I recently got a copy of Joseph Conlon’s new book Why String Theory? and was pleasantly surprised to find that it’s quite good. Conlon is a lively, entertaining writer, generally sensible about the scientific issues involved, and I think does a great job of explaining the point of view of typical physicists now working on string theory. He also very ably explains the “sociology” of the field, the different kinds of people who work in this area and their varying sorts of goals and motivations.
The book is explicitly motivated by the desire to answer a lot of the criticism of string theory that has become rather widespread in recent years (wasn’t always so…). For a typical example from the last few days, see Why String Theory is Not a Scientific Theory at Starts With a Bang. I have mixed feelings about this sort of thing. It gets the main point quite right, that string theory unification is untestable, having failed to make any predictions, and by the conventional understanding of the scientific method, it’s past the time at which most theorists should have abandoned it and moved on. On the other hand, I don’t see at all the point to arguing about the term “scientific theory”. Sure, it’s a scientific theory, a failed one. I’ve personally never noticed any consistent usage by physicists of terms like “theory”, “model” and “hypothesis” in ways that accurately indicate degree of experimental support, don’t see why some writers insist that there is one. I also very strongly object to the article’s standard move of trying to make a failed theory a “mathematical theory”. Mathematics is about well-defined ideas, and there is currently no such mathematical construct as “string theory”. The problems with string theory have nothing do with mathematics, rather have to do with a physical idea that didn’t work out.
To a large extent the problems Conlon is struggling with are ones that the community of string theorists has inflicted on itself. The great majority of writing for the public by string theorists is characterized by large amounts of outrageous hype. For a very recent example, see Daniel Harlow here, who seems to think string theory is a huge success at explaining the standard model
although it hasn’t quite managed to reproduce the complete standard model of particle physics, it comes very close and the obstructions seem more or less technical. I want to emphasize that postdictions are just as good formally as predictions for testing a theory; the distinction is purely sociological.
and that it is also much more (did you know that string theory is what explains the existence of black holes?)
the main reason to work on (or be inspired by) string theory from a scientific point of view is that it may provide explanations of phenomenon that have ALREADY been observed: the existence of black holes, the small positive cosmological constant, and the evidence for an inflationary phase of the early universe.
As for the problem with the multiverse making no predictions, that’s just wrong. We just don’t know what the theory is, when we figure it out, surely it will make predictions:
the issue is not that it doesn’t make predictions. The issue is instead that we do not yet understand it well enough theoretically to know what the predictions are!
I’ve always found reading this kind of thing quite puzzling. My impression of most string theorists is that they’re smart and rather sensible, well aware of the difference between ridiculous hype and an actual scientific argument. Unfortunately such sensible string theorists also have seen no point in trying to write for the public until now, and I’m glad that Conlon’s book finally changes that.
If you followed the reports from the recent Munich conference, you likely heard that the assembled philosophers and physicists nearly unanimously found the anthropic multiverse point of view Harlow advertises to not be legitimate science. Conlon expresses his opinion in this way, and I think it’s the majority one among string theorists, whatever you might have heard:
The most serious problem with the anthropic landscape is that it provides a cheap and lazy explanation that does not come from hard calculation and also has no clear experimental test. It sounds exciting, but does not offer lasting sustenance, and may even act as a deterrent against necessary hard work developing new calculational tools.
Of course, this does no mean that the anthropic approach is necessarily wrong. However the triumph of science has been not because it contains ideas that are not necessarily wrong, but because it contains ideas that are, in some important sense, known to be true: ideas which have either passed experimental test or are glued together by calculation. The anthropic landscape is neither of these. It represents incontinence of speculation joined to constipation of experiment.
Instead of Harlow’s claims that string theory makes lots of postdictions, coming very close to reproducing the complete standard model, modulo some technical issues, Conlon deals with the situation in a much more honest and straightforward fashion. Of the fourteen chapters of the book, chapter 7 is entitled “Direct Experimental Evidence for String Theory.” Here’s the entire content of chapter 7:
There is no direct experimental evidence for string theory.
Conlon’s point of view is different than that of the majority of string theorists in one way, which he explains in detail.
My interest in string theory is in what it can offer to physics that can be probed by experiment.
