This is a story about the speed of light being not what we thought it was, and involving general relativity, neutrinos, and its one data point from a unique astronomical event. Oh, yeaah, riiight. And yet, it is clearly explained, and stands a good chance of being right. I am definitely going to have my weird-o-meter recalibrated.
The speed of light is the same as it always was. Any given photon may, extremely rarely, split into an electron-positron pair, and then recombine. The electron and the positron are not travelling at the speed of light, so this event will stick in a small delay. If you measure the speed of light over most human experimental lengths, this event will be very rare - so the very occasional photon will show a tiny delay. If your light travels over such vast distances that the photon may have experienced so many of these delays that it spent whole hours as electrons and positrons.
Each photon will have a random number of these delay events, so you might expect the light pulse to get blurred out a bit by this randomness. There will be a slight blurring, but because the number of events is so huge, the fractional deviation from the mean is pretty tiny.
Cute, and neat. Some posters still try and argue for gravitational viscosity, or for faster-than-light neutrinos, or that this is a failure of science and only philosophers can help us now. Ho-hum. Too little fog, too late, chaps. Better luck with the next one, eh?