For better or for worse, OGGM just changed the license of all our software (OGGM, OGGM-Edu, OGGMcontrib, rgitools, etc.) from GPLv3 to the more permissive BSD-3-Clause License.
It was not easy to make this decision: when we chose GPLv3, it was by conviction that all software that uses our work should also be free software, hence the choice of a copyleft license. There are plenty of vivid debates online about which license is “more free” than others, and we won’t start a flame war here.
After some thought, we (well, mostly me, but I was supported by most of the OGGM contributors) decided to use BSD3. We had quite some discussion on github about this, and I am going to summarize the main points here:
The OGGM projects provide a framework from which many models and other software can benefit from (OGGM-Edu is one of them, and we hope to engage more models, such as the MIT licensed SERMeQ and PyGEM projects). With the GPL license, we would enforce our license to these downstream projects (or, worse, limit the endorsement of OGGM because of license issues). With BSD3, including OGGM code is as easy as adding our copyright notice to your repository.
The GPL license is very complex and wordy. BSD (or MIT) is simple and clear.
Even with a GPL license, there is no guarantee that modifications to OGGM will be made freely available. With a GPL license, outside modifications of OGGM could not be shared openly unless shared under GPL as well (that’s why we chose this license). However, single authors or organisations could take OGGM, make it better, and publish/produce data based on their modified version of OGGM without having to share it to others while still being compliant with GPL (as long as their modifications remained private within their “organisation”). This problem remains with a permissive license like BSD, but at least the license is now simple and easy: people are allowed to do so if they wish.
After settling on MIT (which is the license recommended by Github and probably the “easiest” and most used one), I changed my mind for BSD 3-clause. The reasons are that (i) BSD3 is the license chosen by many packages in the scientific python ecosystem, and by many packages we love (numpy, bokeh, scipy, holoviews…) and (ii) because the “third clause” (non endorsement) seems appropriate in the context of academic citations and publications. Indeed, although we encourage people to use our model, we don’t want our names to be used to promote bad science (e.g. climate change denialists).
In the “copyleft” VS “permissive” license debate, some people share the opinion that software which is mostly funded by tax payers (this is the case for OGGM) should be available under no restriction to anyone, i.e. this is an argument for the permissive license as well.
That’s it! Not a big deal for many, but still important in my opinion. We hope that you agree with these changes, and we hope that OGGM will continue to grow in the future, regardless of which (open) license we are using!