Despite having a bad cold, yesterday, I attended the DFG Schwerpunk stringtheory workshop at DESY. I was quite surprised that half of the speakers (well, including myself) talked about non-commutative geometry of one sort or the other. Does this mean that this subject is not completely dead in 2005 or does it just mean physicists in Germany have not noticed, yet?
However, the talk that caught my attention was by Michael Ratz from Bonn. He described some ideas to use finite temerature field theory in conjunction with moduli stabilization.
According to his talk, whatever you do (KKLT or race-track or, you name it), the typical moduli potential if it has a minimum at some finite value at all, it is an exponentially decreasing function with some bump in it such that the local minimum is separated form the minimum at infinity by a potential wall of height typically proportional to the gravitino mass (the susy breaking scale) squared.
But if you now crank up the temperature, the thermal effective potential tends to wash out the bump like it washes out the bump in the Mexican hat potential in the electro-weak phase transition. So above some critical temperature (roughly the geometric mean of the susy breaking and the Planck scale) there is no local mimimum anymore and the moduli run off to infinity.
Thus when you design your universe you should make sure that the reheating after inflation does not reach that critical temerature as otherwise you wouldn't have finite couplings and compact dimensions of finite size anymore later.
This bound is not very tight but Michael claims it is quite model independant. I must say I like the idea of having temperature and thus time (at cosmological scales) dependant moduli fields...