Our friends in Seattle seem to be really good with the gravitational duals of QCD and many related questions, and I plan to write something about the interesting things that Andreas Karch and Matt Strassler told me about AdS/QCD and some issues of the LHC later.
It's raining here and the weather has the ability to affect the impression from any city. The first time when I visited New York, it was also raining - and I obviously liked the City much more when it was sunny.
Andreas told me about the successful calculations of meson masses from the holographic perspective. One considers a dual free theory whose structure is more or less dictated by symmetries; she writes down the bifundamental scalar fields that break the chiral symmetry down to the diagonal subgroup. The result is surprisingly good - five or ten percent errors of the meson masses despite a very small number of input parameters.
The three-point functions and higher do not work as well, and the properties of QCD objects that really look like strings - especially those with spin 2 and higher - also don't work too well. Of course, a question is whether the surprisingly good agreement concerning the masses (two-point functions) of the low-spin QCD mesons is just an example of a good luck, or whether we can find some arguments that it is a reflection of some underlying "truth" whose other predictions could also be trusted. It has not been quite explained which of these things should work and which of them should not and why.
Matt Strassler agrees with this broader picture and it is one of the reasons why he focuses on model-independent things.
Matt also argues that the LHC should not be thought of as a gluon-gluon collider. In his picture, the number of very interesting events with very high center-of-mass energies could be dominated by gluon-quark scattering. So we exchanged some ideas and scalings concerning the parton distributions of quarks and gluons (and less importantly also antiquarks) inside baryons, with Matt of course being the much better knowledgeable party in the discussion.
In QCD, the couplings of an excited (spin one) rho' mesons seem to be pretty universal, up to small errors, and this fact can be explained from the dual theory because they correspond to the lowest mode of a five-dimensional gauge field in a cavity whose profile has no zeroes. The relevant couplings are obtained from overlap integrals and if you know that the profile has no zeroes and behaves in a more or less pre-determined way in one extreme region, the result for the profile - and consequently also the overlap integrals - is determined and therefore universal, up to small corrections. That's the reason why the three-point couplings of rho' seem to be pretty universal while you could not say the same things about the ever higher excitations.