Open source software in math

18Nov07

This opinion piece (PDF) in Notices of the American Mathematical Society, argues that since mathematicians increasingly rely on software in generating their results, we should be increasingly concerned about the use of proprietary math software, since (a) not everyone has access to that software and (b) the code used by that software is usually closed-source, so the algorithms used can’t generally be evaluated.

I agree. It’s the same reason why I (and other scholars) don’t like locking my information into proprietary file formats — academia has a tradition of free exchange of information, at least in the sense that academics almost always disclose all of their methods and results.

At the same time, scholarly communication has traditionally taken place through, among other channels, publications in expensive academic journals. If you’ve ever tried to get an article on Science Direct without being hooked up to your campus network, you know that this information is not “freely” exchanged. Often, an individual article goes for about $30.

Increasingly, though, academics are using the internet to facilitate free exchange. Working papers are published on the web. Entire textbooks are posted online for free. And academia is probably one of the richest sources of open-source software.

But as the article also points out, good software is costly to make, and it can be difficult to make money on open-source software. So if commercial programs are easier to use or more powerful (as they often, though not always, are), it seems natural that many serious researchers will want to use them.

So what are we to do? I suppose that, as long as results can be checked in free software, there isn’t really a problem. Everyone is entitled to the information, but it seems reasonable that people with funding should be allowed to access that information with a slicker interface. After all, I think everything you can do in MATLAB you can also do in SciLab, and everything you can do in Stata you can also do in R, though you might need to invest a substantial amount of time and knowhow. Still, there is something unappealing about the widespread use of materials that, to some, are prohibitively expensive, but that’s really nothing new. After all, the armchair particle physicist doesn’t have access to a supercollider.

Advertisements



%d bloggers like this: