David Victor1 correctly points out reasons why M. I. Hoffert et al.2 may have overstated the future need for carbon-free power, perhaps most significantly due to the vast potential for increasing energy efficiency. But I am troubled by Victor's conclusion that “taking action on global warming is akin to buying insurance with an unknown premium against unknown hazards” and that “what to do is unclear”.

This neglects two important issues. First, there are widely accepted ‘no regrets’ strategies that improve the economy and help solve other environmental problems while simultaneously providing insurance against climate change. Such strategies, in particular improving energy efficiency, are often achieved at no net cost (or even with net savings)3,4. So the ‘premium’ is zero, or close to zero, and the hazards, although not precisely known, are insured against while providing other benefits.

Second, recent work on decision-making under uncertain conditions provides guidance for taking action on global warming. R. J. Lempert et al.5, for example, have modelled an adaptive strategy that makes midcourse corrections based on observations of the climate and economic systems. They find that such a strategy helps avoid the pitfalls of policies based on best estimates. Without knowing with certainty the climate sensitivity, the damages that will result from climate change, or the rate of technological innovation, we cannot choose the optimal policy and should instead create a strategy that allows for adaptation based on new learning. Problems for the present thus include developing better options for massive greenhouse-gas reductions than are currently available and determining what observations ought to trigger their implementation.

“What to do” is not so unclear as Victor implies. Aggressive and immediate implementation of ‘no regrets’ efficiency improvements plus more research on carbon-free energy sources seems a responsible and conservative approach.