For anyone who’s not fully grocked Nimtz’s and Stahlhofen’s experiment, here’s a summary:
The inside surface of one face of a triangular glass prism acts as a mirror, reflects effectively 100% of visible light. place 2 well-made (faces perfectly flat) prism together, however, and light passes through the inside face to emerge from the second prisms with no change in direction (see paths 1 and 2 in the attached image).
N & S placed 2 prisms nearly, but not quite, together. According to ideal, classical (Newtonian) optics, light should be reflected as if the second prism were not present (path 3). However, in the real, quantum-physical world, a small (but rigorously calculable) fraction of the light passes straight through both prisms, as if they were not separated (path 4). This occurs because, in quantum physics, each particle of light (photon) is not a classical particle, with a definite position at any given time, but a distribution of probabilities of the particle being detected at a particular position at a given time. There is a definite, non-zero probability that a photon having entered prism 1 will, a moment later when classical optics predict it being reflected by the prism’s interior face, be located inside prism 2, allowing it to continue in a straight line as if the prisms were together. This “doing the classically impossible” is known as “quantum tunneling”.
If you put a detector in the gap between the two prism, you would not detect a photon tunneling between them (path 5) – this effect is not a photon violating the laws of classical refraction and reflection, but realizing a less probable, but not impossible, quantum statistical outcome.
The faster-than-light character of this experiment comes from the fact that this not-in-prism-1-but-instead-in-prism-2 tunneling effect is not movement in a classical sense, so doesn’t require any time to “leap” the distance. So, if as N & S appear to have done, you precisely measure the travel time of the tunneling photons, it will be briefer by at least the amount of time required for a non-tunneling photon to travel across the gap between the two prisms. So, the time required for light to travel from an emitter to a detector may be shorter – as N & S have demonstrated – for a photon passing through a couple of separate prisms (path 4) than through air, or possibly vacuum (path 6).
I can find no fault with their reasoning or experiment, and am revising my previous opinion that this experiment is not much different than previous “faster than light light” demonstrations involving
group velocities. From this, I think it's reasonable to conclude that one could construct from ordinary materials, such as glass, an long optical path, such as a fiber-optic communication cable, capable of transmitting a signal faster than a direct EM signal, such as a visible light or radio beam.
I believe, as N & S and reporting journalists suggest, this really is FTL communication, and really does violate predictions of Relativity. When confronted with such a paradox, one should be mindful that, despite being a modern theory, Relativity is a classical mechanical, not a quantum physical, theory, so, in an absolute, objective-reality-based sense, is wrong when applied to experiments of this kind.
A better article 
I change my mind – N & S really have demonstrated FTL communication 
