Jacques Distler has a new posting about the Swampland, based on hearing a talk by Cumrun Vafa and discussions with him in Eugene, Oregon. Vafa seems to have made clear to Jacques that what he had in mind was just what he wrote about in his paper, investigating qualitative issues such as what gauge groups could arise in string theory. Jacques notes correctly that if string theory is ever to make contact with experiment, it has to have detailed, quantitative things to say. Vafa didn’t think that such things were currently addressable, an attitude Jacques found perhaps overly cautious, although it just sounds to me realistic.
Jacques enlarges Vafa’s swampland question to make it include the obvious crucial problem for string theory: given some arbitrary choice of the 120 or so parameters of the MSSM, can you get this out of the string theory framework? He makes much of the fact that current constructions of flux vacua are parametrized by sets that are not continuous, but discrete, although of such a huge if not infinite number that it is unclear whether this is of any practical significance.
Jacques and many others seem to be of the opinion that the thing theorists should now be doing is studying the details and physical implications of these huge numbers of flux vacua. Besides the fact that this is a horribly complex and ugly business, without the slightest indication from physics that it is a promising thing to do, it seems to me to be something inherently doomed to failure. Without knowing what the non-perturbative formulation of the theory actually is, the reliability of the perturbative string approximation one is using is unclear, with wishful thinking the only reason to believe that the real world will correspond to a region where the approximation is sufficiently reliable. Furthermore, these flux vacua constructions have been accurately described by Susskind as “Rube Goldberg mechanisms”, and it seems to me likely that one can get just about whatever one wants by further complicating the mechanism. This is the completely conventional way wrong scientific ideas often fail: the simplest version of the idea doesn’t work, so people keep trying to fix it by adding more and more ugliness and complexity. Sooner or later the whole thing collapses or fades into deserved obscurity when people finally give up hope of getting anything out of it.
To make the whole question of calculating anything in this framework even worse (something hard to imagine), it seems that there is an inherent theoretical problem with the computational complexity of the question of figuring out which flux vacua correspond to specified observable quantities. Frederik Denef mentioned this in some of his recent talks and Michael Douglas will be giving a talk about this on Wednesday at the KITP, entitled “Computational Complexity of the Landscape”. I guess perhaps the new line about all this will be that string theory is the TOE, but it can be rigorously shown that one can’t ever actually calculate anything with the theory.
Update: The Douglas talk is now on-line. As far as I can tell he has now given up on the idea of doing statistics of vacua, and is instead concentrating on the problem of whether you can show that, given one of the known flux vacua constructions, some flux vacua give you what you want, e.g. a cosmological constant of the right magnitude. Given how poorly string theory on these flux vacua is actually understood, I don’t see that he can even formulate a calculation that makes any sense. But he doesn’t actually calculate anything, engaging instead in a long meta-discussion about computatibility. Kind of a weird performance. Gross seems to have been in the audience, but not spoken up. I hope he hasn’t given up.