This summarizes it well. My take:

1. the researcher gets first crack at data analysis, with the stipulation that data be archived by the end of the grant
2. permit the researcher to post "messy" data — not necessarily requiring lots of time to clean it up and thoroughly document it. It can a rather enormous task to do that. But Steve and others have asked for nothing more than working spreadsheets or raw data files, and that seems reasonable, and pretty easy (if one has nothing to hide)
3. there’s a federal metadata standard — the "FGDC metadata standard" (Federal Geographic Data Committee) that is in use, and software exists that makes it pretty easy to create files describing one’s data set. The metadata files themselves are pretty big and ugly, but nobody needs to create them from scratch. If you had an archived data set, the metadata file would tell the location (lat-long), the parameter measured, the file format, the units, where to find the data, who to contact for questions, etc. Answering questions about data can be a royal pain, and I’d suggest that NSF and others are NOT paying scientists to do that. But if one spends a little time creating good metadata, most questions are automatically answered — "a stich in time saves nine."
4. I am bothered by the elitism I see among climate scientists. As a climate scientist myself, I believe that there should be a really "big tent" here, because the presence of people from other disciplines strengthens our field. Look at what Steve and Ross have done: using rigorous statistical and evaluation processes developed in other fields, they have exposed some true weaknesses in climate science, and fundamentally changed the way we look at paleo data — for the betterment of the science. No one within the field was either (a) capable or (b) willing to do what they did. I tip my hat to you two!

I’ve been working with an electronics engineer (he has some pretty cool signal processing ideas), a civil engineer, geologists, statisticians, Ag Engineering folks … the synergy you get from multiple viewpoints is quite amazing, and these folks are showing me things in my data sets that I never would have seen on my own (or if I only hung out with climate scientists). So my message to Steve and Ross and others is: welcome to climate science — I’m glad you came to the party!

Steve: Thanks. I seem to spend more time on disclosure and due diligence issues right now than anything. Simply viewing IPCC TAR as a “scientific prospectus” is an approach that would be hard to get to without the particular experience that I’ve had, but it’s a good framework. Obviously there’s not much experience with international science assessment reports and appropriate standards of disclosure and due diligence, but there’s a lot of experience with standards of disclosure and due diligence for mining promotions and I’ve never heard anyone suggest that IPCC should apply a lower standard than mining promotions. If anything, we hear from learned societies about what a remarkable standard IPCC sets. I’ve heard a lot of what seemed to me like whining and complaining by skeptics about IPCC. So I simply tried applying the minimum standards of due diligence and disclosure applicable to mining promotions, which are relatively objective and not mere whining, and tried to see how these things worked. Lots of surprises, that’s for sure.

I’d like do some more mathematical work, showing the application of econometric methods dealing with serial autocorrelation to paleoclimate proxies. I’ve also done some really pretty work on making tree ring site chronologies using statistical methods rather than ad hoc recipes (with the byproduct of better statisitcal control of what you’ve really got in one of these chronologuies). I’d like to do something on confidence interval estimation with proper allowance for autocorrelation, but haven’t figured out how to do it properly. But I’m so overwhelmed with other pressing stuff that I hardly ever get to it. I also have some surgery to do on the other multiproxy studies. There’s nothing flashy in these studies. But people keep throwing these other studies in our faces, so it’s impossible not to respond. There will be a few more red faces.

I can understand the personal desires of scientists to hold onto data, especially if there is further work being done, but there comes a point where replicability becomes an issue. Also, privately funded researchers have no specific obligation to release data, but unless they do, public policy should never be based on such research.

Secondly, if someone doesn’t want to keep make their code public, then don’t publish articles. Get private funding and do’t publish any articles. But if your data and methods are being used in scientific prospectuses, then you’ve waived any privacy.

Also and this is often neglected, Mann’s verbal descriptions have been proven to be inaccurate. In a business situation, once you’d found problems in one area (whether it was the inaccurate listing of series used or the inaccurate description of the PC method), you would not trust the rest of the verbal descriptions. Why should you?

Yes if the scientist is receiving government money. Scientists want it both ways. Many claim objective scientific superiority over their peers who are employed in industry because they aren’t tainted by the quest for profit. But the only reason not to publish all source code is for the scientist to profit. That profit may come in the form of larger grants and/or academic prestige.

If a scientist wants to keep his source code private they should get private funding. There should be heavy strings attached to all government grants both to discourage doing science on the backs of taxpayers and to entitle the taxpayer to the use of what they paid for.