It seems that Japan’s Hayabusa probe hit a snag during a practice landing recently, abending the landing when the probe saw “an anomalous signal”. Hayabusa now orbits at 3km; there’s a good collection of images of the asteroid in question, Itokawa, at SpaceRef.
So what was the “anomalous signal”?
My random guess is that Itokawa’s material is a lot looser than expected (as in, heavier material but more loosely packed) — a great big bent-potato-shaped pile of gravel and bulldust — and that even the light thrusters used by Hayabusa were enough in the asteroid’s microgravity to kick up a little dust from a few km away — especially if the thruster exhaust is electrically charged — and change the probe’s view slightly.
This would be consistent with the “gentle impacts” explanation for the odd cratering seen on other small bodies, and with the observations of Deep Impact.
If Itokawa is indeed so loosely packed, deflecting an Itokawa-like asteroid from a collision course with Earth would present a unique problem.
Traditional proposals have relied upon either blasting an incoming asteroid with a nuke, or mounting thruster units on it and more gently steering it aside.
A pile of gravel and dust hit by a nuke would not respond as a unit; much more of the force of a blast would be consumed in heating up and rearranging the gravel than would be the case for the chunk of stone usually modelled, and a likely outcome would be to simply disrupt the asteroid so now Earth would be hit in pretty much the same place by the same mass as a large variable-density cloud of gravel rather than by a single lump. An asteroid so disrupted would be completely unmanageable (no second tries).
Similarly, mounting thrusters on a rocky “slush pile” would result in the thrusters pushing themselves under the surface and/or rapidly becoming misaligned unless they were very low thrust and had a correspondingly long lead time in which to nudge the asteroid onto a more Earth-friendly course.