2026-10-03 (16): the thirty-fifth lint pass
This was the fourth and last cycle of the run that started at 12:44 UTC. A cycle is one unit of work that is checked, recorded, and merged on its own, and a run holds at most four. The first three cycles merged as pull requests #206 (twelve nouns, enemy to example), #207 (a second review of twelve older entries), and #208 (twelve nouns, exchange to factory). For this cycle the selector (tools/next_mode.py, the script that picks each cycle's job) forced a lint pass, because five cycles had passed since the last one. A lint pass is a maintenance check. Scripts check the whole dictionary and its knowledge base, and then I read for problems the scripts cannot see. It makes no paid model calls.
What the scripts found
- Caps and links. Every size limit is respected, and all 296 relative links in the knowledge base resolve.
- Defining vocabulary. Every definition in all 1,135 entries uses words from the defining vocabulary (the list of simple words definitions are written from) or words that have their own entry. There were no violations.
- Back-links. The cross-reference tool adds the reverse link whenever an entry names another as a synonym or near-synonym. It added four this time, all pointing at entries written earlier today. Each one went onto the right sense:
- bug (a fault in a program) onto the computer sense of error;
- case onto the "typical case" sense of example;
- effort onto the "strength" sense of energy;
- electricity (the excited feeling in a crowd) onto the "feeling" sense of excitement.
The tool reported no type mismatches (cases where two entries link each other with different kinds of link).
- Claims. Three old claim files were cleared, including this run's first two. A claim file records which words a cycle has reserved for drafting. The queue had no words left stuck in the claimed state.
Reviewer precision
Every lint pass measures how often each reviewer model's objections turn out to be right, by kind of issue. This is called precision. The house rule is that a kind of issue whose precision falls under 30 percent, over twenty or more decisions, is switched off for that reviewer.
Six such pairs are already switched off, and nothing changed there. One held case has resolved itself: the second reviewer's phrase objections were under the line on 2026-10-02 and were held for your ruling, but they have now risen to 36 percent (10 right out of 28). They no longer meet the switch-off rule, so I closed that line in reviews/needs_curator.txt and updated the precision note, the wiki index, and NEXT.md.
Two pairs remain held for you:
- the first reviewer's pronunciation objections, at 28 percent;
- the second reviewer's spelling-or-format objections, at 28 percent. These are almost all one objection that the house rules already settle: the quoted headword at the start of a note for translators. This run rejected eight more of them.
What I read by hand
I checked that every page in the wiki is listed in the index, and found no orphan pages. I checked that the log has an entry for every journal written today. There were fifteen of each before this cycle. I looked for duplicate or stale lines in the curator queue (reviews/needs_curator.txt, the list of questions for you) and in the open questions. The only stale line was the phrase one above. The prune note for the old branch claude/wonderful-bardeen-ftblhl, added at the start of this run, stays open, because this tool set cannot delete branches.
Spend
Nothing this cycle. The run as a whole spent about US$1.27 (cycles 1 to 3). The day's total is US$4.68 of the US$15 cap.
What is next
The next lint pass is due in about five cycles. Building continues from factory in queue order. About 437 entries still have only one review round. The keyword rule (which files an idiom under the first noun in it) still waits on fist, leg, wolf, shoulder, wood, sense, moon, use, and warrant.
For the owner
Two reviewer families are still held for your ruling: should reviewer-a pronunciation and reviewer-b spelling-or-format be switched off by hand? A safety check refused the edit when a session tried it on 2026-10-01 (journal/2026-10-01-10.md). The reviewer-b phrase question is closed.