Moving my content creation to a home-made Hugo site

Last weeks have been pretty intense. I officially left Equisense and started a joined PhD project between INRIA Lille and École Polytechnique Montreal. I had been preparing for this fresh start for several months and also wanted to evolve in my content creation process.

A journey from plain markdown to Hugo

I started writing articles to keep track of my learning paths on various topics, including numerical techniques, data analysis and programming. My first articles were either hand-written or RMarkdown-generated Markdown files on a GitHub repository.

As I was slowly moving from R to Python, Jupyter notebook became pretty handy to mix paragraphs, code snippets, results and charts. It also forced me to quit declaring and modifying variables, an annoying habit got from always having a REPL and text editor bundled in most scientific computing IDEs (Matlab, Scilab, RStudio, Jupyter).

Great, the articles were not centralized though but split into their GitHub repositories, you have to admit this is not the most user-friendly browsing experience. I found several blogs running on Jekyll and I decided to give it a try. For someone who is not fond of struggles with front-end side layout issues, this was a true gift, I could easily reuse templates from front-end developers and designers (special thanks for the awesome Gravity project) without much struggle and focus on what I liked: building projects and writing content.

Switching to THE writing platform

I kept maintaining the Jekyll blog until almost exactly one year ago. During that time, I was mostly writing in the context of a side-project or thinking on our journey at Equisense. This raised new requirements for the writing process such as collaborative writing, review from friends, seeing the overall picture we were sending as a team from the sum of our articles.

For these reasons, my articles gradually switched to Medium, first published as an individual, then on the Equisense page. This was a very productive time for writing as we encouraged one another and had a direct impact on the way we presented the team, how we work and our learning path: an invaluable tool to help candidates decide whether the company was a fit for them and to ease the onboading.

If Medium works, why would anyone go back to writing everything from scratch?

I really enjoy the writing experience on Medium, with some drawbacks. Medium’s design is very opinionated, that’s a part of what makes it a experience. However, leaving some choices on key topics is essential (at least to me) on the content-creation side. I believe this should be the case on any two-sided platform: be opinionated on the user-side, leave flexibility on the creator side.

The perfect example is the bright screen. It ensures the Medium experience is consistent with the unique font, background etc… But writing on a dark screen is a lot more comfortable, especially when you’re used to it or when your eyes are more light-sensitive: writing late in the evening or early in the morning was not conceivable to me on the Medium interface. The hack I used was to write everything on Atom, then paste everything to Medium once the first draft was ready, still a bit of a pain.

This might seem minor as a reason to switch, but the root behind it is more essential: Medium is a platform, you’re therefore a user, not an owner. Despite its global success, the company is still young and looking for the right model. Nothing tells me the model they choose tomorrow will be one I want to contribute to (how paid content will be managed for instance). Switching platforms for written content is a lot more tedious than choosing well at the beginning. This new step in my professional and personal life is the perfect occasion to rethink that commitment, I will still be re-publishing content to Medium, but not as the primary source.

Mathieu Besançon
Mathieu Besançon
Researcher in mathematical optimization

Mathematical optimization, scientific programming and related.