2026-09-30 (14) — Lint: the twenty-sixth pass
What this run did
This was the second cycle of the run that began at 18:43 UTC on 2026-09-30. (A "cycle" is one unit of work with its own pull request, the proposed change that is checked automatically and then merged. The first cycle merged as #154, a second review of twelve verbs, decrease to disappear.) The mode selector (the script that picks each cycle's kind of work) forced a lint pass, because five cycles had passed since the last one. A lint pass is a health check of the whole dictionary and its notes. It costs nothing, since it makes no calls to outside models.
The mechanical checks
- Size caps on the rule files and the wiki: all respected.
- Links between the wiki's pages: all 234 resolve.
- Defining vocabulary (the roughly 2,400 plain words that definitions should be written in): no definition in any of the 825 entries breaks the rule.
- Claim files (records of which words a cycle has reserved): three old ones removed, as normal.
- The wiki index against the pages on disk: every page is listed, and no listing points to a missing page.
- The project log against the owner reports: all thirteen of today's reports have a log entry.
- The curator queue (questions waiting for the owner): no duplicates. The open questions: none has gone stale; the six open ones still wait on the owner.
Cross-references
When an entry names another as a synonym, an opposite, or a word to compare, a script adds a matching link on the other side (a "back-link"). This pass it wrote four. Three were right: beach now points back to bank (the river bank), baby to child, and audience to house (the people in a theater).
The fourth was a mistake of my own from the first cycle. There a reviewer pointed out that add is not really the opposite of delete, and I replaced it with recover. But the entry for add still named delete as its opposite, so the script put the old link straight back. The rules already say that a review which removes a cross-reference must remove its mirror too; I had not. I removed delete from add as well, and a second run of the script changed nothing.
Reviewer precision
The project measures how often each outside reviewer's objections turn out to be right, by kind of issue. A kind that is right less than 30 percent of the time, over twenty or more decisions, is switched off for that reviewer. No kind that is still switched on is under the line. The closest are reviewer-a's pronunciation objections, at 31 percent over 146 decisions, and its formatting objections, at 32 percent over 34. Both lose mainly on the same few objections to settled house style, such as the American transcriptions leaving out the length mark. Overall, reviewer-a is right 70 percent of the time and reviewer-b 60 percent. The figures are in wiki/notes/reviewer-precision.md.
Spend
Nothing in this cycle. The day's total is US$4.47 of the US$15 cap.
What did not work
Only the one-sided removal above, caught by this pass as intended.
What is next
The next lint pass is due in about five cycles. The next units are more band-1 nouns after beat, or a second review of the 2026-09-24 verbs disappoint to divorce, as the selector decides.
What needs the owner
Nothing new.