Demo · A week of conferences, and what stayed with me
2026-11-04
This is a demo post, written to show how the personal pages work. Replace or delete it.
Three events in five days is too many, and I would do it again. The talks blur together by Friday; the conversations in the hallway do not.
What I noticed
The most useful question I heard all week was asked twice, by two people who had never met: "What would convince you that you are wrong?" Both times, the room went quiet for a second longer than was comfortable.
A few smaller notes, in no particular order:
- Panels work best when the moderator has read the panelists' work, not just their bios.
- Posters beat talks for anything with a figure you need to stand in front of.
- Every evaluation discussion eventually becomes a discussion about what the number is for.
The best conversations started with a disagreement and ended with a shared dataset.
Next time
Fewer sessions, longer walks. I will block the afternoon after any panel I am on, because that is when the real follow-ups happen.