Edit-Language Collective (LM-Dawn class)

Class card for the LM-Dawn cluster of practitioner collectives whose operational task is emitting, curating, contesting, and validating structured-grammar edits to shared civilizational models. The canonical grammar is the Prime Radiant primitive-grammar token set (MVP_PLAN.md §1): <|machine|>, <|coupling|>, <|signal|>, <|stress|>, <|transition|>, <|new_machine|>. The class generalizes to any collective that governs a shared edit-language for a world-model or knowledge commons — the practitioner-side of Prime Radiant's Phase 5 LLM edit-agent loop. EDIT-GRAMMAR PRACTICE as LM form [EXTRAP]: The edit-language collective is the operational complement to the dawn-machine-substrate-knowledge-class (which curates epistemic substrate) and the cross-era-prediction-collective-class (which produces predictions). This class EDITS the predictive substrate: it governs the grammar by which new machines are proposed (<|new_machine|>), couplings are asserted (<|coupling|>), and transitions are signaled (<|transition|>). The edit-language is the lowest-level coordination layer of the world-model — it determines what can be named, what can be linked, what can be contested. Emergence_subtype [v0.2 gap — recorded here]: edit_grammar_practice. Authority is distributed and contestable: any contributor can propose an edit; validity is determined by collective deliberation over the proposals/ queue. Analogy: Wikipedia talk-page NPOV deliberation applied to civilizational-machine ontology rather than encyclopedia articles. Named instances [EXTRAP]: (1) Prime Radiant card authors (Sonnet sub-subagents in this session — meta- instance; the authors of Machine Cards ARE the edit-language collective for Prime Radiant's model of civilizational machinery; every card emission is a primitive-grammar edit). (2) Wikipedia editor community (~250K active editors as of 2024; analog form via wiki-edit grammar; Wikipedia talk-page NPOV deliberation is structurally adjacent to edit-language practice; edit-revert cycles are contested-edit- resolution at scale). (3) Open-source maintainer communities operating around RFC/issue-comment grammars: Rust language team RFC process (~300 RFCs/year; RFC = formal edit-language proposal for language semantics); Linux maintainer-tree patch-review (Torvalds's maintainer tree as edit-routing grammar; LKML as contested-edit arena); Python PEP system (~700+ accepted PEPs; PEPs are formal grammar for Python's operational semantics). (4) Schema.org community (~50 active contributors maintaining structured- vocabulary for the web; schema vocabulary = edit-language for machine-readable world descriptions; every schema extension is a grammar edit). (5) IPLD / IETF schema-design subcommunities (IPLD = InterPlanetary Linked Data; IETF drafts as RFC-grammar for internet protocol semantics; XBRL taxonomy maintenance community ~200 contributors maintaining financial reporting edit-grammar). (6) Stack Exchange moderator network (~5K active mods; moderation policy is edit-language governance — what can be asked, what can be answered, what constitutes valid edit-language in the Q&A form). (7) e-flux journal editorial collective; New Models magazine editorial — curatorial edit-language practices for theory-culture discourse. (8) Cosma Shalizi's analytical-thinker community (Bactra review practice as contested-edit labor on epistemic models). Four mechanism pillars of the LM edit-language class: (1) Primitive-grammar governance — the collective maintains and extends the token vocabulary (adding new primitives, deprecating ambiguous ones, resolving collisions between competing grammars). Analog: Rust RFC process for language feature grammar; Python PEP Steering Council for language semantics. (2) Proposals-queue triage — contested edits enter a proposals/ queue; the collective deliberates, validates, rejects, or defers. Analog: Wikipedia talk-page discussions; LKML patch review; IETF working-group comment periods. Prime Radiant: the proposals/ queue is the machine version of this mechanism. (3) Canon/extrap provenance labeling — the collective labels every edit with epistemic confidence ([CANON] vs. [EXTRAP]) and wave_source, maintaining the provenance chain. This is the edit-language's anti-proletarianization mechanism: inscribed provenance permits re-internalization by future contributors who did not author the original edit. (4) Translation-chain maintenance (Latour ANT) — as the grammar is deployed in new domains, the collective manages translation chains: re-enrolling actors in the new grammar, settling disputes about scope, handling betrayals (edits that violate the grammar's OPP). Analog: W3C working groups; IETF standardization debates; Wikipedia's manual-of-style enforcement. Theoretical anchors: Wave-2 Latour ANT translation chains (edit-language is a translation chain from contributors to world-model); Wave-1 Bryant onto-cartography (cards as material-discursive objects that the collective produces); Stiegler tertiary retention as inscribed-edit substrate (edits are tertiary retentions of civilizational knowledge); MVP_PLAN.md §1 primitive-grammar table; Wikipedia talk-page NPOV practice; collective- discourse-analysis as method. Proletarianization risk [EXTRAP]: MEDIUM-HIGH (0.58). The edit-language is inscribed as a technical artifact (the grammar, the schema, the proposals/ queue). If the contributor community thins faster than LLM- automated editing replaces living grammatical competence, the substrate persists but the collective's ability to contest and evolve the grammar degrades (Stiegler signal). The LLM-as-edit-agent (Prime Radiant Phase 5) is a dual-edged development: accelerates edit throughput but risks de-skilling the human practitioner community that validates contested edits. Capture resistance [EXTRAP]: MEDIUM (0.50). Grammar communities that are open-protocol (Rust, Python, IETF, Wikipedia) are structurally resistant to capture; commercial grammar projects (proprietary schema vocabularies, closed-ontology curation) are vulnerable. The capture-resistance gradient within the class is high. v0.2 substrate gap: `institutional` substrate type is absent from the enum. The edit-language collective is primarily cognitive (grammatical competence), semiotic (the grammar itself), and social (the deliberative community), with an institutional dimension (WMF governance, Rust Foundation, PSF, IETF) captured imperfectly by the `social` enum value. [STUB-substrate-enum-gap].