OK, the Knight et al. paper is here, thanks folks. Clearly they have had some jolly fun dividing the runs up into trees, but the paper is a disappointment to me, as it doesn’t really deal with the main issue, which is the physical plausibility of some of the runs. It *does* talk about “Our findings reinforce the fact that variation of parameters within plausible bounds may have a substantial systematic effect…” but that rather slides over the fact that varying a parameter within a plausible range is *not* the same thing as producing a model with a viable climate. As I reported ages ago, and I’m sure its been said elsewhere, there is good evidence that low values of entraiment are phycially implausible, and those low values give the high CS values, above 9K. Kn et al say: Consistent with this, the highest predicted CSs (>9K) are all for low entcoef runs, associated with high rhcrit and ct and low vf1 (Fig2, supporting Table3) a combination indicative of reduced cloud formation. But what they don’t do is mention the Palmer stuff (maybe its not published?). As I understand it, the tests for a viable model simulation applied to the cp.net runs are very weak, and they seem to be determined not to improve them.