It has to the "best" if it's the "only"
xkcd relies heavily on one particular study of temps (Marcott et al, 2013).
Marcott was the first attempt to reconstruct the temps of the last 11000 years. I think it's still the only one. Science depends on replication of findings, and one single study is not science yet. It's scientific, but it is not "the science".
The study has been criticized. There are plenty of climate scientists who believe the earth has been warmer than it is today. Particularly during the
Late Glacial Interstitial, in which Europe and N. America unquestionably saw a period of a lot of warming. That period is not noticed on the xkcd graph, presumably because smoothing? Whatevs, it's one single study and so it gets a big "mmmmmmaybe".