This article is one of many on this interesting topic. I was interested, since I'm quite up on my x-men, and I know that thought-waves travel faster than light! To have earthquakes zipping around the world at that speed, is quite a scary thing!
Of course, the headline writers belong to a different union than the authors, plus, everybody steals from the original article. It turns out to be a rather mundane study, that shows fault propagation faster than a slow S-wave.
This is just basic physics. As mentioned in the boring stories, when a fault is all primed to go, it just takes a little slip to shear the adhesion points, and change the state to dynamic friction. Most faults are not like your feet in the bathtub, in that they are very rough, so it takes a good hunk of displacement to fly. This is generally called the critical displacement (d-c). (Imagine a subscript!).
When all the little feet are primed, ready to slip, we can see that a fault rupture activates seismic waves which supply the critical displacement. If the fault is rough, and the value of d-c is high, then only an S-wave has enough oomph to knock it into action. The fault rupture moves along at a pokey 2 km/s. The faster P-waves just go flying by.
If, however, we have a smooth bathtub-fault, then d-c becomes vanishingly small. This is what I think happens with those fabulously destructive blind thrust earthquakes. Then, the leading P-wave has enough displacement to trigger the slip. When this happens, the leading wave becomes a giant fist, ready to smash any little stone building (or nuclear power plant) in its way. I think our favourite Japanese n-plant got hit with this.
Anyway, the study claims to have measured this on a strike-slip fault, which is rare, but much easier to measure, since you've got seismometers along the length. It will never be easily measured on the thrust faults.