It takes little imagination to visualize the extreme political and social controversy if we wait for a specific impact to become known before developing the path deviation criteria.I wrote transhumanist polymath Anders Sandberg about the potential problem and a conversation ensued. Here's the thread:
The opportunity for abuse and the underlying human characteristics that concerned Carl Sagan (when he reflected in 1992 on potential "negligence, fanaticism or madness") still remain a challenge while the instantaneous impact point is slowly guided off the Earth. This deflection dilemma arises in the recognition that if one can deflect an incoming NEO such that it misses the Earth, one can as well deflect a NEO that would otherwise miss the Earth such that it now hits the Earth, presumably in a particular location.
This is clearly a challenging task since the few space faring nations of the world are powerful, and such intrusive oversight will not be easily negotiated. It is crucial to realize that the appropriate time-window to negotiate these international agreements is not the time available until an impact occurs, but rather the time available until an impact is predicted.
Me: Here's another consideration for a nihilistic human extinction scenario:
The deliberate engineering of an asteroidal impact: "Tugboat as Lifeboat?"
Anders: As my gaming group remarked as they debated asteroid braking and mining strategies in one of our games: "Don't even think of it! Aerobraking asteroids easily becomes lithobraking!" :-)
Like weather control attempts, asteroid deflection attempts may be subject to huge risks of liability claims - if the defelction does not work 100% somebody is still going to be hurt, and survivors may claim that they suffered because of the attempt. This might actually reduce the incentive to deal with lesser impactors (preventing extinction level events will likely be accepted by everyone, even if a few small pieces are likely to drop on someone). It would be interesting to make a probabilistic analysis of the defense schemes published so far and see in which regimes (size-warning time - method) there are liability problems.
As a strategic weapon asteroids are rather bad - you put your mass driver on a NEO, and it will be in the vicinity of Earth every 10-20 years rather than all the time. Also, it is hard to discreetly send a spacecraft to an asteroid, so everybody will know you did it. Unless of course there are already plenty of space traffic, but then the deadliness of impacts may be blunted.
As for nihilistic kills, asteroids do have a certain charm. Smiting the world from the sky and all that. But I have not seen much evidence for nihilism as motivation on large scales. If you look at various atrocities and the effort that went into them, most seem to be fairly direct projects with fairly non-nihilistic (if nasty) goals. Hence I would worry more about somebody thinking the world would be a better place with a thoroughly smashed America/Europe/Asia/wherever than someone trying for extinction - specific rather than general malevolence.
Smaller asteroids could of course be used instead of nukes, but they appear somewhat unwieldy (as described above). Hence warfare with smaller asteroids is unlikely to kill off mankind, just to make us very sorry. One could imagine automated second-strike systems launching asteroids if one's home nation is destroyed causing some form of chain reaction bombing the Earth into a crater wasteland, but again that seems to be using slow and inefficient means to achieve what lots of nukes (or bioweapons) likely can do more cheaply.
Replicating asteroid mining stations building mass drivers and sending off new stations to other asteroids could set up a very dangerous system. This in addition to the *huge* arms race incentives - if you launch one of these a few weeks before me, I may end up with an immense disadvantage, so I better launch first. Even if all we do is peaceful resource extraction, we still are likely to compete for free asteroids. But such robotization of the solar system is a separate threat from asteroid impacts. The asteroids become mere tools and raw materials to the replicating infrastructure rather than the threat in themselves.
Me: One concern I have is, with all the mass movers working out there, and given the potential for someone to hack into its system, this kind of attack could be instigated remotely and even somewhat clandestinely.
Anders: Yes, if you have automated robofacs across the solar system the system can become a tremendous danger if it is insecure. It ought to be possible to make a fairly secure system: each robofac only accepts commands with the right crypto-authentification, copies are prevented from running if their multiply redundant checksums do not work out, tamperproofing of various kinds etc. But then the owner goes on a coffee break, leaving the master password written on a post-it next to his computer...
Of course, one could also use the fruits of such an infrastructure to set up asteroid defenses fairly easily. Make a few of the robofacs build interception missions, and you will have lots and lots of interceptors distributed across the solar system waiting for an emergency. But even that solution can become a problem if badly implemented (imagine a mistaken command making the interceptors attack anything leaving from Earth, and no easy way to get rid of them: a bit of a whimper).
In general, the greater the amount of mass/energy people can throw around, the greater the risks to isolated or concentrated targets.