CLIMATEGATE AND SCIENTIFIC CONDUCT: Derek Lowe offers a working scientist’s view:

I’m not actually going to comment on the climate-change aspect of all this, though. I have my own opinions, and God knows everyone else has one, too, but what I feel needs to be looked at is the scientific conduct. I’m no climatologist, but I am an experienced working scientist – so, is there a problem here?

I’ll give you the short answer: yes. . . . A third issue I want to comment on are the problems with the data and its analysis. I have deep sympathy for the fellow who tried to reconcile the various poorly documented and conflicting data sets and buggy, unannotated code that the CRU has apparently depended on. And I can easily see how this happens. I’ve been on long-running projects, especially some years ago, where people start to lose track of which numbers came from where (and when), where the underlying raw data are stored, and the history of various assumptions and corrections that were made along the way. That much is normal human behavior. But this goes beyond that.

Those of us who work in the drug industry know that we have to keep track of such things, because we’re making decisions that could eventually run into the tens and hundreds of millions of dollars of our own money. And eventually we’re going to be reviewed by regulatory agencies that are not staffed with our friends, and who are perfectly capable of telling us that they don’t like our numbers and want us to go spend another couple of years (and another fifty or hundred million dollars) generating better ones for them. The regulatory-level lab and manufacturing protocols (GLP and GMP) generate a blizzard of paperwork for just these reasons.

But the stakes for climate research are even higher. The economic decisions involved make drug research programs look like roundoff errors. The data involved have to be very damned good and convincing, given the potential impact on the world economy, through both the possible effects of global warming itself and the effects of trying to ameliorate it. Looking inside the CRU does not make me confident that their data come anywhere close to that standard.

Read the whole thing.

UPDATE: Related thoughts here.