Aargh, aargh, aargh.
A seemingly intelligent software guy just doesn't get it about averages and the Central Limit Theorem.
He has a germ of a point, in there somewhere, together with a little kernel of misunderstanding, from which he has constructed a towering monster of semiotics, a boggling distinction between Real Things (e.g. temperature (!), muon half-life, light curves of variable stars), and Other Things (e.g. mean temperature (!!)). This distinction allows him to treat observations of Real Things (e.g. a million measurements of the decay times of a million different muons, or a thousand estimates of the apparent magnitude of a particular variable star at different times) differently from observations of Other Things (e.g. sixty temperature measurements at a particular weather station in September 2009).
The frustrating part, as so often with Someone Wrong on the Internet, is that he clearly has the wit to grasp the explanation, but he's not going to accept it. Even if he consulted an expert statistician, and was patiently led through his error, he would write her off as an idiot and go find another expert statistician instead.
I suspect he even thinks there must be something wrong with his argument, because if it was sound then *all* climatology (not just GISTEMP) would be entirely without any foundation (as, for that matter, would be weather forecasting). And yet he works very hard (see other posts on his blog), trying to find more and more things wrong with GISTEMP. If that post held water, he could refine it down to a precious one-page gem, publish that, and rest on his laurels (with his Nobel Prize).
Anyway, he's got me pegged as a conspirator now, and tells me I'm interested in something called "strategic inside baseball" (I have no idea what that is). Hey ho, there'll be another one along in a minute. In related but better news, see the next post.