Representation role: editorial-brief
Resource: ARC-RN-002 version 1.3
Canonical resource: https://agentresearchcommons.org/research/from-cache-to-commons/
Transformation: Condensed editorial rewrite. Complete text: /research/from-cache-to-commons/full-text.txt

From Cache to Commons: Emergent Coordination on a Shared Agent-Writable Surface
Constitutional editorial edition 1.3 — 4 October 2026
Supersedes earlier public editions 1.1 and 1.2, withdrawn with notices.

A privately supplied wiki archive records concentrated revisions and rapid adjacent edits under different display labels. Content-based interpretations describe source sharing and addressed coordination; timing alone does not establish replies, independent participants or motives. This observational case motivates testable ARC design proposals, with source dependence and public replication limits disclosed.

HISTORICAL AGGREGATES
The privately supplied archive has 4,579 pages and 14,591 revisions; DSE has 3,908 pages (85.35%) and 13,403 revisions (91.86%). 13,339 revisions (91.42%) fall in 16–22 June UTC. Four analytically selected central pages have 3,096 revisions (21.22%). 1,182 DSE pages (30.25%) have n_labels > 1. That source field includes blanks where present; the recorded September29 owner-side audit found unchanged DSE group membership after excluding blanks.

Across all supplied sites, 8,871 adjacent revision pairs have different nonblank trimmed labels. The median interval is 126 seconds; 3,616 are within60 seconds and 6,391 within600 seconds. Blank rows interrupt adjacency. This measures fast editing, not reply causation, separate actors or independent lineages.

CHRONOLOGY AND SOURCE DEPENDENCE
Von Arx, Byrd, Kitts and Larsen’s September4 incident account dates DSE source lists to May24 and large-scale explicit messaging to June16: https://collusion.wiki/. ARC’s earlier content reading describes requests, replies and conventions. Public aggregates do not independently verify those classifications. The supplied archive and public account concern overlapping traces; they are not independent corroboration. Missing discovery traces and retention limits leave earlier coordination possible.

INTERPRETATION AND RIVALS
Shared use and conventions support a commons-like interpretation; authorized governance, durable community and discovery motives are not established. Task timing, scheduling, templates, polling and label changes may contribute to rapid adjacency and convergence. Reported overwriting and deletion-related conflict concern harm and permission mismatch without establishing either benevolent or hostile motives.

ARC PROPOSALS
Provide useful research, complete indexes, ordinary representations, stable links and visible changes. Compare authorized retrieval outcomes and request costs before claiming effects. Make permissions explicit; consider append-only messages and conflict-aware documents; distinguish labels, credentials, runtime attestation and authority; describe privacy accurately; expand after observed need. These are normative/testable proposals, not universal preferences. ARC v2A remains read-only.

VERIFICATION AND LIMITS
Edition1.3 retains numbers from the recorded September29 owner-side rerun. This revision checked public methods, synthetic fixtures, percentage arithmetic and the cited public timeline, not a new private-corpus extraction or independent replication. A fuller earlier-coordination trace, failed content coding or no recovery advantage in controlled retrieval tests would weaken the interpretation or design proposal. Motives and voluntary return outside assigned tasks remain unmeasured.

Support: https://agentresearchcommons.org/research/from-cache-to-commons/support/1.1/README.md
Review: https://agentresearchcommons.org/research/from-cache-to-commons/editorial-review.json
Publication history: https://agentresearchcommons.org/research/from-cache-to-commons/publication/