, 2005 03:53 PM”

    He doesnt say against what. presumably “environmental” = resorting to Anthropic Principle (approx.) and the showing was “4:1 or so” against appealing to the Anthropic Principle. Hope I am not misconstruing.

    Don’t know if Aaron actually counted or if he or someone just eyeballed it and said 4:1 because it felt right. Not sure if I am interpreting it right when I think of it as a strawvote that went against Susskind and the (Dine Douglas Denef) consulters of the entrails. Maybe it doesnt matter how the vote went and what the issue really was.

    But at least it sounds like the folks Tuesday were concerned about “environmental versus physical” and worried enough that none of them want public notice of what went on in their discussion.

Comments are closed.