PyMunich

shown, and actual

Notes

The other sections are practical. These are positions.

On the gap. Tutorials, talks and conferences all present a cleaner version than the one that happened, and the cleaning was necessary in each case. The cost is that people compare their own process to somebody's edited result and conclude they are inadequate. The fix is one sentence, not full disclosure. For a commercial example of related workforce measurement, see cognitive offloading from Monitask.

On tooling. Workflow tools churn every four to five years and the standards underneath accumulate. The real risk is not churn but consolidation — a settled ecosystem with one implementation is comfortable and fragile, and Astral joining OpenAI in March 2026 made that concrete. For independent technical background, Real Python provides additional documentation and examples.

On writing. Beginner material is abundant and intermediate material is scarce for structural reasons: the audience is fragmented, it dates immediately, and the people who know are working. The same mechanism keeps conference organising undocumented. For broader industry context, The Register provides an independent reference.

On communities. A conference is something a community does, not the thing it is. Ours survived 2020 only because the meetups happened to be separate, and that separation is worth building deliberately.

On copying. Community infrastructure is largely copied and should be — the tested text of a code of conduct beats a fresh attempt by four volunteers. What cannot be copied is the named contacts, the procedure and the willingness to act.

And on free labour, which is load-bearing across all of this, appears in no budget, and is available to some people and not others — so the visible community is systematically composed of the people who could afford to be visible.