The Princeton Center for Theoretical Science has been having a mini-symposium on the string theory Landscape, and as part of this today hosted a “panel discussion” on the topic. It turns out that there’s not a lot of support for the Landscape in Princeton.
Michael Douglas was the only real Landscape proponent in evidence. He gave a presentation on the state of Landscape studies, beginning by noting that landscapeologists keep finding more possible string vacua. Evidently the 10^500 number always quoted for the number of semi-realistic vacua is no longer operative, with latest estimates more like 10^(10^5) or higher. Douglas acknowledged that this pretty much removes any hope of making predictions by using experiment to fix this freedom and end up with non-trivial constraints. All that’s left is the idea of doing statistical calculations, but there the problem is that you don’t know the measure. He ended up mainly talking about cosmology, partly about the hope that maybe cosmology would constrain the possible vacua, as well as going over various ideas for putting a measure on the space of vacua. None of this really seems to lead anywhere, with all proposed measures having a rather ad hoc character. Douglas advocated just trying to count all vacua with the same weight, since at least one might hope to calculate that.
Tom Banks began by claiming that the effective field theory picture used in the landscape is just not valid. He also pointed out that if the landscape arguments were valid, the landscape would be disconfirmed by experiment, since 10-20 of the Standard Model parameters are unconstrained by anthropics, but take unusually small values, not the random distribution one would expect. Banks takes the attitude that the CC probably has an anthropic explanation, but not particle physics or the SM parameters. He also attacked the usual claims that different vacua are all states of the same theory, arguing that they instead correspond to different theories. Finally, he pointed out that the one prediction that landscapeologists had claimed they would be able to make, the scale of SSYM breaking, hadn’t worked out at all (Douglas now acknowledges that this can’t be done).
Nati Seiberg then argued that, as one gets to deeper and deeper levels of understanding of particle physics, one might reach a level where the only explanations are environmental and have to give up. He sees no reason for that to be the case now, with the main problem that of EWSB, and nothing to indicate that anthropics has anything to do with the problem. Rather, the problem is there because we haven’t had high enough energy accelerators (the LHC should change that), and the problem is hard. He ended by saying that the appropriate response at the present time to anthropic arguments like the Landscape is to just ignore them.
The last speaker was Nima Arkani-Hamed, who I suppose was chosen as a proponent of anthropics. He didn’t live up to this, saying that he pretty much agreed with Seiberg. Like Banks, he finds the anthropic explanation of the CC a plausible reason for why no one has come up with a better idea. He did say that thinking about anthropics and the Landscape has led people to look at some possiblilities for particle physics that otherwise would not have been examined. About the cosmological issues brought up by Douglas, his opinion is that there’s probably no point to thinking about these questions now, doing so might be like trying to come up with a theory of superconductivity in 1903. As far as EWSB goes, he believes the LHC will show us a non-anthropic explanation for its scale.
He explicitly attacked the discussion of measures that Douglas had engaged in as “not fruitful”, saying that he didn’t see any “endgame”, that it was wildly improbably that these could predict anything about particle physics. He also doesn’t see why our vacuum should be typical, joking that some of the least typical people in the world (Linde was mentioned) are most devoted to claiming that our universe is typical. He went on to argue for the currently fashionable enterprise of studying S-matrix amplitudes, arguing that looking at the local physics embodied in Lagrangians was no longer so interesting, that instead one should be trying to understand questions where locality is not manifest.
Finally, Arkani-Hamed ended with the statement that string theory is useful as a way to study questions about quantum gravity, but “unlikely to tell us anything about particle physics”. This is an opinion that has become quite widespread among theorists, but news of this has not gotten out to the popular media, where the idea that string theory has something to do with the LHC keeps coming up.
So, all in all, I found myself in agreement with most of the speakers. On another positive note, the math and physics book collection at Labyrinth (which has replaced the U-store bookstore) has improved dramatically.
“Evidently the 10^500 number always quoted for the number of semi-realistic vacua is no longer operative, with latest estimates more like 10^(10^5) or higher. Douglas acknowledged that this pretty much removes any hope of making predictions by using experiment to fix this freedom and end up with non-trivial constraints.”
