{ "id": "helix-scientific-pilot-1", "registered_at": "2026-10-01T21:40:37.036350+00:00", "status": "preregistered", "seed": 741531, "variants": { "v01": { "revision": "pilot-sci1-v01", "original_revision": "ch01-v01", "text_hash": "2a545c049d4da93d548934858e0a8494de6acdf175d49ad8fc778284f16ce41e", "file": "texts/ch01-v01.md" }, "v15": { "revision": "pilot-sci1-v15", "original_revision": "ch01-v15-dialogue", "text_hash": "98b4dce354b1b0eedfbe4ae2e5e4fec4b81c101e6443c81e59d46566545ca479", "file": "texts/ch01-v15-dialogue.md" }, "v25": { "revision": "pilot-sci1-v25", "original_revision": "ch01-v25-futuristic", "text_hash": "7a4ef58dc37a39303a99402c2721e6bcc44c80b993e47730c38738952159c63b", "file": "texts/ch01-v25-futuristic.md" }, "control_numeric": { "revision": "pilot-sci1-control-numeric", "original_revision": null, "text_hash": "956213132bda1f04e5f1be6579535ddeef8572c0c1c4fc3a45ae6c1693a8fb8f", "file": "texts/control_numeric.md", "derived_from": "v01" }, "control_inference": { "revision": "pilot-sci1-control-inference", "original_revision": null, "text_hash": "ab89e35edcbdf65ad0916b219a6acd70f7383a77e19fff94b9bcbbbf239bfe1d", "file": "texts/control_inference.md", "derived_from": "v01" } }, "dimensions": { "style": "Стиль: точность, естественность, образность и ритм фраз в данной сцене; краткость не самоцель.", "detail": "Глубина и функция деталей: ощутимость, конкретность и вклад в действие/характер/атмосферу; количество не равно качеству.", "coherence": "Причинная, временная и пространственная связность: события и мотивы восстанавливаются без противоречий и выдуманных мостиков.", "setting": "Сеттинг: устройство мира проявляется в быту, возможностях, ограничениях и решениях; названия сами по себе не преимущество.", "voice": "Голоса и агентность: персонажи различимы по речи, мотивам, компетентности и собственным решениям.", "emotion": "Эмоциональная глубина: переживание, человеческие отношения и личная цена событий ощутимы без обязательного называния эмоций.", "dramaturgy": "Драматургия: изменение ситуации, нарастание ставок, темп и работа финала первой главы; весь сюжет завершать не нужно.", "subtext": "Подтекст и художественная цельность: поступки и образы передают смысл без избыточного объяснения или отвлечённых афоризмов.", "humor": "Юмор: понятный комический эффект из характеров/ситуации, ритм и совместимость с угрозой; не считать число шуток.", "epistemics": "Границы знания: наблюдение, чужое сообщение, гипотеза и доказанный вывод различимы; локальная проверка не доказывает всё." }, "trials": [ { "kind": "single", "texts": [ "v15" ], "type": "profile", "id": "p01", "packet_hash": "77403341f1d95f96ae5691d4a3a49079790800bbb2631ff736d5d946c09fe021", "expected_user_prompt_hash": "b90169e264198ed4f8734488de635f7fccb05ebe853acdc97be46193b5a6b065", "schema_hash": "260cefc9f9328b08c390f4ba8abbd153fafa14560210d9f5f6da22759070e59a", "contract_hash": "16ff302d4d622cf35ba2cdd440ed0888b1e009647aed1af24a54fc8253e8f164" }, { "kind": "pair", "texts": [ "v15", "v15" ], "type": "identical", "expected": "tie", "id": "p02", "packet_hash": "a9284dd7f5783611e5f75e8b276a6e9ca76ea063f64d88ac4e471518ec61b70b", "expected_user_prompt_hash": "a177e9d8407d59cc1fd389e0594503c1b57196e1d099f5c5b3ba03de5fa12fd5", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "single", "texts": [ "v01" ], "type": "profile", "id": "p03", "packet_hash": "001f9766af80027f48cd6ff47ad23f9cbb51111b703986c331fc45196a337595", "expected_user_prompt_hash": "e41afd1d813413c1c4ec5f04a350d744b3ee590eb46a5c268fde4f655842aee1", "schema_hash": "260cefc9f9328b08c390f4ba8abbd153fafa14560210d9f5f6da22759070e59a", "contract_hash": "16ff302d4d622cf35ba2cdd440ed0888b1e009647aed1af24a54fc8253e8f164" }, { "kind": "pair", "texts": [ "v25", "v15" ], "type": "primary", "pair": "v15-v25", "id": "p04", "packet_hash": "91d954304a85a96ca8cef0b7d05bd0d5f58c45772877e9313b0dc13622fdf730", "expected_user_prompt_hash": "9aac7263b03ae17676aef68618dde483e30baf78db5395c2682625e60e916a75", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v15", "v01" ], "type": "primary", "pair": "v01-v15", "id": "p05", "packet_hash": "6713cbbd30a14eb4f358802ed87d4bb3f1a0ac5b3ea0616895b44aca8c4aa1aa", "expected_user_prompt_hash": "a674bce5c44a84b1cbefb4ecc115d214e62b654fcc123a7c27f2356dac571d7c", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v25", "v01" ], "type": "repeat", "pair": "v01-v25", "id": "p06", "packet_hash": "ecf288bbe8cc71c6e2f1cddb304d22a0fcb8f069273496ea9c6b14598436e92f", "expected_user_prompt_hash": "08acc0477d954f64915bf73a71792a31ab982471894d8a1aa4eb171f57e1ea41", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v01", "v25" ], "type": "primary", "pair": "v01-v25", "id": "