Fabien Maussion bio photo

I gave a talk at the “Advances in Glacier and Ice Cap Modelling” session at the IUGG General Assembly 2019 in Montreal, CA. It was a solicited contribution, so I decided to talk about something less… conventional than I would normally do at a scientific conference.


Download the slides: pdf

Preparing the talk was very challenging for me, because I was going to talk to an audience of glaciologists. Modelers, mostly, but most certainly not well aware of the Jupyter ecosystem and a very different crowd that you would expect at, say, a SciPy conference.

Lessons learned

I think that the main message went trough, but I still took a few mental notes I wanted to share here:

  • 12 minutes is too short for such a new topic! I have four main take home messages, and 2 or 3 would have been enough.
  • JupyterHub / Web-browser computing is two steps ahead already for many. It’s hard to convey the advantages of such a set-up when most people don’t even know what a Jupyter Notebook is.
  • We should continue to have low-level, entry tutorials for these tools at scientific conferences. This offer won’t be used by the “older”, established scientists but by their staff, who will then convince higher level faculty of the usefulness of these tools.
  • A vast majority of people agree with the main premise. Some know about Google Earth Engine for example, and are concerned about the proprietary aspects.
  • Most still report back that they don’t know how to start and find all of this intimidating. I can’t really argue against this argument, we still have some work to do before wide adoption.

What now?

I have spent a lot of thought and energy in the past months for this topic. The “ball is rolling”, and I have several projects going on:

  • OGGM-Edu is still at an early stage but is gaining adoption. We have some funding to move forward.
  • we have a “proof-of-concept” JupyterHub (more info here), and I’m very optimistic about it. I wrote a roadmap to keep the momentum going. The roadmap specifies how we have to use Pangeo instead of vanilla JupyterHub in the future.

I will continue to monitor and support these efforts, but I also welcome some help! Please let me know if you want to help out! ♥

Image missing Photo: Beatriz Recinos