In the long comments thread below, Ken Parish makes the point that the observed rate of global warming over the past few decades has been about 0.2 degrees C per decade, and that the IPCC and others (including some Ken has cited with approval) suggest that about half of this is due to human activity, mainly the emission of greenhouse gases. The rest reflects factors such as the cyclical fluctuations that led to warming in the first part of the 20th century and cooling in mid-century. (There are both higher and lower estimates of the share due to global warming, but half seems reasonable, especially when we compare the recent rapid warming to the more gradual warming and cooling observed in the past).
The simplest way to interpret these numbers is to suppose that there are no lags, sinks or feedbacks, so that current emissions feed directly and immediately into increases in global concentrations of greenhouse gases and then, also directly, into higher temperatures. On this basis, a continuation of the late 20th century rate of emissions would imply a temperature increase of 1 degree C, which seems unlikely to cause major disasters.
But of course, under business as usual, emissions will not remain stable. If no mitigation policies are adopted, it seems likely that emissions will at least double over the next 50 years suggesting warming of at least 2 degrees C.
Now lets think about lags, sinks and feedbacks. To the extent that lags and sinks are important, and that sinks are gradually filled, the final impact of CO2 emissions will be greater than the initial impact. Thus, if emissions stabilise at a high level, temperatures will keep rising. This again supports a higher estimate of damage. It’s hard to estimate, but I’d suggest that this supports a baseline business as usual estimate of around 2.5 degrees C.
Feedbacks can go in either direction. The IPCC argues that positive feedbacks will dominate, critics such as Richard Lindzen disagree. I find the arguments too complex to resolve, so lets call it an even-money bet that feedbacks will either enhance GW by 50 per cent or reduce it by 50 per cent. On this basis, it’s about equally likely that the impact will be 1.25 degrees C (not too bad) or 3.75 degrees C, enough to wipe out a wide range of vulnerable ecosystems. (This is spurious precision, but the range of possible outcomes is what matters)
The basic question is whether to start preparing for the bad scenario now (Kyoto) or to wait for more information. If we start preparing and it turns out that things are not as bad as we thought, we incur a cost (about 0.5 per cent of world income on most estimates) for no benefit. On the other hand, if the news is bad, we will be well placed to take the action needed to prevent a disastrous rate of warming.
If we kill Kyoto (negotiated over nearly the whole of the 1990s), wait ten years and then start negotiating an alternative when we get bad news, we will have lost 15-20 years at least, and possibly more. The cost of achieving any given level of atmospheric concentrations will be greatly higher, and the amount of unavoidable damage much greater.