Ross McKitrick of Climate Audit has a new paper out which he discusses in depth on his blog.The paper makes the familiar denier claim that models overestimate warming.
However, the study has major limitations. It relied on the average of model runs instead of using an ensemble. By using an average, McKitrick eliminated important differences between model scenarios. It's like averaging the Rockies and Grand Canyon to claim the whole region is completely flat, so topographical maps are useless!
Worse still, McKitrick all but admits this flaw by linking to James Annan's dismissal of McKitrick's previous work — which also used problematic averages — before hand-waving away Annan's fundamental debunking of this type of analysis.