I think he missed a couple important points. The sweep effect, shaking and aiming are more amenable to signal processing than it seems, depending on the resolution and noise in the system. I tried to illustrate in pictures: 1) Just to demonstrate that a nonuniform sweep is amenable to detection or processing, consider this picture, with the correct estimate in green: http://imgur.com/FsPAwCn 2) Now imagine the device…
It's strange how he mentions the sampling speed, but then pretends it's 1/3 second. Basically what he is attacking is the worst design of a speed gun that he could imagine from the public information. A more fair challenge would be to find sufficient error from a competent design of a speed gun, but then it wouldn't be as self serving.