Software code can provide important insights into the results of research, but it's up to individual scientists whether their code is released---and many opt not to.
The following letter was published in the Letters to the Editor in the December 2010 CACM (http://cacm.acm.org/magazines/2010/12/102133).
About the software of science, Dennis McCafferty's news story (Oct. 2010) asked "Should Code Be Released?" In the case of climate science code, the Climate Code Foundation (http://climatecode.org/) answers with an emphatic yes. Rebuilding public trust in climate science and support for policy decisions require changes in the transparency and communication of the science. The Foundation works with climate scientists to encourage publication of all climate-science software.
In a Nature opinion piece "Publish Your Computer Code: It Is Good Enough" (Oct. 13, 2010, http://www.nature.com/news/2010/101013/full/467753a.html), I argued that there are powerful reasons to publish source code across all fields of science, and that software is an essential aspect of the scientific method despite failing to benefit from the system of competitive review that has driven science forward for the past 300 years. In the same way software is part of the scientific method, source code should be published as part of the method description.
As a reason for not publishing software, McCafferty quoted Alan T. DeKok, a former physicist, now CTO of Mancala Networks, saying it might be "blatantly wrong." Surely this is a reason, perhaps the main one, that software should be published — to expose errors. Science progresses by testing new ideas and rejecting those that are wrong.
I'd also like to point out a glaring red herring in McCafferty's story — the suggestion that a policy in this area could undermine a modern-day Manhattan Project. All design and method descriptions from that project were top secret for years, many to this day. Such secrecy would naturally apply to any science software of similar military importance.
Displaying 1 comment