This view is far from universal. It may seem odd, but most of those who work on string theory are essentially uninterested in any connections with experiment, any public claims that they may make to the contrary notwithstanding.
He backs this up by the observation that less than 10% of talks at recents Strings 20XX conferences have any connection to observable physics.
Here, I’m again in the majority, with his colleagues, who I think have made an accurate evaluation that connecting current string theory to experiment is a failed and hopeless project (I differ with them on prospects for this changing). Conlon has a research program to investigate potentially observable effects of moduli fields, something his colleagues are skeptical about. While I’m also skeptical about this, it does seem like a reasonable thing to investigate, especially since such things may be generic to all theories with extra dimensions, not just string theory. The chapter of the book describing this research is one with material you won’t find in other popular books.
Many of his colleagues have adopted the attitude that, while connecting string theory to experiment is hopeless, it deserves investigation purely as an idea about quantum gravity. While Conlon devotes a fair amount of space to the arguments about quantum gravity and string theory claims about them (including some criticism of loop quantum gravity) he avoids much of the usual hype, and also makes it clear that he himself isn’t interested in pursuing this because of the lack of any hope of ever testing one’s ideas. In some sense I think he and I agree here: it is only if one’s idea for quantizing space-time degree of freedom connects up somehow to our successful theories of other quantized degrees of freedom that one will have any hope of ever knowing whether one has the right theory of quantum gravity. Absent a connection of this kind, one is doomed to become just another cog in an endless fruitless ideological argument about whose quantum theory of gravity is better (or at least, whose sucks less).
Conlon claims that at this point, most string theorists are interested in string theory not as a theory of quantum gravity, but because of applications of ideas that have emerged from string theory to other fields (e.g. AdS/CFT). Here he gives a reasonable account of attempts to use AdS/CFT to say something about condensed matter physics. One place in the book where he, unusually, descends to conventional incantations of hype is his account of applications of AdS/CFT to heavy-ion physics, where he says nothing about the fact that this doesn’t work very well, just repeating some rather stale hype.
There’s a lot else to like in the book, for instance a chapter of highly perceptive descriptions of the different kinds of theorists and the different ways they work, including some rather amusing and mostly friendly caricatures of common behavior. For an example of the kind of thing you’ll read here but not in any other popular string theory book, he notes that certain persons have recently received multi-million dollar prizes based upon model-building ideas that didn’t work out.
There’s a lot more in the book than I have time to discuss, some of which I agree with, some of which I don’t. Obviously I have a different point of view than Conlon’s, but his at least I find to be one with serious arguments behind it, unlike all too much of the popular string theory literature. One thing I found rather discouraging after my book came out ten years or so ago was what seemed to me a lack any serious response from sensible string theorists. Quite a few years later, it’s great to see that Conlon has written such a thing, and I recommend it highly to anyone who cares about these issues.
And, Happy Holidays!
Update: Sabine Hossenfelder has a posting with a similar take on the Siegel piece. I also like her description of the Munich workshop:
There was, however, not much feud at the workshop, because it was mainly populated by string theory proponents and multiverse opponents, who nodded to each other’s talks. The main feud, as always, will be carried out in the blogosphere…
I haven’t seen the full piece, but New Scientist now seems to be covering the multiverse as theology, which is about right.
Update: Over on Facebook Dan Harlow explains the “technological problems not relevant for questions of principle” needed to get string theory predictions
the idea is that in order to view string theory as a theory of nature, we need to view it as providing a unique probability measure on the space of low energy theories. This would be computed by understanding both the structure of the landscape and the dynamics of eternal inflation. We can then compare our observations to the predictions of this measure, and if they are atypical the theory is ruled out. We are far from doing this though, except for the imprecise cartoon that seems to more or less work for the cosmological constant. This seems just as scientific to me as quantum mechanics, except that we don’t yet know how to compute the probabilities.
I see a bunch of problems of principle here, starting with not knowing the underlying non-perturbative theory and going on from there. Some commenters over there think “It’s hard to even begin to imagine how one can even take Woit seriously.”, but it looks like they take seriously Harlow’s claims that this “seems just as scientific to me as quantum mechanics”, with the minor difference that you can’t calculate anything.