[KarlInOhio]

If this is happening, then it will happen multiple times to some photons and not at all to others unless it happens so many times over the 168,000 light years that it causes a very tight bell curve. That means there should be some smearing of the wave front as the photons which didn't split into electron/positron pairs get in front of those which did split. Also, lower energy photons like radio waves should be far less likely (or even zero chance) to split and thus travel full speed all the way, while gamma rays with more energy would be more likely to split. Is there a frequency distribution with the radio waves getting here before the visible light, x-rays and gamma rays from the supernova?

Or is it possible that our theories about how light emerges from the last gasp of fusion from the supernova is wrong and it takes hours longer to make it through the star's outer shell?

[FredZarguna]

He's not postulating the existence of real position-electron pairs, which would require either two γ's or a γ and a nucleus in order to conserve momentum; he's talking about virtual p-e pairs. Virtual pair production happens to all photons, all the time, regardless of energy. You don't need to conserve energy and momentum because the virtual particles don't exist long enough to have measurable physical effects. Essentially it's a quantum fluctuation in the electromagnetic field. It has to "exist" or the path integral doesn't work, and there are [presumably an infinite number of] Feynman diagrams describing it.

See, among many places, http://en.wikipedia.org/wiki/Virtual_particle.