Week ending 8/9

Executive summary

This week built the mechanism these reports have been asking for since June: a way to actually settle doctrine terminology rather than accumulate it. Library Review shipped end to end, from database schema through a volunteer reviewer interface to an admin promote step, alongside the offline pipeline that merges per-book output into one canonical registry per language. Content growth paused for the first time in three weeks, with New Testament and Old Testament counts flat, because engineering went into review infrastructure instead. A database migration failed every production deploy until it was repaired, the fourth consecutive week with a deploy-time incident.

What's new

• Change

  Shipped Library Review end to end. Volunteers can now review contested doctrine terms in a browser, see other reviewers, and agree or disagree.

  Impact

  Terminology decisions can finally be settled by speakers of the language, not left pending indefinitely.

  Next action

  Name the first volunteer reviewers for German and Spanish.

• Change

  Added a volunteer invitation flow and weighted voting. Experienced Bible teachers who are vouched for count more, but no term settles on one person's answer.

  Impact

  Review quality scales with reviewer expertise without letting any single voice decide doctrine.

  Next action

  Send one real invitation end to end before inviting a group.

• Change

  Built the offline pipeline that feeds review: a conflict analyzer, per-language normalization rules, a canonical merge step, and a promote step.

  Impact

  Knowledge from every loaded Bible book now merges into one requirements package per language.

  Next action

  Promote German once its 78 pending decisions are settled.

• Change

  Corrected every language package that described itself as covering only Romans. The fix spanned 62 languages.

  Impact

  The Library no longer understates its own coverage to partners and translators reading a package.

  Next action

  Confirm the corrected framing reads correctly in a published bundle.

• Change

  A migration failed every production deploy until repaired the same day. A separate older migration was made idempotent.

  Impact

  This is the fourth straight week with a deploy-time database incident, despite last week's written recommendation.

  Next action

  Make idempotent migrations a merge requirement rather than a recommendation.

• Change

  Onboarded Twi as a new language, with Phase 1 output for nine books. Only Romans is published to the Library so far.

  Impact

  Language coverage grew to 67 published bundles, though eight Twi books await publishing.

  Next action

  Publish the remaining eight Twi books to the Library.

• Change

  Tier-filtered the translation memory in the Crowdin system prompt. Added password recovery and structured language and teaching fields to profiles.

  Impact

  Translation prompts stay affordable as registries grow, and volunteers can now record their qualifications.

  Next action

  Use the new profile fields to identify qualified reviewers for the queue.

Business impact

The doctrine-review backlog is now measurable rather than theoretical. 203 terminology decisions across German and Spanish sit in a queue volunteers can work through in a browser, with weighted voting and a quorum that no single person can satisfy alone. The review-capacity gap leadership has been asked to staff for three weeks now has a working tool waiting on people. Separately, every language package that described itself as covering only Romans was corrected, across 62 languages, removing a claim that understated the Library to anyone reading it.

Metrics

• Registry decisions awaiting a reviewer

  203(was 0) ▲ Up from last period

  German 78, Spanish 125. Newly measurable this week; previously these conflicts were unresolved but uncounted.

• Languages with a complete New Testament

  57(was 57) → No change from last period

  First flat week after two consecutive jumps. Engineering went to review infrastructure.

• Languages with a complete Old Testament

  10(was 10) → No change from last period

• Book pages live in the Library

  1,939(was 1,938) ▲ Up from last period

• Languages with a published Library bundle

  67(was 66) ▲ Up from last period

Risks

Four consecutive weeks have now included a deploy-time database incident. Migration 0025 failed every production deploy until it was repaired the same day. Last week's report asked for idempotent-by-default migrations; one older migration was hardened this week, but the new one still reached production broken, so the recommendation has not yet changed outcomes. The 4 August target to begin a theologian review pass on Turkish Matthew passed with no evidence it started. Review capacity remains unstaffed while the queue is now 203 concrete decisions, and a deploy size ceiling allows only three languages under review at once.

Decisions needed

Leadership: name the first volunteer reviewers, because the capacity request is no longer theoretical and 203 decisions are queued behind it. Decide which language joins German and Spanish under the three-language review ceiling. Consider making idempotent migrations a merge requirement, after four straight weeks of deploy-time database incidents and one week where the written recommendation did not prevent a repeat.

Next Week

• Name the first volunteer reviewers for the German and Spanish queues
• Settle the Spanish SEÑOR and Jehová convention before reviewers begin
• Choose the third language for the review batch, at the three-language ceiling
• Make idempotent migrations a merge requirement after four weeks of incidents

Key dates

• August 15, 2026 — Target: Translation Studio feature completion (per the site's own under-construction banner)