On first seeing the pink and blue limits I realised at once that these were estimates of scatter of available results at various times, which is altogether different to the REAL error in the estimates. If they were real, of course they’d get thicker with age. (Like I do).

On the other hand, some eggshells in the British Museum were found to be thinner as time passed. Not all, mind you. See for example the whole paper of -

“Rhys E. Green & Jör n P. W. Scharlemann 165 Bull. B.O.C. 2003 123A
Egg and skin collections as a resource for long-term ecological studies”, from which I selectively quote:

A recent study by Green (1998) found that eggshell thickness of four species of thrush has declined over the past 150 years. This decline was evidently not caused by organochlorine pollution, because it began before the introduction of DDT and a step-like decline is not observed.

Despite observations like this, many people still believe that Rachel Carson’s “Silent Spring” book was a landmark in environmental science. Later papers appeared to show that DDT caused eggshell thinning – without explaining the observation I quote, almost as if authors WANTED to involve DDT.

In similar vein, the repetition of those blue and pink graphs has the capacity to become a landmark. This is despite them being wrong and unexplained. That is where I see the danger. Repetition begets Popularism begets Acceptance begets Belief.

Just like the Hockey stick.

In short, these guys believe their models. They use the term “ensemble” ambiguously, and yet they do not understand the most basic problem of ergodicity. Which is the problem of non-interchangeability of time-series statistics and ensemble statistics when your system varies chaotically at all time scales (Wunsch, Smith, Pielke, Hurst).

This belief is spreading and it needs to be revealed for what it is: blind faith.

I don’t know what the answer is to the latter, but, depending how you choose your denominator, I’ll bet it’s higher than 50%. Maybe even higher than 90%. Point is: there’s an awful lot of “trust” going around in this camp. Maybe the paleoclimatologists should take a closer look at what the GCMers are doing, and each of them should keep a closer eye on the instrumentalists, and so on – instead of merely trusting each other all the time.

If you look *really* closely at the GCM projections, you’ll notice that the widths are typically narrower at the beginning of the projections and wider at the end. I used to think this was because they were reasonably showing an increasing uncertainty along the time-coordinate. We’d all expect that sort of appearance if the widths reflected true physical uncertainties — which always increase with projection distance — and so we get a kind of visual cue that this is the case with the temperature projections, too. But that’s not what they’ve done. Their plots show anomalies, usually starting near zero. What they’ve done is scale the standard deviations of the numerical divergences by the magnitude of the anomaly. Where the anomaly increases, the widths of the SD’s increase, and where the anomaly declines, so do the SD widths. This is not the way physical uncertainties work. But the net result is to make the SD’s of the numerical variations look generally like projected physical uncertainties.. And in showing SDs that reflect only numerical divergence, they’re making the implicit assertion that the GCMs are producing temperature anomalies that are physically exactly accurate. I suspect all of that is deliberate

People typically make their conclusions by visual cues, and one really has to pay attention to the text to know what they’re actually showing. There’s almost a kind of plausible deniability at work here. They’re visually communicating one message, and textually communicating another.

And look at the so-called observational data. No measurement error bars on those lines, either. More incredibly, the African and SA data are represented to be as accurate as the European, Australian, and NA data. The total figure is classic IPCC nonsense, in which the climate models are represented as physically complete, but numerically noisy, and the observed surface temperatures are represented as perfectly accurate.

Bender’s reference (#28) to IPCC confidence envelopes:

The Thompson slide that says Glaciers have no “political agenda”:

I like this slide. It shows all the sites where his OSU team have drilled ice cores–and of course, kept most of the raw data unavailable for study by others: