Mozilla Science Lab, a branch of the Internet nonprofit Mozilla, is running an experiment to see if review processes could up the quality of researcher-developed software, Erika Check Hayden reports this week at Nature News.
While many researchers develop and use software in their work, most have not received "formal training in coding best practice," Check Hayden writes.
This has led to a variety of issues, including some high-profile incidents — for instance, she notes, some have suggested that the fraudulent research of Duke University scientist Anil Potti "would have been exposed much earlier" had he "been compelled to publish his data and computer code along with his original papers."
Less scandalous, but perhaps even more alarming, Check Hayden writes, are the instances where researchers presumably operating in good faith are led astray by poor coding practice. She cites the example of Scripps researcher Geoffrey Chang, who in 2006 had to retract five papers on crystal structure "after finding a simple error in the code he was using."
In their experiment, the Mozilla researchers have subjected nine papers from PLOS Computational Biology to the sort of code review typically given to commercial software prior to release. A preliminary report on their findings is expected with the next few weeks.