views:

101

answers:

1

There's a big hullabaloo about hacked source code from the University of East Anglia's Climate Research Unit that Global Warming skeptics are saying proves that the scientists who wrote these IDL and Fortran programs were intentionally fudging data.

http://wattsupwiththat.com/2009/12/05/the-smoking-code-part-2/

To rebut the claim, one commenter said: I’ve been programming in IDL to produce plots for scientific papers for about 10 years and this pro clearly isn’t producing a plot for publication. The pro just plots to screen – if it were for publication then it write the graph out as a postscript file to submit to the publisher. To me, this looks like someone experimenting with the data, which doesn’t really mean anything.

Is that true? Coders make drafts and iterate their code and leave their imperfect drafts in their directories all the time. Or, is their a more insidious explanation?

I haven't worked on any open source projects or scientific research so I'd like your input on this. I was wondering if anyone in the open source community or scientific computing community has ever heard of anyone trying to intentionally program garbage output or to 'fudge' their data just to validate good data? Is this a known practice or taboo?

Can anyone think of any major projects they've seen or worked on where there's code that intentionally gives garbage output?

+2  A: 

Doing programming for scientific publications myself I can confirm that

  • one writes lots of prototypes, looking at the data from different perspective, playing with the data, making hypothesis, rejecting them, etcetera, and yes this produces tons of scripts that are later not used in the final paper
  • after a paper has been submitted, the directory with the data, scripts, prototypes, etcetera (including all "garbage output") is typically just left as it was at the moment of rushing toward the deadline

As scientists we need to publish and thus we try to present data in the most favorable way. As long as the (raw) data is provided together with the publication this is not a problem, since other scientists can come and verify the findings. Thus fudging is not the true scandal here, plotting to delete (raw) data is what might turn the leaked emails into climategate.

Adrian
I think I will try to find the full directory structure of the supposed emails and try to find emails that describe what the author of the source code was doing, as well as find the other iterations of the source code.
zilbo79
Thanks for sharing.
zilbo79