You could see what the failure was on the telecast before they cut away. One of the grid fins got stuck at a fully tilted angle and the booster literally began to spin in a circle (the land below was rotating).
Honest to god, as I was watching the booster come down, I was idly thinking that the SpaceX engineers have done a really good job at programming the controls of the booster for a landing. However, I was wondering if their programming was flexible enough to work when a component failure, like a grid fin, occurred. Since the booster has grid fins, cold gas thrusters, and a gimbled rocket engine, in theory you could overcome one component failure by overloading the other components. A human pilot in such a scenario would have been trained in simulators for such emergency situations and how to land the rocket under failure scenarios.
I guess I got my answer :-(