Each morning, as soon as I arrive in front of my computer, I generally spend a couple of minutes browsing articles of the day on arxiv.org, in particular in the "Numerical Analysis" section (and sometimes on Computational Physic, but that's when I'm in tramping mode). This systematically browsing might look a bit fastidious, but I like to have an large overview of the newest developpements.
In that topic I came across an article named "General Complex Polynomial Root Solver and Its Further Optimization for Binary Microlenses" which has a very interesting appendix (B, page 20) about the utility of patenting algorithms, in particular the discussion stands on the classical book "Numerical Recipes". In fact the author happened to write this article because they needed a different method to compute roots of complex polynomials than the one proposed in the NR book, because of copyright violation. So in some sense, they argue that "copyright and patent law are stimulating intellectual innovation".
But they point out a more general problem about the reluctance of researchers to release publicly their research code, in particular for Intellectual Property concerns which poses an other problem: reproduciability of [numerical] experiments, which is at heart of the scientific method. I thought that was interesting, so I decided to have a post on that topic.
Moreover, the cite the four page article "Practices in Code Discoverability". The title says it all.
Comments