25 Research Journal
The raw, dated account of the research process — what was tried, in what order, and why — as opposed to experiments/ (structured results) or research-platform/findings/ (polished write-ups). Entries are allowed to be messy and wrong in the moment; that’s the point of keeping a journal instead of only publishing conclusions.
25.1 How to write an entry
Add a file under the current year, journal/YYYY/YYYY-MM-DD.md. No fixed template, but a useful entry usually says:
- what you worked on and why (link the research question it bears on, if any),
- what happened — including partial/negative results,
- what you’d do differently next time, or what’s now an open question.
If something clearly failed or was abandoned, add a line to dead-ends.md with a pointer back to the entry — dead ends are easy to forget and expensive to re-discover.
If something generalises beyond the specific entry — a modelling mistake worth not repeating, a data quirk that will bite again — add it to lessons-learned.md.
25.2 Entries
25.2.1 2026
- 2026-09-25 — auto-generated weekly research summary.
- 2026-09-18 — auto-generated weekly research summary.
- 2026-09-11 — auto-generated weekly research summary.
- 2026-09-04 — auto-generated weekly research summary.
- 2026-08-28 — auto-generated weekly research summary.
- 2026-08-27 — auto-generated weekly research summary.
- 2026-08-21 — auto-generated weekly research summary.
- 2026-08-14 — auto-generated weekly research summary.
- 2026-08-09 — dead end: delete-then-insert factor-value lift (reverted to the
ON CONFLICTupsert). - 2026-08-07 — auto-generated weekly research summary.
- 2026-07-31 — auto-generated weekly research summary.
- 2026-08-03 — repository restructured around a research-notebook shape (this journal included).