I always get a kick out of theory-of-everything papers. This title seems to cover a fair amount of ground in a short space. The public-access article is 30 pages, with two figures.
The idea they're picking up on is that maybe SETI is looking for the wrong sorts of things. Specifically, the most advanced civilizations would be characterized by machine intelligences, possibly persisting long+wide after their biological forbears had snuffed themselves. (This includes a pretty loony discussion in which the fact that humans haven't immolated themselves, 60 years after inventing nukes, is used to strengthen the argument that civilizations don't always self-destruct. Glad to hear it.) Thus the information content of advanced civilizations makes an easier (spatially more dispersed, longer lasting) target for long-range detection than biosignatures. Information storage, per se, has different constraints than carbon soup, and might be seen in different places than terrestrial environments.
But I'm still confused. I thought the SETI project was searching for nonrandom radio signals (think Leave it to Beaver, starring Wookies). Won't the dispersed machine progeny need to communicate with one another, that is, via radio signals? I can't understand which assumption of SETI is undercut by this admittedly very interesting idea.
UPDATE, 17 June: I REALLY hate to bring this in, but this discussion of Star Wars touches on the same points about civilizations on a galaxy scale. Bottom line: machine intelligences will be doing the civlizing, thank you very much. Bio-intelligences have too long generation times.