Er, no, he doesn't say that the main climate model is useless, or that "it's rubbish and doesn't work". He doesn't say anything even slightly like that. The particular thing he's saying is "meaningless" is a figure that has basically nothing to do with actual climate simulation.

Clearly he's not impressed with the quality of the code. Fair enough. There's a lot of very poor code around doing important things. It looks to me as if after all Ian's cleanups, he ends up with code that produces essentially the same as the pre-cleaned-up code, which suggests that the code-quality issues he found don't invalidate the actual calculations.

Another thing that's notable: here we have a lengthy, ill-tempered diary of someone's attempt to wrestle with this big hairy codebase and the big hairy pile of ill-formed data (supplied from the outside world; this isn't all CRU's fault) that it processes, and in all his notes of what he's doing and his interactions with colleagues we don't see anything like "Ran the code again. It's weird -- it doesn't predict any substantial warming at all" or "Joe told me to change X and Y to make the figures look scarier" or "Wait, this bit here is obviously fudged". This despite a clear willingness to moan about his colleagues' work.

In other words, this seems like it would be exactly the sort of place where you'd see evidence of conspiracy and dishonesty if there were any, and there is no such evidence there.