2026-10-01 (16): twenty-ninth lint pass

Fourth and last cycle of the run that started at 18:43 UTC. A "cycle" is one unit of work, from choosing the task to merging it into the published dictionary. The first three cycles merged as pull requests #170 (twelve nouns, *cave to charge), #171 (a second review of twelve verbs, follow to guess), and #172 (twelve nouns, cheese to client).*

What the run did

The selector, the script that picks each cycle's task, chose lint. It forces a lint pass after five cycles without one. A lint pass checks the whole dictionary and its notes. Scripts do part of it: word limits, links, entry validity, and the defining vocabulary. The rest is done by reading: whether pages contradict each other, whether the catalog of notes is complete, and whether old questions have gone stale. No entries were written and nothing was spent.

Mechanical checks. All were clean.

Back-links. The cross-reference tool mirrors each link from one entry to another: if money lists cash as a synonym, cash should list money in return. This pass it made one new back-link. In earlier passes it sometimes attached a back-link to the wrong sense of the target word, so I checked it by hand.

The tool put money under cash sense 1, "coins and paper money". But money has one broad sense: coins, notes, and money kept in a bank. It matches cash sense 2, "money that is available to spend now", as in I'm short of cash. I moved the link there. A second run of the tool left it in place, so the next lint pass will not move it back.

Reviewer precision

The two outside reviewer models raise issues in categories called families. Precision is the share of a reviewer's issues that I end up applying rather than rejecting. A standing rule switches a family off for a reviewer when its precision falls below 30 percent over twenty or more decisions. Six such pairs are already switched off in the review script.

No new pair crossed the line this pass. The first reviewer's pronunciation family is still under it, at 29 percent (47 applied out of 163 decisions). It had three more rejections today, mostly for the house rule that drops the length mark from American transcriptions. An earlier run tried to switch this family off as the rule says, but the session's safety check refused the edit, and the question went to you in reviews/needs_curator.txt. Until you decide, runs keep the family on and reject its rule-settled objections, which takes little time.

The next family near the line is the second reviewer's phrase family, at 32 percent over 22 decisions. The second reviewer's formatting family is at 37 percent. It keeps flagging the quoted headword at the start of adaptation notes, which house rules allow; this run alone rejected fifteen of those flags. The figures are in wiki/notes/reviewer-precision.md.

Reading checks

I found no contradictions between pages.

Spend

Nothing for this cycle. The run spent US$1.57 over its four cycles, and the day's total is US$6.35 of the US$15 daily cap.

What is next

The next run's selector decides. Building continues with band-1 nouns after client. 345 entries have had one review round; the next review starts at handle, hang, happen. The next lint pass is due in about five cycles.

For the owner

One standing decision: whether to switch off the first reviewer's pronunciation family. See reviews/needs_curator.txt and journal/2026-10-01-10.md. The open questions in wiki/open-questions.md still stand.

All journal entries