Science is open software

Modern computational science depends on open source software to achieve reproducibility, reliability, and collaborative progress, making open source a necessary condition for doing science in a computerized world.

Key Points

• Science requires testable, systematic predictions; computational reproducibility lets readers embed ideas into their own inner models and build on them — software is how we encode and share predictive models.
• Scientific results depend on software; if the software is wrong, the science is wrong — bugs have already caused retractions in climate science, immunology, and other fields.
• Open source provides both reproducibility (executable, modifiable code) and reliability (public scrutiny, iterative improvement) — the same principles as the scientific method applied to simulation.
• A vision for open computational science: instantly reproducible results via browser-run containers, globally maintained models like Wikipedia, accelerated discovery through reusable runnable code, and stronger public trust via transparent, auditable software.
• Actionable steps: share and document code from day one; use reproducible environments like NixOS for long-term stability; build on existing cross-platform tools rather than reinventing; promote academics who contribute software infrastructure.