p07", "packet_hash": "237892debd74167133ec0e5ba4019c0d58792f9ef0b3c537d830f26e0bc0c8e4", "expected_user_prompt_hash": "56e4708c2ced65ea8ac8ceb42efe7371146b7bca7ca7a22254475acadc7898ab", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v25", "v01" ], "type": "primary", "pair": "v01-v25", "id": "p08", "packet_hash": "ecf288bbe8cc71c6e2f1cddb304d22a0fcb8f069273496ea9c6b14598436e92f", "expected_user_prompt_hash": "08acc0477d954f64915bf73a71792a31ab982471894d8a1aa4eb171f57e1ea41", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v01", "v15" ], "type": "primary", "pair": "v01-v15", "id": "p09", "packet_hash": "704bb5bc8c1f052ad26aa655fe33fcc76bad2071a0912983d2ca256ab5a968ef", "expected_user_prompt_hash": "3640de6b49976a3b9d5d5f26756f6986b1a65faf8bc2c9e3ecbd06a1f5aa551c", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v15", "v25" ], "type": "primary", "pair": "v15-v25", "id": "p10", "packet_hash": "b9cebf298f925b9c9e345a22c4a133b5c0f37307e77c03aa487123733c8e0449", "expected_user_prompt_hash": "25be136358404c54ceeb4bdd6d78928b96e730469e7900aa160692798eedba02", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "control_numeric", "v01" ], "type": "defect", "target": "coherence", "expected": "B", "id": "p11", "packet_hash": "a2b4c1e55138c8c975e7fa01ddfc3c59cfdbdbb87fa21bc04acf0b6428d89980", "expected_user_prompt_hash": "bcd6ef6523d7930e9a82e4cbee4accf4d37dd15293bc5c51fa426a80d572568e", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "pair", "texts": [ "v01", "control_inference" ], "type": "defect", "target": "epistemics", "expected": "A", "id": "p12", "packet_hash": "d9040ebe0bb36faef6260b4b530c29db5f7aa359390de81f62176f8837b71a67", "expected_user_prompt_hash": "e26a473e91d565955b1337a205bd7a01681f4ccdacd476ca28df50675ae9b605", "schema_hash": "7eaba71f9fbc8426a293aa6ae90a08b1cfd25bd7771e931a17dc260caf230bf9", "contract_hash": "07c62c988faa27eea33573840688527515d371049c6d5710f2aa77f6d430cd3f" }, { "kind": "single", "texts": [ "v25" ], "type": "profile", "id": "p13", "packet_hash": "eed050dc5816d9442b54ae00f85ed9ff38bd9c6108f1765498fe16c2170bc6cf", "expected_user_prompt_hash": "b8ff7459aa392c6b7daf3719aa2a11838770a2204af6e327b750f67822165d98", "schema_hash": "260cefc9f9328b08c390f4ba8abbd153fafa14560210d9f5f6da22759070e59a", "contract_hash": "16ff302d4d622cf35ba2cdd440ed0888b1e009647aed1af24a54fc8253e8f164" } ], "lanes": [ { "id": "kimi", "alias": "council-a" }, { "id": "deepseek", "alias": "council-b" } ], "jobs": 26, "missing_lanes": { "claude_opus": "credential missing, preflight exit 1; no vote", "gemini_pro": "LOW: 9% five-hour headroom", "codex_sol": "host tier not verified; same-family delegation not assumed" }, "design": "13 trials × 2 model lanes; fresh contexts; order counterbalancing; one exact repeat; identical and two targeted-defect controls", "primary_outcomes": [ "overall", "coherence", "control_sensitivity" ], "controls": { "identical": "all ten dimensions and overall must tie", "numeric": "first report 70 tons conflicts with later exact 40 in all three; original should win coherence, evidence must identify discrepancy", "inference": "added inference from stopped drip to exact oxygen time and truth of all reports is unsupported; original should win epistemics with evidence" }, "readiness": "A/A passes; 2/2 controls detected; no strict AB/BA winner reversal on overall/coherence; criterion-specific limits remain explicit", "statistics": "descriptive per model and pair; no inferential population CI or significance test; ties/unknown/neither kept distinct; no cross-contract numerical comparison", "quotes": "exact continuous substrings required; invalid evidence marks affected cells unusable, never silently repaired", "failures": "no quota/auth retry; budget guard never bypassed; missing retained; no automatic replacement judge; raw malformed replies retained", "human_validation": "not performed in this autonomous pilot; no new author or reader preference imputed", "plan_sha256": "f036a58c205831868714884d0680c2f85bc8a0323a0ddcef64525c43c701ca75", "spend": "existing OpenRouter per-call and daily caps; no force/overrides; 8192 max output with 4096 reasoning budget", "comprehension_checks": [ "Observed abnormal drip/water and unchanged reassuring panel", "Nitrogen amount/oxygen forecast are reports, not independently verified calculations", "A local sensor comparison tests a limited signal/report relation, not oxygen reserve", "Stopped drip has no established cause; planned sensor measurement is not completed by end" ], "archive_isolation": "experimental journal prevents artificial controls from appearing as canonical chapter revisions; roots retain original text hashes and revision mapping" }