John Mashey is a well-known and well-respected greybeard in the computer business. It was a treat to read his many articles on comp.arch. He has forgotten more than I will ever know. But such flattery needs to be tempered by his forays into the global warming debate.

He has posted many comments to http://www.realclimate.org/, generally making fun of the people outside the IPCC clique – those who dare question the work.

Some year and a half ago, he opined, verbosely but wrongly, that releasing all the data and the code used to produced publications by James Hansen’s group at NASA is somehow inappropriate or wasteful. Here’s one, responding to this comment. He seemed to assume that no software + climate expert promised to review the entire code base of whatever widget the NASA people were running. Therefore, he argued, no one else should feel entitled to review anything at all.

Bollocks. That is a ridiculous standard. Nonspecialists routinely contribute meaningfully in open-source projects. Nothing even close is done in the esteemed peer review process that is supposed to give such authority to the published mainstream papers.

His most recent entry is just as obtuse. Here, he’s defending the typical practice of climate researchers to NOT release the snapshots of the data they used in particular published papers, even though many of the data sources (historical temperature tables) are frequently – and retroactively — updated. How warm was it at historical sensor X last year? Well, that depends on when and whom you ask… scary!

Anyway, here is a guy who knows all about software (and hardware). He even comments about how software version control solves many problems. He still seems to be so invested in the position that the mainstream climate people should not have to stoop to fully disclosing their methods that he goes back thirty years in some company history to find a time when some researchers couldn’t be bothered to do version control. And that was swell then, so it must be swell now.

Bollocks and shame. It should be obvious to any software practitioner that simple source control of the changing data sets would make it trivial to both track history and propagate changes. And he should know full well that simple version control is now trivially cheap. (There exist many free public web services to do it if one can’t even host a repository oneself.) There is no excuse for not saving data except negligence … and encouraging enablement by folks such as Mr. Mashey.