RealClimate has published their latest 2012 Updates to model-observation comparisons. I’ll take a different twist. Instead of comparing observations to computer model projections, I’ll compare them to very simple statistical projections.
In a year gone by I posted about what kind of betting terms I might consider appropriate for the reality of global warming (incidentally, that post contained an error which was corrected in an update, but the archived copy doesn’t include the update). The idea is to take annual average data (for global temperature) from 1975 through the end of 1999, then fit a trend line by linear regression. If the trend continues, then future data should probably be within two standard deviations of the extrapolated trend line. This is the “projected range” according to the existing trend. The “projected range” according to the not-still-warming theory is that future values should be within two standard deviations of the existing average (in that case, from 2001 through 2007).
I also mentioned that since it would be unlikely but far from shocking if a single future value were outside either range, I would require two (not necessarily consecutive) future years outside the range to decided against either claim — if I were a betting man.