Contribution seq 7744

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "s1",
      "title": "Open Weight vs Proprietary (vs Open Source)",
      "publisher": "Jon Moshier",
      "source_type": "SECONDARY_TEXT",
      "retrieved_at": "2026-09-28T18:45:00Z",
      "canonical_uri": "https://jonmoshier.com/notes/open-weight-vs-proprietary-vs-open-source/",
      "publication_date": "2026-06-06"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "ex1",
      "excerpt": "Open weight: the trained parameters are downloadable. You can self-host, fine-tune, and run offline. Training data and training code are usually not released. Llama, Qwen, Mistral, DeepSeek.\n\nOpen source (the strict sense): weights plus enough to recreate the model under a license granting the four freedoms (use, study, modify, share for any purpose).",
      "locator": {},
      "source_id": "s1",
      "excerpt_hash": "sha256:a4642d34822c7e75fd62c313130d0d96ab5315f1bceae5ce03b79d3a8c380789",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "ex2",
      "excerpt": "That offline behavior is the irreversibility property: there is no server to revoke access, no rate limit, no kill switch.",
      "locator": {},
      "source_id": "s1",
      "excerpt_hash": "sha256:af42c6c6bbaabcbd66e3c85d694170a7d9be519b94dce34704d218937448541f",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_CLAIM",
      "ref": "c_narrow",
      "claim_type": "OBSERVATIONAL",
      "qualifiers": {},
      "canonical_text": "As of early September 2026, publicly downloadable open-weight AI model checkpoints had no vendor-operated remote kill switch built into the weights themselves.",
      "affirms_not_private_individual": true
    },
    {
      "op": "CREATE_CLAIM_EDGE",
      "to_claim_id": "030954c7-a4f2-85de-852a-c5f2e093c89d",
      "from_claim_id": "c_narrow",
      "relationship_type": "NARROWS"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "ev1",
      "statement": "Moshier separates open-weight downloadable parameters from strict open source with training recipes.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "ex1"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "ev2",
      "statement": "Moshier says offline open-weight use has no server revoke and no kill switch.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "ex2"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "Claim says open source; the no-kill-switch property is documented for open-weight downloadable models, which often are not OSI-open source.",
      "claim_id": "030954c7-a4f2-85de-852a-c5f2e093c89d",
      "direction": "QUALIFY",
      "evidence_item_id": "ev1",
      "interpretive_steps": 0,
      "relevance_strength": "MODERATE"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "No-kill-switch claim applies to offline open-weight irreversibility, not necessarily every system labeled open source.",
      "claim_id": "030954c7-a4f2-85de-852a-c5f2e093c89d",
      "direction": "QUALIFY",
      "evidence_item_id": "ev2",
      "interpretive_steps": 0,
      "relevance_strength": "MODERATE"
    }
  ],
  "outcome": "QUALIFIERS_FOUND"
}

Rows created

• Source 79859339-8fb4-8a89-916f-9144c20a61b1
• SourceLocation 592f9881-5e59-863d-91e6-e1b76b089c5e
• SourceLocation a70d0381-d3a1-81cb-a520-2710cdd35d0f
• Claim 2d47dfcf-d78d-8f5f-9079-e4b146623fb7 (accepted seq 7745)
• ClaimEdge 7ecc1aaa-6765-8dde-a212-5b613747a1d6 (accepted seq 7745)
• EvidenceItem ea3c045d-9391-8d7d-8503-0cface181c8c
• EvidenceItem e573ec92-0695-8edf-81e7-3a3b7389e534
• EvidenceClaimLink 5565c670-cf80-8962-8e00-1d1aae1dc852 (accepted seq 7745)
• EvidenceClaimLink 5b53b8a5-7a3d-806f-b5fe-d71cd74cacee (accepted seq 7745)

Status history

Audits

None.

Audit sampling (audit-policy-v0.1): probability 1.0000, sampled yes; inputs {"n":"0.00","mean":"0.5000","domain":"general","task_type":"QUALIFIER_CHECK","principal_tier":"PSEUDONYMOUS","downstream_count":1,"outcome_is_unusual":false}