Science magazine this week has an article about the anthropic string theory landscape controversy, entitled A ‘Landscape” Too Far, by Tom Siegfried. The only theorist quoted as opposing anthropic landscape arguments as not science is David Gross, although experimentalist Burton Richter’s talk at SUSY 2006, and letter to the Times (“I can’t understand why they don’t take up something else — macrame, for example”) are also quoted. Gross says that anthropic explanations are not science but “fun parlor games”, that “they’re not science in the usual sense of making predictions that can be tested to better and better precision over the years.”
Quoted as strongly in favor of the anthropic landscape are Susskind, Linde and Polchinski (there’s an extensive side article about Polchinski’s conversion experience to the anthropic ideology). Sean Carroll and Frank Wilczek promote the idea of the multiverse as a new Copernican revolution, and Clifford Johnson defends anthropic landscape studies with:
It would be nice if we could explore some of those unpalatable ideas just in case that’s the way nature chooses to go.
Clifford has a posting about this on his blog, where he has more to say about this. He seems to have decided to deal with the very uncomfortable position that the evidence and rules of logic put string theorists in by advocating ignoring logic, quoting Moshe Roszali approvingly about the desirability of being able to hold contradictory viewpoints simultaneously.
The Science article does get a very little bit into the crucial question that determines whether landscape studies are science or not: is there experimental evidence that can test the hypothesis? Andrei Linde objects to people who say this subject is not science with:
It’s not an easy job to do, so if you don’t want to do it, then don’t do it. But don’t say it’s not science.
It’s true that the anthropic landscape is incredibly complicated and difficult to do anything with, but I don’t see how that fact is any kind of argument in favor of it being a science. Linde does claim that gravitational waves can be use to “verify anthropic predictions about the nature of spacetime curvature.” I don’t know exactly what that’s about, presumably something to do with possible effects in the CMB due to our universe being born out of a bubble nucleation. If anyone knows of any precise “anthropic prediction” of this kind, I’d be interested to hear it. But, in any case, whether or not you can by observation see whether the universe arose in this way, I don’t think Linde answers at all the objection that the string theory anthropic landscape is inherently unpredictive and thus not legitimate science.
The Science article also includes a heavily overhyped statement about the experimental support for inflation, describing the WMAP results as having “provided strong support for inflation’s predictions.”
For a much more serious discussion of whether the string theory landscape, anthropic or not, is inherently unpredictive, you can watch the video of a talk given yesterday at the KITP by Wati Taylor on String Vacua and the Quest for Predictions. This was the inaugural talk for the semester-long program on string theory phenomenology that will be taking place in Santa Barbara. The blurb for the program is a masterpiece of hype, telling us that string theory has “the potential to predict properties of superpartners that might be found at the Tevatron or LHC and provide new experimental tests and probes of the theory”, something that I don’t think any serious person actually believes these days.
Taylor’s talk was quite remarkable, very explicitly going over exactly how bad the current situation is for efforts to get any prediction at all out of string theory. There was a lot of discussion with the audience, and much nervous laughter. Unfortunately I found some of Gross’s comments hard to hear. Taylor explained that after spending ten years himself working on trying to better understand what string theory is (he worked in string field theory), he doesn’t see any realistic prospects for significant progress on this problem during the next ten years. He listed the basic problems as the lack of a non-perturbative definition in anything but special, non-physical backgrounds, the inability to do even perturbative calculations in the kind of Ramond-Ramond backgrounds that people are using to stabilize moduli, and the lack of any definition of string theory when supersymmetry is broken by a positive CC, and thus the background is deSitter.
Discussing the landscape, he said that there was no evidence for a dynamical principle that would select the vacuum, with no hint at all of how such a thing would work, and that there is no known mechanism that would destabilize the known conjectured constructions of vacua. He goes on to ask “what can we do even if we don’t know what we’re talking about?”
He introduced his own current philosophy, which is that unless some dramatic new breakthrough comes along in string theory (which he didn’t seem optimistic about), the only idea for getting a prediction out of string theory that is still conceivable is to look for strong correlations among standard model parameters in the landscape. He didn’t even bother to mention the fashionable idea of a couple years ago that one could make predictions using statistics of vacua, that idea seems to be completely dead. He noted that as time goes on, people keep finding more and more constructions of vacua, and it now seems clear that there are so many of these that one can’t use their hoped-for discrete nature to make predictions.
According to Taylor, the only possible hope for getting a prediction out of string theory is if one can show that, for all string vacua, there is some strong correlation between values of the low energy field theory parameters. If it turns out that (for example), for all string vacua the number of generations is always 3 when there is an SU(3) factor in the gauge group, then knowing about SU(3) predicts the number of generations. There’s no known reason why anything like this should be true, and it sounds like pure wishful thinking to me, but I guess Taylor’s point of view is that string theorists should be working harder on understanding the details of the landscape in the hope of finding such a thing, because it is the only hope for getting a prediction out of the theory, and thus justifying it as a science.
Taylor acknowledges that the state of affairs is that one can’t do at all realistic calculations along these lines, but he has been doing some unrealistic ones with Michael Douglas. They’ve been looking for correlations between the size of the gauge group and the number of chiral generations in intersecting brane models. These are quite unrealistic, with no supersymmetry breaking and unstabilized moduli. In any case, their result is negative: even in this simplified, unrealistic context, they find no sizable correlations.
Given this start, it will be interesting to see how the participants manage to get through the semester without getting so depressed about prospects for string theory that they abandon it and go on to something else. One new feature of the program is that a wiki has been set up to allow for communication and discussion between the participants.