A Popeye Moment
We all have our Popeye moments.
I’m here on this site. This is a site for Smile, a Bayesian inference engine. Smile was written by the good folks at the Decision Support Laboratory at the University of Pittsburgh.
Fine site, liberal software license, recognized as excellent, fast, proven, robust software. I busily click about looking for the source code repository. Ominously the Downloads page contains only a plethora of binary tars and zips, must be a link around here somewhere for Developers.
Clicking clicking clicking … and ten minutes later, I start to get that sinking feeling and sure enough I end up here.
Here is the passage of doom. “We have considered making GeNIe and SMILE open source and discussed internally arguments for and against, doing some research as well. We have finally decided against it. So, compiled version only :-).” Later on in the back and forth it is stated that this decision was not done cavalierly. Apparently, after contemplative discussion and self reflective application of the software itself to decide its fate, Decision Theory argmax utility analysis dictated that the source code shall remain latent to the public.
My Vulcan half takes solace in the fact this decision is of rigorous method and scientifically sound. My Human half takes umbrage that the utility of the quantified public at large (me!) is so minuscule. It laments the loss of common sense methods such as plucking a daisy from the window box, and petal by petal pluck, “now we open source, now we don’t, …”. 50-50 shot with that one.
Benefit Of The Doubt
No doubt, from all appearances, the Smile folk, are good folk. Superb research. Nicely done site. Supportive and responsive in generating binaries upon request. If the rational as to why the code is unavailable were right out there in a little blurb on the Downloads page, I’m open to the possibility I’d agree with the reasoning. The intent is not to pick on Smile, but it serves as the exemplar of something that appears all too common.
Publicly Funded Research
This software was developed at a private, state-related, nonsectarian institution, where the University receives an annual appropriation from the Commonwealth of Pennsylvania. Looking at the Credits page, Smile received funding from DARAP, NASA, National Institute of Health, NSF and ARPA as well as others. All public grant monies, my money, your money.
This software was developed at a (partially) publicly funded university, led by a tenured professor, regardless of all else should not all basic research coming from this institution be publicly available?
Should scientific journals associated with field of computer science, accept for publication papers concerning specific software and algorithmic implementations without ensuring the software, core to the research, is accessible for peer review and to the public at large?
Should the departments of the federal government, DARPA, NASA, NIH, NSF, and ARPA et al., use public tax dollars to fund general research without stipulating access to the funding public to the results as a condition of the grant?
Insight please. Any one, any one?