These experiments are always about the group delay, not the information propagation speed.

Group delay can have all sorts of counterintuitive things done to it with the right setup, but that’s just an illusion.

To use an analogy: if you sweep the beam of a laser across the surface of the moon, the green dot can be made to move faster than the speed of light. However, nobody on the surface of the Moon can use this for superluminal communication because the information is encoded at the source of the laser. They’d have to relay their information to the person on Earth controlling the laser.

More aptly: to someone on the surface watching the green dots moving, they’ll seem to be moving backwards in time.

That is, an observer waiting for a dot to reach them would see a dot appear at their location, then zip away from them!

This is a bit like the sonic boom where a bullet can make a crack as it passes your head and then you hear the gunshot a moment later.

These experiments in the article are similar, except instead of a “dot” they’re tracking the summed up peaks and troughs of multiple waves of light. The final effect (and the mathematics) is basically the same.