Wednesday, October 13, 2010

Is it in their Nature to lie?

In his real life, your humble Devil is a Product Manager for a small software company. Given that it is a small software company, your humble Devil actually delves into the methods and programming of said software.

As such, I know a little about how software programming works, and what is considered acceptable and what is not—both by the programmers themselves, and by those performing the "acceptance tests".

Having established some vague credentials, I would like to draw your attention to this article in Nature—as highlighted by His Ecclesiastical Eminence—regarding the ClimateGate data releases last year.

As most people will know, most of the forensic fury was focused upon the emails exchanged between the key players in this fraud, but a few people started delving into the data that was released alongside those communications.

In fact, your humble Devil highlighted a large part of this in my collation of comments around the HARRY_READ_ME.txt file (a post that resulted in over 24,000 absolute unique visitors in one day).

What this file displayed was not what Nature dismisses as "wonky code", but an utter failure of any kind of systematic programming ability, plus a total lack of verification and testing.

As far as I—and, I am sure, most programmers—are concerned, the construction of models based upon such obviously inaccurate software is tantamount to fraud. Regardless, Nature does not agree...
When hackers leaked thousands of e-mails from the Climatic Research Unit (CRU) at the University of East Anglia in Norwich, UK, last year, global-warming sceptics pored over the documents for signs that researchers had manipulated data. No such evidence emerged...

Where to start? The fact that there was far more data than the HARRY_READ_ME.txt file to examine, and I hadn't the time to collate the results—if anyone can donate links to those who did, please leave them in the comments.

But the HARRY_READ_ME.txt is enough: it details the lack of raw data, the rough estimates, the use of rainfall as a substitute for temperature, the use of synthetic data (i.e. "data" that was made up to fit the climatologists' prejudices) and any number of other really poor practices.

Are they fraudulent? Maybe not.

But the fact that the software programme created by Harry was used to construct the next lot of models—despite the fact that the file existed and that it is inconceivable that Harry didn't tell his employers what a fucking massive pile of shit it was—most certainly is.

These people knew that the software did not operate according to specification, but they used it anyway. FAIL.

These people knew that much of the original data was missing, corrupted or faked, but they used it anyway. FAIL.

These people knew that, together, these factors would produce results that were incorrect. FAIL.

These people knew that, regardless, the software would produce the result that they wanted. FRAUD.

But the killer comment is made by Bishop Hill...
Now correct me if I'm wrong, but none of the inquiries actually looked at the computer code, apart from there being a brief word from Tim Osborn in evidence to Muir Russell, denying that the bodges he'd mentioned affected published results. I'm pretty sure the Harry Readme was not looked at by any of the inquiries.

You are not wrong. None of the "independent" enquiries looked at the code, and this was for the same reason that none of the media rebuttals mentioned the code.

The reason that it was only the emails that were mentioned was that they had some kind of plausible deniability. Excuses were wheeled out, along the following lines...

"Oh, don't worry! Scientists are always having little spats. These were personal emails, not intended for release."

Well, we know that they weren't intended for release because the scientists in question were all urged to delete data and emails to prevent them being released under FoI.

This was to ignore the fact that the data had been examined—the code had been examined too. And from looking at those files, there were only two conclusions to draw:
  1. the climatologists were deliberately defrauding the community about their results (very likely), or

  2. the climatologists were so fucking incompetent that their data and results mean nothing at all (even more likely), or

  3. both.

Either way, there is simply no way that we should be restructuring the world economy—and, by the by, killing fuck-loads of poor people—on this evidence.

Of course, facts, logic and science are seriously unlikely to trouble the idiots at Nature—they might lose some of their share of "the money flood"...


chris edwards said...

I wonder how many of our tax pounds were paid to this cowboy for this crap? surely that is fraud, as is using it to hike taxes? I know no words low enough to describe these cheats.

Sean O'Hare said...

I have been a software developer for over 40 years, much of that time for various sofware companies and consultancies. Much of my work has been in the fields of defence and engineering.

For interest I started an unofficial quality audit on the ClimateGate code as soon as it was released and I can only confirm that it is totally awful. My first reaction was that the code I was looking at couold not have come from the sophisticated and complex models we were led to believe that we have spent millions developing. It appears to be the work of an undergraduate working alone in their spare time.

I was expecting to see use of advanced mathematical libraries but all there is are 'A' level statistical algorithms, and as you observe even these appear to have been fudged to massage the data.

Twisted Root said...

This link has all of the documents and analysis of the emails,but I can't see any analysis of the data and software.

Andrew said...

I have a copy of the UAE computer code:

Results desiredResults = UAE.GetDesiredResults();

Results actualResults;

Program program = UAE.GetLastResearchersProgram();

actualResults = program.GenerateResults();
while (actualResults != desiredResults)


Brian H said...

This is, of course, of a piece with the refusal of the CRU-Crew to take input or analysis or critiques from competent professionals in any other relevant area: physics, oceanography, hydrology, modeling, forecasting, chemistry, etc. Jackasses of all trades, masters of none.

Anonymous said...

I'm not a software quality expert - but I'm an administrator who over many years had responsibility for software quality in large organisations - I still sit on a government advisory body in this field. I read the HARRY-READ-ME file too - and reached the same conclusions as you do. I wrote to Prof Acton, the UEA vice-chancellor - making many of the points made in this post and preceding comments. I urged the appointment of someone to the Muir Russell enquiry with some sort of standing in software quality. I got an emolient response.

As far as I can see, not only was no such person appointed but no attempt was made even to investigate the questions.

The quality and accuracy of the modelling of past temperature data is absolutely basic to the man-made global warming narrative: it is barely believable that the Russell enquiry either didn't realise this and/or didn't bother to investigate the evidence.

Roger Vaughan Carr said...

A wisp of smoke yet lingers at the end of the barrel which is the HARRY_READ_ME files… or gun ─ so why is it not the subject of forensic debate pressed by all who doubt the integrity of both the scientists and the science emanating from the East Anglia CRU and flowing into policy worldwide?

Anonymous, above, comments: “I urged the appointment of someone to the Muir Russell enquiry with some sort of standing in software quality. I got an emollient response.”

So even the Muir Russell inquiry was alerted and chose not to hear?

Thank you for publishing this essay, Devil’s Knife. Perhaps it will inspire demands that this matter is pursued much farther. At the very least it keeps it alive.

Francis Turner said...

John GC writes similarly here - - John followed up on the initial work I did but since he's a much better programmer did stuff quicker and better

vervet said...

"Barrage ditched as nuclear plants get green light"

Oh happy day ! - this is going to piss off the green meanies big time.