The current standards of scientific collaboration and publishing are fundamentally broken, and are actively hindering progress. Fortunately, the world of software engineering has already come up with a solution, version controlled systems. We build tools to facilitate transitioning from the obsolete journal publishing model to a modern, versioned and open collaboration ecosystem. We do this by publishing research on github.
At its core, the current model consists in sharing static text documents published periodically after an opaque peer review which happens behind closed door. This model worked well in the pre-internet scientific academies of the 1900s, but we can now do better.
Much like writing scientific papers, writing and publishing code consists in a coordinating large group of people to edit a set of files which evolve over time. Software engineering solved this problem with git, and collaborates through websites like github. We can do the same for research publications and datasets.
No need to try and hunt down an article's authors to troubleshoot a protocol, just publish an issue in the repository and anyone can try to help
Did you find new results which invalidate an old conclusion? Open a pull request to bring everyone up to date
No need to wonder if the paper you are reading is the latest one on the subject, you'll always see the latest document and all the iterations it took to get to it
Negative results are valuable, find dead ends before you run into them and make sure no one else falls into the pitfalls you did.
Any collaborators will be automatically acknowledged and their contribution to the repository tracked.
Any updates or issues are discussed publicly, ensuring the authors don't get flooded with the same questions.
