WUWT press-release on study of upward adjustment of NOAA temperature data

It matters in terms of statistics and machine learning. If, and that's if, these guys are correct then the data collection, pre-processing and normalization can have a large effect on the end result when you are trying to predict a trend or train a classifier. This is essential to good science and possibly more interesting to some of us than, say, a new node.js framework or a standing desk.

The argument they make for classifying the temperature sensors into different grades seems solid. There is a clear methodology extending from Leroy on how to class the sites into five different classes depending the surrounding terrain which we know affects the readings. They have used ground and aerial photography in giving the stations these classes using this methodology. So this isn't just a rehash of general Urban Heat Island arguments of the past it's a new, seemingly methodologically robust, chapter in that debate which must be answered by the other side. As it stands the best two classes of site have very good coverage over mainland USA and it's reasonable to compare them to the worst three classes of site. It turns out there is quite a large discrepancy between them[1]. This is science, not politics.

But, yes, the science has political implications. On that political side this is an issue worth addressing if we are going to put trillions of public money into dealing with the problem posed by climate change, according to the models which are operating on this very data. Redirecting those trillions has the potential to put the economic growth of billions of people living in extreme poverty on hold, no small matter. In light of that arguments like "if I, Prof. A., open-source my code and data Prof. B. will get a publication I want to write out before me" seem rather trivial. If your research says the public should spend trillions, you can expect to show your working and to be scrutinized heavily in public. On the face of it these results are huge, and I for one want to see them discussed and reviewed.

I don't think we should just accept arguments from authority - and even then, only a majority of the authorities, as there are dissenting professors from institutions like MIT and Cambridge - given the gravity of the consequences. Data processing, science, institutional incumbents versus disruptive outsiders, trillions of dollars: I think it is hitting a lot of HN notes actually.

[1]http://wattsupwiththat.files.wordpress.com/2012/07/watts_et_... And the normalization question is also at issue here beyond the different classes.