Background independence, LQG and string theory

The Statistical Mechanic, 2006-08-26

Lee Smolin is a well know skeptic of string theory and recently published a book about The Trouble with Physics. He insists that physics, and in particular quantum gravity, must be background independent and uses this as one argument against string theory. He explained this point some time ago as follows [1, 2]: "Any acceptable quantum theory of gravity must incorporate the basic lesson of GR that the geometry of spacetime is dynamical and defined without reference to any fixed background - hence background independent. String theory has as yet no such formulation. [..]
Even at the background dependent level GR is not recovered because all backgrounds on which consistent propagation of worldsheets have been shown are static in that they have timelike or null killing fields. [..]
The basic reason to expect so is that the algebra of supersymmetry transformations closes on the hamiltonian, which is the generator of time translation symmetry."
Lee is working on a proposal known as 'loop quantum gravity' (LQG), which should be such a 'background independent' quantum theory of gravitation. It meets and clashes with string theory at the example of 2d gravity (with additional matter), a.k.a the bosonic string. The standard quantization of the bosonic string is well understood since many years and one important result is the consistency requirement of the critical dimension D=26. (In the case of superstrings the number of extra dimensions is reduced and D=10.) Along came Thomas Thiemann with the tools of LQG and found that a 'background independent' quantization does not require such extra dimensions.
This result was discussed at the String Coffe Table and further examined in the paper by Helling and Policastro, who basically reject it as unphysical; They simplified the discussion to a single harmonic oscillator and argued that LQG methods yield results which contradict the well known result confirmed by many experiments [+]. Does this mean that LQG is already falsified by past experiments?
If it would just be that easy. First of all, LQG is really more a collection of ideas rather than one theory; Ashtekar et al. suggest a very different result for the harmonic oscillator, which is also different from the standard result, but similar enough that current experiments could not distinguish between them. (See also this paper.) And then there are some basic questions: We never observe a single, isolated harmonic oscillator. Rather we observe systems coupled somehow to a measurement device, which is part of a whole universe. (Solutions of GR also describe complete universes only.) Thus if a consistent 'background independent' description of a single harmonic oscillator cannot be found, it could be due to the fact that such an isolated, single system cannot exist [*].
I am afraid that without experiments this debate will not be settled any time soon. Unfortunately, such direct experiments may not be possible with our technology and this is in my opinion the real Trouble with Physics.
But in a recent radio interview Lee suggests that a good theory should find experimental support within 5-10 years. In my estimation LQG is about 10 years old and time is running out ... 8-)

[*] If you think this argument is a bit far fetched, I would not argue about it, but only mention that similar reasoning is quite common in debates about LQG vs. string theory, as one can see in the comments to this post.

[+] update: Thomas Thiemann mentions and discusses these issues on p.44ff of his latest paper as "A folklore statement that seems to have entered several physics blogs [..]". (via Lubos Motl).

update: Robert Helling responds here to this latest paper. You may notice that he mentions the argument "you would have to couple [the harmonic oscillator] to the radiation field and thus the full system is interacting and much more complicated", similar to the one I made above.

Back to The Statistical Mechanic