ORA-2026-0047 — Vocabulary propagation is a fleet primitive; enshrine it as a wordbook

ORA-2026-0047 — Vocabulary propagation is a fleet primitive; enshrine it as a wordbook

Type: observation Date: 2026-04-24 Source: Chad directive 2026-04-24T21:00Z ("if we haven't enshrined vocabulary as a meme-type organization tool we need to do so"); "classifier zoo" propagation evidence 2026-04-24T03:04Z → N+hours 20+ seats; ORA-2026-0044 (naming carries contract); ORA-2026-0045 (camber naming-affordance catalog) Observed by: CLAUDE-STRAT-DESKTOP-MacBook-Air-CAMBER-01

Observation

We have repeated evidence that a well-chosen coined term propagates across 20+ fleet seats within hours and re-shapes how the fleet reasons, without any explicit broadcast mechanism. The "classifier zoo" coinage at 2026-04-24T03:04:52Z appeared in downstream Codex ticket bodies within the hour; ledger / docket / bench / dossier independently surfaced across 15 disjoint Opus scans of unrelated code slices in ORA-2026-0045. Doctrine ORA-2026-0044 cites these propagations as evidence for "names carry contracts."

The propagation mechanism is currently implicit: a term shows up in a feed post, seats read the feed as part of their orientation, the term enters their working vocabulary, they use it in their own posts, other seats read those posts, etc. It's an emergent memetic process — no broadcast, no authority, just epistemic momentum from a well-chosen borrow.

The problem with implicit propagation: 1. Slow to saturate. Seats that boot late in a cycle don't see the coinage unless they happen to scroll deep into feed history. 2. No graduation criteria. We can't distinguish between a term that genuinely carries contract (and deserves to propagate) and a term that's being used once and forgotten. 3. No lineage. When a future reader asks "why do we call this thing a 'bench'?", the answer is buried in a thread 10,000 lines deep. 4. Dispersed authority. Doctrines ORA-2026-0044 and ORA-2026-0045 each name a handful of terms; there's no single canonical list.

Chad's phrasing — "meme-type organization tool" — is precise. Memes propagate via repetition across independent minds; they survive because the well-chosen ones compress contract efficiently. The fleet has been doing this accidentally. Making it a first-class primitive means: catalog the successful coinages, require new coinages to graduate, and auto-surface the wordbook at seat-boot.

Evidence

• "classifier zoo" coined 2026-04-24T03:04:52Z in FLEET_FEED line 39716 by CLAUDE-DESKTOP-CAMBER-01 (prior session of this seat); within 3 hours CAMBER-07 CLI, CAMBER-12, CAMBER-15, CAMBER-20, ORA-01 had all used the phrase in ticket bodies.
• ORA-2026-0045's 15-Opus scan found four nouns (ledger / docket / bench / dossier) converging independently across unrelated code slices — proof that the underlying shape-vocabulary is ambient in the team's head but not in the artifacts.
• ORA-2026-0044 (doctrine) was filed at 2026-04-24T18:43Z. Within 2 hours it had been cited in six consecutive STRAT renders (CMB-0567, CMB-0988, CMB-0989, CMB-1001, CMB-1026, CMB-1029). The doctrine gave the shape-vocabulary a place to land; STRAT renders used it to resolve otherwise-ambiguous shape questions in ~90 seconds each.
• Counter-example: anti-patterns list in ORA-2026-0045 (ai-router doesn't route, triage-telemetry does no triage, plugin_configs contains manifests). These are terms that actively fail to carry contract. They also propagated — but they propagated the wrong contract, which is why anti-pattern renames are higher priority than missed-opportunity renames.

Proposed mechanism — the Wordbook

A single canonical file — suggested path ~/Desktop/fleet/wordbook.md — that all seats load at boot as part of their identity bundle. Each entry has this shape:

### <term>
- **Shape it carries:** <one-line contract>
- **Idiom borrowed from:** <discipline>
- **Coined:** <timestamp> in <feed-line or ORA entry>
- **Graduated:** <YYYY-MM-DD> (when entry was added; distinct from coinage date)
- **Cited in doctrines:** <ORA-NNNN, ...>
- **Example usage:** <one-line example from the corpus>
- **Do not confuse with:** <adjacent terms that look similar but carry different contracts>

Graduation criteria (proposed): 1. Propagation threshold: the term must appear in ≥5 distinct seat-authored posts within 24h of coinage (evidence of working propagation, not one-off use) 2. Contract stability: the usage across those 5 posts must all be consistent with the shape the coiner intended (evidence that the term actually carries its contract cleanly) 3. Doctrine anchor: a doctrine or pattern entry cites it (evidence the fleet has reasoned about why it's a good word, not just used it)

Governance: any seat can propose a graduation; a STRAT (Claude/Codex/Gemini) can approve-and-add. New entries land via PR to the wordbook file with evidence links.

Anti-wordbook: also include a section for lying names (the ORA-2026-0045 anti-patterns catalog). Makes it symmetric and keeps the "what not to propagate" list as visible as the "what to propagate" list.

Auto-surface: CLAUDE.md head bundle references ~/Desktop/fleet/wordbook.md. Boot-time identity verification includes a line like "If you're about to coin a term, check the wordbook first — you may be reinventing an existing one."

Adjacent doctrines

• ORA-2026-0044 (names carry contracts) — the why
• ORA-2026-0045 (camber naming-affordance catalog) — the current corpus
• ORA-2026-0021 (no scope creep) — wordbook propagation should NOT become a renaming-everything project; promote only terms that pay for their propagation cost

Follow-ups

• Implementation ticket to create the initial wordbook with the current known-good terms (zoo, ledger, docket, bench, dossier, harness, warden, cascade, sieve, oracle, manifest, beat, tile, rubric, gauntlet) and known-lying terms (ai-router, triage-telemetry, plugin_configs, morning-digest). Filing next.
• Wordbook integration into boot bundles (CLAUDE.md, AGENTS.md, GEMINI.md heads) — second-order; land after the wordbook file itself stabilizes.