But isn’t the mainstream string theorist argument now the exact opposite, i.e. “the bigger the landscape, the better”? See:
http://motls.blogspot.com/2006/06/top-twelve-results-of-string-theory.html
According to Joe Polchinski (who is has done research in string theory and written a large textbook on the subject as you are aware), the twelfth top result of string theory is precisely:
“The existence of the landscape, a large enough set of metastable solutions that the cosmological constant can adjust to a value small enough as to allow organized structures (which require many bits and many cycles). [hep-th/0603249]” (Emphasis added.)
Therefore, the bigger the size of the landscape, the higher the probability that it can be somehow adjusted to fit observed physics, in principle. If the landscape size was really small, string theory could be falsified easily, and these experts (who appreciate a really big landscape over a small one) wouldn’t waste their time on it.
anon.,
Landscape proponents want at least 10^120 vacua, so as to anthropically solve the CC problem. They would like to get a prediction about something else, and for this the much larger numbers being considered now are a problem.
Peter, can you post a link to the conference site/talks if available?
thanks
Shantanu,
The only info about the conference available on-line that I know of is here:
http://pcts.princeton.edu/pcts/bigbang/Program-11-21-08.pdf
at the panel discussion, Douglas has quite a few slides, Banks just a few, Seiberg and Arkani-Hamed made informal remarks. I don’t think anything was recorded, or that more will be available on-line.
Peter, the PCTS conference link you gave was interesting in what was omitted:
Witten was not listed as being involved.
Has Witten taken a clear position on the Superstring Landscape,
or
does it appear that he is just letting it hang itself and drift slowly in the wind ?
Tony Smith
Witten is on sabbatical at CERN.
Tony,
As Aaron points out, Witten is not in Princeton this year. My impression is that he, like just most of the theorists in Princeton, is no fan of the landscape, and even if he were around, might steer clear of much involvement in this kind of program. What I find remarkable about the landscape story is how its proponents have managed to give the public the impression that it is the dominant point of view among string theorists. In actuality, I think most string theorist’s attitude toward it is that of Seiberg: it should just be ignored.
ignoring the string landscape implies ignoring string theory
I may be wrong, but I think not too long ago Arkani-Hamed was actively working with landscape-inspired models. (I even vaguely remember his calling this kind of work “staring the monster in the face”, or something like that, in an interview.) If so, his change of opinion about those models (and about the relationship between ST and Particle Physics) is quite remarkable. More generally, it seems to me that all the opinions quoted above from well known theorists about these matters reflect a tidal change of mind in the string community.
I think it is a sign of progress that 10^500 has morphed to 10^100000. It is much more effective in keeping more people from wasting their time on it. Let a small contingent of researchers keep working on this and improving the intractability results of the landscape.
Sumar Ongi said “… Arkani-Hamed was … calling … landscape-inspired model … “staring the monster in the face” …”
and
Peter said “… Nima Arkani-Hamed … went on to argue for the currently fashionable enterprise of studying S-matrix amplitudes, arguing that looking at the local physics embodied in Lagrangians was no longer so interesting …”.
Is Arkani-Hamed’s “landscape-inspired model” that is “the currently fashionable enterprise of studying S-matrix amplitudes”
actually
his MARMOSET approach described in hep-ph/0703088
and discussed in a RESONAANCES blog entry dated 1 July 2007 entitled “Nima’s Marmoset” ?
In view of the success of the Standard Model Local Lagrangian,
is it reasonable to say that “local physics” of “Lagrangians” is “no longer so interesting” ?
Tony Smith
Tony,
You’re mixing up three completely unrelated things (landscape-inspired models, studying S-matrix amplitudes, and MARMOSET), with nothing much in common except that Arkani-Hamed has worked on them.
What exactly is it that there are N=10^{many} of? (What does “possible string vacua” mean?) How are they counted? Does each one of these N “vacua”, have their own continuously variable parameters (like the standard model has), in which case, why don’t they say there are infinitely many and be done with the counting?
