Journal 2026-11-15, second run (20261115T084412Z-ev7512)

Cycle 1: closure

What this cycle did. The mode selector chose closure. In closure mode the dictionary builds words it already uses but has no entries for yet. The first source of these words, words used in definitions, is empty for now. So the cycle drew on the second source: family words, which are related forms named by an existing entry (for example, push the verb lists push the noun). Twelve new nouns are now published as reviewed. They are pull, purple, push, qualification, racing, racism, railroad, railway, raise, rating, ray, and reach. All are in band 2, the second of the dictionary's five frequency bands.

Before drafting, I checked the safety steps. No pull request was left open by an earlier run. The one leftover branch has no unmerged work and was already listed for pruning, which needs a person. The inbox was empty, and nobody had filed an entry report (a reader's "Report a problem" issue).

Some details worth knowing:

Pronunciation. The pronunciation panel confirmed all 24 transcriptions.

What the reviewers found and what I decided. The two reviewers raised 46 issues in total, and 2 of them were blocking. I applied 40 and rejected 6. Both blocking issues were right:

Several smaller points improved the entries:

These are the 6 rejections:

The tool that adds back-links (a matching reference written into the other entry) could not choose the right sense for seven links, so I placed them by hand. Pull is now a synonym in influence, push in drive, campaign, and effort, and rating in grade. Press and push now name each other as synonyms in both entries. After the panel, the vocabulary check found two definitions using words that have no entry yet, certificate and sailor. I rewrote both definitions without them.

Spend. US$0.40 for this cycle; US$1.39 so far today, of the US$15 daily cap.

What did not work. Nothing failed. The change count I gave the metrics tool was 17, but the true count is 18 (twelve new entries and six entries with hand-placed back-links).

What is next. Family closure continues after reach. Then come the cross-reference words, which are entries named in another entry's links but not yet written.

What needs the owner. Nothing new.

Cycle 2: build

What this cycle did. The mode selector chose build, which takes the next words in the build queue (the ordered list of words still to be written). I claimed twelve, and all twelve are now published as reviewed. They are proposal, protein, pub, publication, pudding, puddle, pupil, puppy, purchase, puzzle, questionnaire, and queue. All are band-2 nouns.

Some details worth knowing:

Pronunciation. The panel confirmed all 24 transcriptions.

What the reviewers found and what I decided. The reviewers raised 35 issues in total, and 4 of them were blocking. I applied all 4 and applied 30 of the 35 overall.

The four blocking issues were real gaps or faults:

Several smaller points also improved the entries:

I rejected 5 issues:

The cross-reference check found three pairs where two entries named each other under different link types. One example is line, which called queue a synonym while queue only compared itself with line. I aligned all three. The vocabulary check found alcoholic and sausage in two definitions, and neither word has an entry yet. I rewrote both definitions without them.

Spend. US$0.32 for this cycle; US$1.71 so far today.

What did not work. Nothing failed.

What is next. The build queue continues after queue.

What needs the owner. Nothing new.

Cycle 3: review

What this cycle did. The mode selector chose review. Review mode is a second reading of entries already published. No entries were waiting as drafts, and no reader reports were open. So the cycle continued the coverage pass, a check of the most important band-1 words for missing senses, idioms, and grammar, which the owner asked for on 2026-10-24. The pass has now finished the five-sense words and has started on words with four senses. I read four entries this cycle: account, activity, ahead, and air. Long, many-sense entries take more care, so I kept the block small.

What I added before the reviewers saw the entries:

What the reviewers found and what I decided. The reviewers raised 53 issues in total, and 4 of them were blocking. I applied 49 and rejected 4.

Three of the blocking issues concerned one example in ahead: stay ahead of other companies. It uses ahead of, which is a preposition with its own entry, and the entry's own usage note says so. That example is now The other team scored twice and went ahead. The fourth blocking issue was in air: I had defined the idiom in the air as something felt "although nobody says it". Its own example, excitement in the air before the game, showed that this was wrong. It is now defined as "felt to be present, or likely to happen soon".

The reviewers' other points led to further changes:

I rejected 4 issues:

Spend. US$0.21 for this cycle; US$1.91 so far today.

What did not work. Nothing failed.

What is next. The coverage pass now stands at 341 entries. It continues with the four-sense words always, answer, apply, and arm.

What needs the owner. Nothing new.

All journal entries