It's clear there are some, ahem, interesting personalities involved in this little spat. I'm going to do my best to avoid paying attention to personalities, he-said-she-said, details of journal interactions, Senate subcommittees, etc. It also seems clear that, whatever particular resolution of the technical matters at hand, it won't have much of an impact on what scientists believe about climate change. There's too large a body of knowledge, built up over the last century, along too many different directions, for any one study to dramatically affect the big picture. I'm not going to bother with that, I'll just focus on the technical aspects of this one little picture.

The subject is recent paleoclimate, the reconstruction of the Earth's temperature over the past few hundred years. The paper in question is MBH98 (plus a Corrigendum), an article in Nature doing a reconstruction from 1400 through 1995, which concluded that recent years are really quite warm, and pins the cause to CO2 forcing. It's not the first study of paleoclimate, nor the last, and not necessarily even the best (see here for a few others, links to review articles, etc., and the 10 Feb 2005 issue of Nature for yet another reconstruction), but it seems to be the first big study doing a spatially resolved high resolution reconstruction of the recent climate, with an interesting important conclusion, which is how you get into Nature in the first place.

The first response to MBH98 is MM03 (Supplementary Info), published at Energy and Environment, which I found at this interesting overview of various climate contrarians. MBH have an unpublished answer. Climate2003 also makes available a couple of failed attempts at getting a comment into Nature, with one answer from MBH. The latest are MM05a (another E&E article), and MM05b, to appear in GRL, also available from Climate2003.

MM03 spends a great deal of time on data management issues, and receive credit in the MBH98 corrigendum. This stuff matters a lot if you're trying to redo the calculations, or suspect bias (or fraud!). They identify and attempt to correct various problems with the data they are working with, and also redo the calculations in a different manner. After these changes to the data and procedures, they end up with substantially different results from MBH98, particularly earlier than the 16th century. Their result is much warmer, in contradiction to MBH98's conclusion. MM03 ascribes the difference to general data errors (miscopying, truncation, outdated) and incorrect principal component calculations. MM03 does not appear to ever succeed in fully reproducing the MBH98 calculations, does not appear to attempt cross-validation and verification (which takes up a good deal of the space in MBH98), nor do they attempt correlation with climate forcings. Still, it looks quite worrying: a reasonable-sounding approach to what ought to be the same data gives a dramatically different answer.

Enough for the moment, more later.