I am in the process of updating a paper reviewing uncertainty in sea-surface temperature measurements. I’ve converted the original paper into html, separated the sections into separate pages and stuck it all on GitHub. The result can then be viewed through GitHub pages. It’s not a beautiful, typeset journal paper, but it is functional and easy to maintain. The added bonuses are as follows:
- It’s now under version control, so the change history is viewable by anyone.
- GitHub has an issues tracker which allows specific issues to be flagged, discussed and resolved. In principle, this can be used to provide a traceable account of each decision, or addition to the document.
- The issue tracker is a way that people, other than me, can easily contribute to the document.
- Issue tracking is also a form of peer review. If someone disagrees with a particular point, they can raise it as an issue.
- The repository can be forked, or cloned, so other people could, in principle, work on it themselves and then it can be merged back in to the main version via a pull request.
- Once sufficient progress has been made beyond the original, a “release” can be made. A release can then be archived and issued with a doi, which makes it citable.
Things I haven’t yet worked out about the practicalities:
- Whether this will work in practice.
- How to ensure that contributors are appropriately acknowledged. GitHub does a pretty good job of tracking contributions to specific authors. What I’m not sure of is how to go the extra step and ensuring that their contributions feed through to the citations.
- Whether there’s something better out there already that does this.
- How to make the review process effective. It would be nice if there were a larger number of contributors who are using in situ data in different ways. Off the top of my head, I’m thinking: satellite calibration and validation, validation more generally, NWP users, observing network bods, other dataset creators, people using ICOADS directly in their research.
I’ve already recorded some generic issues that need thinking about. There are bound to be others that I just haven’t thought about.