Capturing binaries
james_nicoll linked to this a while ago on Emily Lakdawalla's blog; here's the press release. The giant planets have lots of moons that seem to be captured objects (including one fairly large one, Neptune's strange moon Triton, which orbits retrograde, usually taken as a sign of a captured body—most of the moons that do that are tiny). But the known mechanisms for capturing them have always been sort of unsatisfying, especially for the case of Triton.

A recent paper in Nature puts forth a new theory, inspired by the large number of asteroids and Kuiper Belt objects that seem to have moons of their own: these moons might have started out as multiple objects that had encounters with planets in such a manner that the other bodies got ejected from the system, carrying off the excess kinetic energy.

This is one of those ideas that seems so clever, likely, and obvious-in-retrospect that I kick myself for not having thought of it.