The following post contributes to an ongoing debate about the role of open-source software in research (see Gezelter, Krylov et al. and our own post). The following comments reflect the personal opinion of John Herbert and are not necessarily an indication of the views of the other authors of Krylov et al.

By Professor John Herbert, Department of Chemistry and Biochemsitry, Ohio State University.

I want to provide a rebuttal to a few of your points [raised in your post], one factual and two others that are matters of opinion.

First, there is one place where I feel that you’ve (perhaps unwittingly) made a factual error. Namely, you state that we are "pre-emptively" arguing against a mandate by the funding agencies, when (to your knowledge) "no such mandate exists". Please see Ref. 1 in our Viewpoint, where we provide links to three different major funding initiatives in the US for which the program solicitations explicitly come with an open-source requirement. These are things that are explicit and citable, but "off the record" both Anna [Krylov] and I have been told by various DOE and DOD program officers that essentially all program calls from those agencies are biased in favor of open source. Our piece is not pre-emptive at all, it’s reactionary. Specifically for this reason, Anna  [Krylov] and I have talked about writing such a piece for several years now, and were finally prompted to act by the appearance of Gezelter’s Viewpoint.

On the opinion side, you agree with us that students and postdocs should not be charged with software maintenance, rather this should be done by professional programmers but the current funding model makes it difficult to hire such people. We are 100% in agreement on this. However, your solution seems to be: change the funding model. Ok, long-term this is good, and let’s work for that kind of social change, but you don’t give any specifics about what a new model would look like and without such, this argument is disingenuous. It’s the same kind of wishful thinking as can be found in Gezelter’s piece, and for which we explicitly called him out in our Viewpoint. We are offering an imperfect but practical solution, namely, the commercial model. My overall view is that the open-source community has failed to offer realistic near-term solutions to what we all agree is a problem.

My second subjective point regards the meaning of the term "open source". You’re not the first to take exception to our Viewpoint on the grounds that the term is defined explicitly on opensource.org. One person, who I do know personally and respect a great deal, was irate that we had referred to GAMESS as "open source" because the GAMESS license explicitly forbids redistribution and therefore, apparently, does not conform to the aforementioned definition. This was interesting to me because I didn't know about this objection, but it's further evidence of the point that Anna [Krylov] and I tried to make, which is that the term is not used consistently. There are many, many people in the computational chemistry community who would use the term "open source" to refer to GAMESS, because you can get the source code for free, even if you’re not supposed to redistribute it, and can only commit changes if Mark Gordon and Mike Schmidt say so. Similar comments apply to NWChem and CP2K. I suspect this may be something that rankles the open-source purists, but the fact is that just about anything that can be downloaded as source code, for free, gets referred to as "open source" in my community. That’s a very heterogeneous set of licensing agreements, and that was our point.