Science Is Open Software
Points and comments are a snapshot, not live.
Open source software is a necessary condition for computational science to be reproducible and testable.
The author argues that modern computational science must equate with open source software because science requires testable, systematic knowledge. Software encodes predictive models, and without open, reproducible code, findings cannot be verified or built upon. Citing retractions caused by software bugs, the author asserts that scientific results depend on software reliability. A vision for future science includes instantly reproducible results, community-maintained models like Wikipedia, and accelerated discovery. Practical steps include sharing code, using NixOS for reproducible environments, building on existing tools, and valuing software contributions in academic promotions.
What commenters are saying
Commenters largely engage with the central claim but diverge on its strength. A top comment critiques the title's imprecision, noting that saying "science is open source software" is not the same as saying open source software is necessary for science. Another thread argues that science is a social process of resolving ambiguity, irreducible to static digital files, and that theory-not software-is the object of science. Several commenters point to practical reproducibility obstacles: data hoarding, poor career incentives for data management, and messy "run-once" code that rarely meets the ideal of instant reproducibility. The Journal of Open Source Software is cited as a constructive initiative.