He refers to some experiments that were done a few years back which were misinterpreted by journalists as having created "faster than light communication". The experiments involved sending microwaves through wave conductor tubes of a certain geometry which caused the propagation velocity of some individual waves making up the wave packets [any signal sent in the form of waves -- electromagnetic, acoustic -- consists of wave packets which are made up of superpositions of individual sine waves, the mathematics of this process is called
Fourier Transform] to exceed "light speed" c ~ 3e8 m/s. But this didn't enable superluminal communication because the entire wave packet still traveled at c...
...actually, the wave conductors can be designed to make the entire wave packet exceed c! So, have we cucked Einstein and can send messages into the past, thus getting easily rich by winning the lottery? NO. In order to do that we'd need to
send information faster than c (and thus, according to relativity, into the past). While a wave packet can indeed be made to go faster than c, the wave packet's "nose", that is, it's front part where the intensity of the electromagnetic field rises from zero to some value given by the wave packet's amplitude, can never exceed c. If you want to state Special Relativity really really autistically precise, you should say: "A change in field intensity sufficient to transmit one bit of information can never exceed the velocity of c ~ 3e8 m/s".
Note that nearly all experimental physicists get this wrong. They usually say: "Individual frequencies can go faster than c, but the wave packet cannot." Wrong. Individual waves can exceed c, wave packets can as well -- but the packet's front part, where intensity changes and thus information is transmitted, can never exceed c.
A professor of theoretical physics once explained this to me over lunch.