There was more fuss in the media this week about the politics of the science of global warming. A paper by a group of researchers headed by Professor Lennart Bengtsson, a University of Reading research fellow, was turned down for publication by Environmental Research Letters. The Times had a front page headline 'Scientists in cover up of 'damaging' climate view' suggesting that the paper had been rejected for political rather than scientific reasons.
Yesterday the publishers of Environmental Research Letters fought back by publishing the peer reviewers comments in full. What it reveals is more interesting than the story itself - the sloppiness of the peer review process. Consider this bit of nonsense from one of the reviews:
"The comparison between observation based estimates of [warming] … and model-based estimates is comparing apples and pears, as the models are calculating true global means, whereas the observations have limited coverage." (my emphasis)
What on earth is a 'true global mean'? How do you distinguish it from an 'untrue' one? I don't know what makes a mean 'true' in any sense other than it is a calculation which must also be the case for 'observations'. I don't understand how an 'estimate' can be 'true' other than in the sense that the estimate was made. I have no idea what the 'truth' is. An estimate may approximate the 'true' figure but it is still an estimate!
Science as we know uses 'observations' - i.e. measurements. This review regards model 'estimates' as better than 'measured' ones. But this can only be true if the model is fit for purpose and that depends on the validity of the assumptions used in creating the model.
I am making no judgement about the right and wrong of the rejection of Lennart Bengtsson's paper but the peer review process is not revealed in a good light by this saga. There is no example here of a rigorous process. On the contrary it appears very sloppy indeed.
Yesterday the publishers of Environmental Research Letters fought back by publishing the peer reviewers comments in full. What it reveals is more interesting than the story itself - the sloppiness of the peer review process. Consider this bit of nonsense from one of the reviews:
"The comparison between observation based estimates of [warming] … and model-based estimates is comparing apples and pears, as the models are calculating true global means, whereas the observations have limited coverage." (my emphasis)
What on earth is a 'true global mean'? How do you distinguish it from an 'untrue' one? I don't know what makes a mean 'true' in any sense other than it is a calculation which must also be the case for 'observations'. I don't understand how an 'estimate' can be 'true' other than in the sense that the estimate was made. I have no idea what the 'truth' is. An estimate may approximate the 'true' figure but it is still an estimate!
Science as we know uses 'observations' - i.e. measurements. This review regards model 'estimates' as better than 'measured' ones. But this can only be true if the model is fit for purpose and that depends on the validity of the assumptions used in creating the model.
I am making no judgement about the right and wrong of the rejection of Lennart Bengtsson's paper but the peer review process is not revealed in a good light by this saga. There is no example here of a rigorous process. On the contrary it appears very sloppy indeed.
Comments
Post a Comment