The MIT group is, fortunately, not claiming to violate the laws of thermodynamics. I'm trying a new list-based way to organize my replies, but it won't help if I'm reading these things wrong. Take this with a whopping grain of salt!
1. Thermodynamics is a statistical study.
Everything moves, but it's hard to model this. One of my professors main interest is in modeling single molecules of complex biological things - like "When this protein folds, exactly how does it interact" or "What are the exact polymerization kinetics of Nylon 6,6?" In solids, the moving atoms can be modeled as a lattice of balls separated by springs. Kick one ball and its neighbors wiggle. This energy spreads through the substance in waves. Some is expressed as sound, some as light. We call these waves 'Phonons' because in their early study it was thought they were only sound. (Incidentally, one can calculate, quite accurately, the mass of a given metal's atoms with just the speed of sound in that metal).
2. Warm objects vibrate, and this leads to interesting effects.
Pardon the Wiki Link:
Phonon: Acoustic and optical phonons. Phonons are the vibrations of atoms in solids, and they can emit photons under certain circumstances met in many diodes. I THINK that this is what's happening: That they are measuring energy that is always emitted. The LED running at low power may regulate the wavelength of the phonon emissions, or I may be spouting complete crap.
