This is…..big:
And impeccably sourced 🙂
It’s not beyond criticism, actually – the timeline doesn’t resolve possible rapid changes in temperature, so the comparison of smoothed paleo reconstruction with modern observations could be considered a bit misleading (see tweets here for example). But this is a relatively minor quibble on what is really a very nice graphical representation.
BlueSkiesResearch.org.uk 
He mentions the smoothness in the text and diagrams around 15500-16000 BCE. If you have an improvement to suggest, contact him. He already fixed some bugs (“Oregon” in an earlier version became “Washington State”). Seriously, drop him a line: XKCD gets a lot of eyeballs.
Yes, I did see that, it’s not the smoothing per se but more the fact that the 20th century has not been similarly smoothed. OTOH smoothing the present (as in Kevin’s 9th tweet) would be misleading as it would not properly represent where we are now and where we are heading. We know that we aren’t just at the top of a short-lived 1C perturbation and don’t think such large excursions happened in the past but this sort of reconstruction wouldn’t show them if they had. Could be considered an inherent limitation of this sort of (frequentist) approach to climate reconstruction.
Right. Could maybe be done by extending the chart further into the future, keeping the instrumental and projected temperature lines (which look like RCP 2.6, 6, 8.5?) but also showing how they would look to a paleoclimatologist at +20K. Perhaps.
I’m also disappointed by the volume of whinging from archaeologists.
Pingback: Independence day | BlueSkiesResearch.org.uk