How Does Open Source Affect CS Research?

by Marc Hedlund

I was reading a computer science research paper today in which the authors frequently referred to how Linux operated with respect to their thesis. They explicitly stated in several places, and left implicit in others, that they were assuming other Unices or even all other OSes worked the same way, when in fact that assumption was wrong in at several cases (Windows being the obvious "outlier," if you can call it that, calling their ideas about operating system security into question). I found myself wondering how much the existence of Linux as a mainstream, open-source operating system is affecting computer science research, at least with respect to operating systems?

Linux is freely available and runs on commodity hardware, suiting academic budgets well. Its source is available, so a question about OS underpinnings can easily and definitively be discovered. And as a UNIXy system, it is familiar to and comfortable for computer science academics in a way Windows is not. Together these qualities make it very attractive to a researcher trying to prove a point.

(Of course a lot of these things are not new -- look at the availability of Berkeley UNIX and others long before Linux. Yet as more and more commercial entities -- even OS vendors like IBM -- adopt Linux, its credibility as a "reference implementation" will grow. )

In general I think this is a good trend and one very well suited to, for instance, the goals of the Free Software Foundation -- providing source provides education. I hope, though, researchers remember that there are other OSes, and other ways of thinking, than the ones most readily available to them.