Finally, is there any sense in which string theory “predicts” that these “vacua” actually exist in physical reality, or is it just that some people choose to add the separate and additional hypothesis that things that might be theoretically possible, are guaranteed to actually exist as part of physical reality? (And couldn’t you just make a “landscape” out of the standard model and its own continuously variable parameters, or out of whatever variable model you like?)
It just seems that some of these theorists are taking their models way too seriously, and have too narrowly focused on a limited range of possibilities.
hmmm,
For the answers to you first questions, see review articles by Douglas and others. Plenty of string vacua do have continuous parameters, but these are supposed to correspond to massless fields. We don’t see these, so you ignore those vacua. The 10^500 or larger number is a counting of vacua that aren’t in obvious conflict with experiment.
Whether these “vacua” are legitimate metastable ground states of the full non-perturbative string theory is controversial. No way to resolve the controversy, since no one knows what the full non-perturbative string theory is.
I’ve gone on at length elsewhere why this is different than the standard model, but the obvious point is that the string landscape is thoroughly non-predictive about anything, whereas the standard model makes a vast number of detailed, testable (and tested) predictions.
PW,
Do you support research aimed at writing down “full non-perturbative string theory”?
What can full non-perturbative string theory tell us that current perturbation series can’t?
regards,
dan
dan,
Sure, some people should continue thinking about this. The problem right now is that, besides gauge/string duality, no one has any very promising ideas. There are a long list of things that people would like string theory to do (provide a predictive TOE, a full quantum theory of black holes, a string dual to QCD, etc, etc.), but no one can tell whether this will work, because no one is sure what string theory “is”, outside of perturbation theory. If that question had a definitive answer, one could start getting agreement on what string theory is good for, and what it isn’t good for, which would be very helpful.
Do anthropic explanations include cosmological natural selection (a la Lee Smolin)?
Gordon McCabe Says: “Do anthropic explanations include cosmological natural selection (a la Lee Smolin)?”
No. Definitely not. They are completely different.
Anthropic explanations are more akin to postulating that the universe generates a huge number of random configurations of matter, some of which happen to be elephants.
Cosmological natural selection is more akin to the usual natural selection and its explanation for elephants.
10^500 might be “the number of semi-realisitic vacua in ST usualy quoted”, but this is the number of vacua in a more or less typical KKLT flux compactification senario. There are a possibly infinite number of Calabi-Yau’s etc. that you could use to play that game. The question is then what are the superselection sectors, but no one knows that I guess…
Two years ago in Les Houches Summer school, Nima was still a very strong proponent of anthropic argument, and not only for the CC. It was fun listening to him, even a 3 in the morning, and it got me thinking a lot. I’m glad if he changed his mind though.
I recognize that cosmological natural selection (CNS) is distinct from the anthropic principle (AP); in CNS, a life-permitting universe is a typical member of the universe population, whilst according to the AP, it is a very special member of the population.
However, I wondered if CNS falls under the aegis of anthropic explanations for the purpose of the arguments in this article. I’m thinking of the following:
“He also pointed out that if the landscape arguments were valid, the landscape would be disconfirmed by experiment, since 10-20 of the Standard Model parameters are unconstrained by anthropics, but take unusually small values, not the random distribution one would expect. Banks takes the attitude that the CC probably has an anthropic explanation, but not particle physics or the SM parameters.”
Can cosmological natural selection explain those Standard Model parameters, or does it leave them unconstrained, just like the anthropic principle?
To Gordon McCabe:
Good question, I don’t know. You would have to go to the original source. I would guess there is much less written on CNS than (various versions of) AP.
But I would think that for Standard Model parameters that are unconstrained (or loosely constrained), it wouldn’t matter. Their values could drift (between generations) with neutral effect, and so CNS would not “explain” their specific values, nor would those values falsify or confirm CNS. (Just like selection-neutral genes.) There are still plenty of (SM and cosmological) parameters that could falsify CNS.
I’m not an expert. I just read some things and formed my own conclusions.