A manifesto for agile (data) science

Parentheses indicate that pretty much all science is data science, since management of data is essential and integral to the labor of scientists.

To fix the reproducibility and replicability crisis of modern science, and reduce the anxiety and bad career prospects of researchers, science needs to follow the path of software development and become agile.

Here is a paper and a presentation to justify a set of new hypotheses that would improve the situation of science. They're open to suggestions, just open a pull request.

Hypotheses

1. Reproducibility over publishability
2. Testing at all levels over hypotheses proved once
3. Open over closed
4. Stakeholder collaboration over vertical chains-of-command

Sign the manifesto

If you agree with these principles, add yourself to the list of adherents via a pull request.