The Incoherent Ponderer has a fascinating analysis up of the statistics of the PhD-to-faculty pipeline in physics. The one thing missing (for lack of a good source of statistics) is how many physics PhDs go on to become faculty in a different discipline. This is increasingly common in this age of interdisciplinary work. For example, while by the IP's rankings Rice only places 1.9 percent of its PhDs as faculty members in top-50 physics departments, I can think of a few who are now faculty in, e.g., EE, Mat Sci, BioE, Chemistry, etc. It would be very interesting to look at the trends over the last twenty or thirty years. One reason for the pedigree effect is that good science is correlated with having cutting-edge resources - as fancier facilities (at least in condensed matter) have trickled down to the masses, so to speak, have things become more egalitarian?
Two more points.... First, I have some nagging doubts about the validity of some of those numbers. I can already count 7 Stanford PhD alumni that I know who have assistant/assoc. faculty positions in top-50 universities. According to the AIP numbers, that's 25% of all of the ones out there. That seems hard for me to believe. Second, Chad Orzel has a very valid observation that goes to the heart of a pathology in our field. 93% of all colleges and universities are not in the top 50. As a discipline I think we do real sociological damage to our students when we brain-wash them into thinking that the only successful outcome of a graduate degree is a tenured job at Harvard. That kind of snobbery is harmful, and probably has something to do with attrition rates. People should not decide that they're failures because R1 academia isn't what they want to do. I thought hard about taking a job offer from a college, and I still resent the fact that some people clearly thought I was loopy for even considering that path.
arxiv:0706.0381 - Fiebig et al., Conservation of energy in coherent backscattering of light
This paper is at once a very nice piece of experimental work, and an example of the kind of argument that I really don't like. In mesoscopic physics, there is a phenomenon known as weak localization for electrons. Consider an electron moving through a disordered medium, and look at one particular trajectory that contains a closed loop (made up of straight propagation pieces and elastic scattering events). Feynman says that the amplitude corresponding to this trajectory is a complex number whose phase is found by adding up the phase from propagation along the straight segments plus the phase shifts from the scattering events. Now consider a second trajectory, identical to the first, but traversing the loop in the opposite direction. It turns out that the amplitudes of these two trajectories interfere constructively for backscattering by the loop. That is, the quantum probability for getting through the loop is below the classical value, and the quantum probability for getting reflected by the loop excedes the classical value. It turns out something very analogous to this can happen for light propagating through a diffusive medium, and this can be the basis for some really cool things, like random lasers (where the back-scattering itself acts like an effective cavity!). The authors of this paper show the physics of this beautifully, but they present it in the form of a straw man argument, saying that the coherent scattering result (with greater than classical backscattering) looks at first glance like it violates conservation of energy. No, it doesn't. It looks like coherent scattering. It doesn't look like a violation of conservation of energy any more than typical diffraction does.
arxiv:0705.4260 - Huang et al., Experimental realization of a silicon spin field-effect transistor
For nearly 17 years people have been trying to make a spin transistor of the type discussed here. The idea is that spins are injected from a magnetically polarized source, traverse a channel region, and then try to leave through a magnetically polarized grain. Depending on the gate electric field, the moving spins precess and either get out of the system or not depending on their eventual alignment relative to the drain magnetization. This has historically been extremely difficult for many reasons, not the least of which are the difficulty in injecting highly polarized carriers into a semiconductor and the annoying fact that spin polarization, unlike charge, can relax away to nothing. Well, this is a pretty convincing demo of a device quite close in concept to the original idea, though it's not a field-effect geometry as first conceived. Very pretty data.