Information and energy?

Is information energy, and can you measure the energy of information? "Experimental protocol for testing the mass–energy–information equivalence principle" by Melvin Vopsona says maybe. SciTech Daily linked to an article about this, or I'd never have heard of it. The premise is that information carries a certain amount of energy, over and above the energy of the object.

Were that true, and if an (e.g.) electron carried some information about itself, then he proposes that electron/anti-electron annihilation should release that energy in the form of a couple of infrared photons in addition to the familiar .511 MeV photons. To be fair, he recognizes that the presence of extra photons is a strong assumption, though I suspect he doesn't realize how strong: considerations of spin with extra photons would turn the .511 MeV emissions into a spectrum--which we don't see.

I'm not quite clear what sort of self-information he's talking about. Things like spin direction are relative to other things, and it's rather banal to note that the relationship of an electron to its surroundings will involve a certain amount of energy.

He estimates how the mass of a 1TB disk drive would change if the information were erased (far too small to measure), but what is information? If your compressed x-ray telescope data (which looks pretty random thanks to compression) is overwritten by my home movies (which are also compressed and whose bits look pretty random), you would say information was lost and I would say it was gained. It looks like a naive definition of information is relative too. If all the bits were in one direction, you could get different energies than if they were random, but that's not what Vopsona is talking about.

Is the premise crazy? Maybe not. This approach won't test it, though.