Parliament misled over Climategate report, says MP 9th July 2010 14:54 GMT

Parliament was misled and needs to re-examine the Climategate affair thoroughly after the failure of the Russell report, a leading backbench MP told us today.

“It’s not a whitewash, but it is inadequate,” is Labour MP Graham Stringer’s summary of the Russell inquiry report. Stringer is the only member of the House of Commons Select Committee on Science and Technology with scientific qualifications – he holds a PhD in Chemistry.

Not only did Russell fail to deal with the issues of malpractice raised in the emails, Stringer told us, but he confirmed the feeling that MPs had been misled by the University of East Anglia when conducting their own inquiry. Parliament only had time for a brief examination of the CRU files before the election, but made recommendations. This is a serious charge.

Full story: http://www.theregister.co.uk/2010/07/09/stringer_on_russell/

And while safety-critical software is indeed expensive to write, a lot of the problems we see here could be fixed very cheaply. We have used configuration control software in our work for years. You can download it free, and it’s very easy to use. And the basics of writing technical code with replication and reliability in mind can be taught in a couple of days. It’s mostly common sense, once you have the attitude to want to do it. Any university with a computer science department knows all this.

I think that if you had gone to the government and said you wanted to do some work getting some professional help making the software unassailable, now that it was a more important subject and the basis of policy making, they’d have agreed. It isn’t the funders that are the problem. I know the attitude of academics on this – they want to concentrate on the fun and interesting science, and regard all the nagging to follow coding standards as a wasteful distraction. They want to explore, and to play. They don’t want to spend hours of every day filling in reams of stupid forms and records, guarding against eventualities that will probably never come up.

I know, because I’ve been there. In academia, you can get away with it. But when there’s big money riding on your science, they won’t let you do that. And once you have got used to the idea, it really is much better.

Provided that there is a sufficiency of data for verification, calibration, and validation, even poorly structured and poorly documented computer code can be debugged. It’s just a lot harder. Unfortunately, funding agencies fail to understand that there is a trade off between saving money at the front end by producing such code and spending money at the back end for debugging it. They seem to feel that the cost benefit analysis favors not spending the money up front to produce high quality well documented code, particularly for programs that have limited numbers of users (we’re not talking Windows 7 or Snow Leopard here; of course, even commercial programs such as these are not fully debugged when released, Windows XP is up to Service Pack 3 and still downloading updates while Leopard is up to 10.5.8).

I agree with all of that. The software generating the results on which AGW science is based are frequently (although not always) written by students, and people who still think like students; in a rush and without any coding standards, configuration control, replicability, documentation, or proper archiving. It has been long suspected; was partially confirmed when the GISTEMP code was released (after several bugs in it had been located through laborious reverse-engineering); and now HARRY most definitely puts the cherry on top.

And had the IPCC reports, media statements, Oscar-winning films, and Nobel-winning global campaigns happened to mention that a bit earlier, I’m sure we would all have been much happier.

It’s true that code that has been hacked together can give correct results, but beyond a certain level of complexity the probability becomes vanishingly small, and HARRY is certainly testament to that.

And with regard to normal academic research, I don’t have a big issue with that. It’s an inefficient use of tax-payer’s money, but government in general so abounds with those that this is surely lost in the noise.

But it isn’t normal academic research. It is, we are told, THE END OF THE WORLD. It is the most important issue facing our generation. It involves mass extinctions, wars, famines, refugees, death and destruction. And it requires hundred million dollar advertising campaigns to drive trillion dollar decisions to restructure our entire industrial civilisation – energy generation, travel, manufacturing – and the potential denial of energy-driven development to the poorer nations just as they were on the cusp of emerging from poverty.

And people are telling me that we can’t afford to employ some software engineers to make sure we’ve got it right? That we’re going to rely on some amateur grad students knocking up some code? That issues of intellectual property, academic funding politics, claims of academic priority and reputation are going to be allowed to get in the way?

Do these people actually believe it themselves? Because I can tell you, if I was sat there staring at the end of the world on my computer screen, I’d be screaming for help. I’d be telling anyone who would listen anything they wanted to know, everything I’d got, to repeat the calculations, replicate it, prove it. Somebody do it properly, because I can’t, and it’s too important for my petty pride to get in the way. I certainly wouldn’t be flying to the IPCC or standing up in front of the TV cameras while sitting quietly on a code disaster, hoping nobody would find out. Not if the world really was at stake.

But people do the strangest things.

Anyway, I guess that’s a lot nearer than I expected to get to an acknowledgement that there might be an actual issue with the code. I’m impressed. My thanks, sir.

The scandal reported by Tom Johnson and taken up by Mooney. Johnson , said to be a graduate student, claimed to have witnessed atheists haranguing religious believers at a scientific conference.

Mooney claimed that he checked and verified Tom Johnson.

Not sure how you managed that Chris, since it was all made up. Tom Johnson is a pseudonym of the person behind the “You Are Not Helping” blog. The same person who commented here under a number if different names and was undetected by Mooney or Kirshenbaum.

So tell us Chris, just how did all that happen and you failed to notice ?

Hey, failing to display the proper degree of openness is typically in sync with HAVING SOMETHING TO HIDE like cooked books or data. Who spearheaded the major investigation of the CRU? AlGore the Carbon Hore (aka “The Poodle”)?

Oh, and its’ so fun to see scientific morphology at work. g l o b a l w a r m i n g takes on tarnish as it’s scrutinized so all media shifts to c l i m a t e c h a n g e. Too funny!

I don’t know what Mr. Nullius’ experience is in software development but he is obviously ignorant of how it is done in university environments where most of it is done by graduate students. Graduate students are not interested in taking the time to produce well structured and well documented code. Their only interest is in producing programs which can be shown to be accurate with whatever data may be available and in getting their degrees as fast as possible so they can go out and make some money. Thus, code developed by graduate students at universities is almost always poorly structured and poorly documented. Even worse, if the computer programs are at all large and complicated, they were almost certainly written by many different graduate students at different times, making them almost incomprehensible.

The fact is that, for the most part, the institutions that sponsor the research don’t want to spend the money required to develop high quality computer code or allow the PIs to hire professional programmers. This is true in all scientific fields, not just climate science.

This does not mean that the programs turn out bad results. Given sufficient data and intuition, one can determine whether the programs appear to be operating correctly, i.e. whether their predictions agree with observations. Thus far, the overwhelming consensus in the climate modeling community is that they do.