PyGambit is looking for contributors.
Recently, Ed has been focusing on writing documentation.
Get in touch with him if you're interested!
PyGambit is looking for contributors.
Recently, Ed has been focusing on writing documentation.
Get in touch with him if you're interested!
Live demo: Ed constructs a game with PyGambit! You'll have to check the video for the demo, of course.
The game is stripped-down poker. "Not to be confused with strip poker", he adds.
It's not a recent project, it originates in the 1980s!
It allows you to construct games in a few lines of code, and it computes aforementioned Nash equilibria with various algorithms
The Gambit project Ed presents is a set of stools for doing computation on finite, non-cooperative games. It is typically used by researchers.
gambitproject.readthedocs.io/en/latest/
Game theory!
You know, John Nash in the 1950s ("Nash equilibrium").
Nowadays, involving any intelligent agents, not necessarily humans.
Last talk of the day for our devroon,
*PyGambit: an open-source software for game theory*,
by Ed Chalstrey.
fosdem.org/2026/schedul...
Key message: design with the intend of public data!
01.02.2026 15:20 — 👍 0 🔁 0 💬 0 📌 0
In her project, based on DNS TAPIR, the logic is privacy-first, open source, based on local decisions.
They try to make data tracking impossible by design, which runs against many commonly held intuitions.
Today, if we work with health data for instance, people will ask you: those findings you had with small data, couldn't you get the same finding with better, bigger data?
Provocatively, Ulrika aims instead at "throwing away as much data as possible".
"Data you don't have can't be leaked!"
*Working with small data that you dare to share*, by Ulrika Vincent and Mikael Kullberg.
fosdem.org/2026/schedul...
Try the prototype:
giovannipro.github.io/wikipedia-cl...
Iolanda settled on a visualization tailored to volunteers, to identify pages where to intervene.
Her solution is based on a timeline of bubbles that you can place on the Y axis depending on a given variable.
The solution is still in the co-design phase.
Depending on the question you have, different visualizations make most sense. There is no universal solution.
01.02.2026 14:52 — 👍 0 🔁 0 💬 1 📌 0
Iolanda has spent a lot of time on Wikipedia, including as a contributor, but it is a big problem to get a sense of the landscape.
She tested a number of solutions, from networks to custom designs of all kinds.
*Visualising Wikipedia*, by Iolanda Pensa.
fosdem.org/2026/schedul...
They have a core team from many organisations and lots of contributors. If you want to learn more or collaborate, check them out 👇🏽
01.02.2026 14:44 — 👍 0 🔁 0 💬 0 📌 0
MyST in JupyterLab blends computation and narrative within an interactive environment.
Because it uses structured data it means that you can publish once and render many times! So you can refine your ideas
Their goal is to create a single toolchain to serve the entire knowledge creation lifecycle.
They used the MyST ecosystem for that!
You write in Markdown and it renders in PDF through MyST AST.
MyST Superpower:
"Simple to use, easy *for users* to extend"
Angus will focus on version 2 of Jupyter Books, released in October 2025.
01.02.2026 14:22 — 👍 0 🔁 0 💬 1 📌 0
Jupyter Book lets you create enriched and interactive multi-document computational narratives.
jupyterbook.org
Angus Hollands @agoose77.bsky.social will now present:
*Introducing Jupyter Book 2: Next-generation Tools for Creating Computational Narratives*
fosdem.org/2026/schedul...
In short: use practices that build trust!
01.02.2026 14:15 — 👍 0 🔁 0 💬 0 📌 0
- Scale your comms alongside your community. You'll need to expand outside of issues at some point. But minimize entry barriers!
01.02.2026 14:15 — 👍 0 🔁 0 💬 1 📌 0
- Pull requests: use them. Even if you are the only developer (it creates a track record of what you expect in a PR).
01.02.2026 14:15 — 👍 0 🔁 0 💬 1 📌 0
"Radically open communication"
Here are a few pieces of advice by Niko:
- Use issues as a scratchpad
=> There are perks to a *public* scratchpad because it creates *trust*
After the abstract: "release early, release often."
(as we know from Eric S. Raymond's "The Cathedral and the Bazaar")
All you need is a MVP, which boils down, as Niko argues, to:
1. At least one useful feature
2. Good documentation
3. It installs seamlessly
That last point is underrated!
If you're a scientist: write an abstract before you even start! It's productive, not as a destination, but as a compass
What's the equivalent in software engineering, asks Niko
Often you'll find a one-sentence, high-level mission statement, sometimes followed by a few additional details (eg: aims)
So, you want to make a research software package...
But how do you persuade others to use it and contribute to it?
It's a lot more about the community and trust than the quality of the software
*Trusted by design: set up your research software for community adoption*
By Niko Sirmpilatze @nikosirmpilatze.com
fosdem.org/2026/schedul...