FOSS & Research

I am somewhat opinionated when it comes to the relationship between academic research and FOSS (or the lack thereof). The following is mostly a rant on the subject, so please take it with a grain of salt :-)

One of the tenets of scientific inquiry is that research results should be contributed back to the scientific community through peer-reviewed publication, and that anyone should be free to take a published idea and expand upon it, use it as a building block for another idea, or even prove the idea was actually wrong. In predominantly technical areas of Computer Science (such as Distributed Systems), there should ideally be both an intellectual contribution, in the form of a peer-reviewed paper, and a technological contribution, in the form of software that others in the community can use and improve upon.

In my opinion, FOSS is the best way to produce this technological contribution, since it allows other researchers to take your work, look at how you actually produced the results presented in your paper (instead of trusting that whatever graphs or conclusions you lay out in your paper are true), improve upon your solution, and even replicate your results (reproducibility of results being another important tenet of scientific inquiry). However, in my experience, the technological contribution is often treated as a second-class citizen; something that gets placed permanently at the bottom of the TODO list or that is haphazardly put together just for the sake of telling people that you've released something, but without really treating or maintaining it as a FOSS project (with a public code repository, a user mailing list, proper documentation, etc.)

Typically (and, again, this is just my opinion), this is done because of the pressure to move on to the next paper, instead of "wasting" time on implementation (beyond what is necessary to get the results presented in the paper). Additionally, maintaining a FOSS project involves producing documentation, answering questions from users, etc. Of course, neglecting the technological contribution is not an entirely unreasonable strategy: publications count for promotion within academia (particularly for getting tenure), involvement in FOSS projects does not. However, I don't think that necessarily makes it a good strategy. I believe that involvement in research-related FOSS projects (which is, essentially, a form of "publishing software") should count as much as publications.

Thus, as a matter of principle, I try to spend equal amounts of effort on the intellectual contribution (writing papers) and the technological contribution (releasing my work as FOSS). My research code is released through the Haizea project, and my typical modus operandi is to release a stable version of Haizea, use that version to produce research results, and use those results to improve the next stable version. This is strategy that produces less papers and, thus, arguably detrimental to my academic credentials. But I believe it's the right thing to do.