That got me wondering: what happens when you run the code? I mean, the whole IPCC Thermageddon story is based on this code, right? Running the code might be interesting.
So I ran it. "Interesting"? Boy, Howdy.
The first thing that became obvious is that this is not at all a manual process. There is a lot of operator intervention, with the program asking you for all sorts of input. So much so, in fact, that it looks like a programmer at CRU took a shot at trying to automate the process:
I'm not interested in grants, and it looks like the editor of GRL has already been properly dealt with, so I selected the first option and forged ahead.
Unfortunately, the code looks to be very fragile, with poor error handling. I'm in the middle of stack dumps, but thought that these results so far were interesting enough to post. Here's the (admittedly limited) error report that I have so far.