"Think of the climate as a small boat on a rather choppy ocean. Under normal circumstances the boat will rock to and fro, and there is a finite risk that the boat could be overturned by a rogue wave. But now one of the passengers has decided to stand up and is deliberately rocking the boat ever more violently. Someone suggests that this is likely to increase the chances of the boat capsizing. Another passenger then proposes that with his knowledge of chaotic dynamics he can counterbalance the first passenger and indeed, counter the natural rocking caused by the waves. But to do so he needs a huge array of sensors and enormous computational reasources to be ready to react efficiently but still wouldn't be able to guarantee absolute stability, and indeed, since the system is untested it might make things worse.

So is the answer to a known and increasing human influence on climate an ever more elaborate system to control the climate? Or should the person rocking the boat just sit down?"