Att lära sig av misstag: en artikel om felaktiga resonemang i "klimatskeptiska" artiklar

 Jeremy Brett och Edward Hardwicke som Holmes och Watson i Granada Televisions Sherlock Holmes-.serie.

En artikel har just publicerats som går igenom ett antal felaktiga resonemang i "klimatskeptiska" artiklar som har publicerats i vetenskapliga tidskrifter.

Rasmus E. Benestad, Dana Nuccitelli, Stephan Lewandowsky, Katharine Hayhoe, Hans Olav Hygen, Rob van Dorland, John Cook. Learning from mistakes in climate research. Theoretical and Applied Climatology, August 2015.

De 38 undersökta artiklarna var författade av bl a Humlum, Loehle, Scafetta, Beck, Svensmark, McIntyre, McKitrick, Michaels, Miskolczi, samt  Gerlich and Tscheuschner. Det är artiklar som i många fall har figurerat på olika "klimatskeptiska" bloggar. Vanliga fel var avsaknad av relevant kontext, utelämnande av motstridig information, bristfälliga utvärderingar, felaktigt användande av statistiska metoder och missuppfattningar om fysik. Detaljer finns i artikelns supplement.

Artikeln (Benestad et al) är lite ovanlig till form och innehåll, och den avvisades av två andra tidskrifter innan den accepterades av  Theoretical and Applied Climatology. Artiklarna som undersöks tas knappast på allvar inom den seriösa klimatforskningen, så det inomvetenskapliga värdet av den här artikeln är begränsat. Så här skrev en av de redaktörer som avvisade artikeln:

The root logical flaw in many of the papers discussed in the appendix is that showing a statistical correlation between some non-CO2 variable and some observed climate time series somehow disproves the hypothesis that CO2 is a driver of climate change. This is as silly as saying the cost of my sneakers is correlated with how fast I run and therefor I have invalidated the hypothesis that training makes me run the 100 yard dash faster. Do we really need 70 pages of text and two dozen R routines to recognize the logical problem here?
And therein lies the real problem. The climate science community has strong theory (dating back more than a century) and good, physics-based models that underly the attribution and prediction endeavors and these guide the interpretation of observations and their statistical characterization (i.e. what the null hypothesis is). If one ignores that foundation as most of the studies being criticized in this submission do, then one is left with unconstrained statistical analyses or curve fitting exercises that have no clear plausible, physically viable explanation. The reality is that many of the authors whose work is being criticized are on the record as thinking that either climate theory and/or climate models are fundamentally flawed, hence the adopt the kind of approach which leads them to conclusions that are in opposition to the vast majority of climate scientists. Again, this can be said in two sentences.
A stronger paper would show exactly how what we know based on first principles physics and more sophisticated models guides the choice of null models, statistical tests, and curve fitting techniques, and how such physically-informed choices would alter the conclusions of various previously published papers. Rather than concentrating on ignorant mistakes made in prior work, it would help to identify the key parameters in those studies and justify a priori what the value of those should be. A fair amount of that is in the submission, but it is buried within invective and verbosity (or within source code.)

Men artikeln kan vara intressant för de av oss som stöter på de undersökta artiklarna och deras felaktiga argument på bloggar och andra internet-miljöer. För "klimatförvillare" brukar tyvärr inte lära sig av sina misstag...