July 30, 2010

Long fuses can turn nice risks embarassing

Silly signsThe paper Long-term impact risk for (101955) 1999 RQ36 by Milani et al. has caused the usual we're all doomed, maybe, in the future reactions.

Asteroid risks are among the "nicest" global catastrophic risks. They follow fairly predictable orbits that can be extrapolated far in advance, and the only kind of uncertainty is parameter uncertainty of exactly the orbit, the size of perturbations like the Yarkovsky effect and of course being aware that a risky asteroid is there - there doesn't seem to be any unknown laws of nature happening here. Even better, asteroid sizes follow a pretty steep power law and the danger from an impact becomes significant just for fairly large impacts. While we might be suffering from an anthropic shadow if we look at impact craters we have astronomical observations on other bodies and of free space NEOs to callibrate our risk estimates.

The interesting aspect of 1999 RQ36 is the relatively high total impact probability of 1 in 1000 given current data, and the far data of highest collision risk, 2182. In this kind of long-term prediction getting accurate data in the near future might not reduce far future uncertainty: even if the current orbit is perfectly determined, the uncertain Yarkovsky effect will make the 2182 prediction uncertain.

Another aspect is that deflecting the asteroid away from the "keyholes" it needs to pass through during a series of passes 2060-2080 is relatively easy (it needs to be moved about one kilometer to the side, within the technology we have today), but afterwards deflection becomes much harder. This suggests that there is a (soft) time limit to fixing a potential problem. The authors note:

The current impact monitoring covers a time span of about a century. If this were to continue, and the asteroid 1999 RQ36 were indeed on a collision course for 2182, then the warning about this would be issued only in 2082, that is at a time when the opportunity to deflect with requirements compatible with current technology would already be expired. To extend the predictability horizon of impact monitoring seems to us to be a better solution, at least economically, than waiting for future and hypothetical technological advances.

But extending the predictability horizon may be very hard for some asteroids; this one probably is well-behaved enough that knowing its thermal properties would be enough to tell whether it ought to be deflected, but it is not impossible that there are others that are too chaotic to be predicted far in advance yet end up with hard-to-shift orbits.

Of course, as a technological optimist I have full confidence in that we as a species could develop good asteroid pushing technology if we felt we had to. It is just that this motivation probably declines exponentially with time distance. If 1999 RQ36 was going to hit in ten years the effort would likely be massive, but if we can delay any effort (and costs) to 2070 or so we will likely do it. I expect technological capability to grow exponentially over this time so maybe this is for the best. But the logic of procrastination may well be strong enough to avoid doing anything by 2080 - after all, there is a century of time left. Then a decade, then a year and then nothing at all.

However, delaying can be smart if the necessary technology becomes cheaper and more groups have access. It only takes one group to jostle the asteroid out of the dangerous orbit to fix the problem. The more groups who can act, the more likely that one might do it (unless there is a very strong dilution of responsibility effect). If lifespans increase we should also expect a lower discounting of the far future, although this might still not be enough. We might even become better at avoiding collective procrastination, although so far we have no historical data to base this hope on. And of course, we are likely to be richer and hence better able to handle sudden costs or the need for building resiliency.

Global catastrophic risks with a very long fuse might be predictable but hard to fix even when the cost of doing it is low. It would be embarrassing if we suffered from such an event that had been predicted more than a century before.

Posted by Anders3 at July 30, 2010 10:46 PM