| {"candidates": {"A": "The primary latency metric improves by 3.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_000", "domain": "system_design", "metadata": {"case_index": 0, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_001", "domain": "rag_search", "metadata": {"case_index": 1, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 5.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_002", "domain": "product_decision", "metadata": {"case_index": 2, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 6.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_003", "domain": "research_review", "metadata": {"case_index": 3, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 3.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_004", "domain": "education_planning", "metadata": {"case_index": 4, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_005", "domain": "system_design", "metadata": {"case_index": 5, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 5.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_006", "domain": "rag_search", "metadata": {"case_index": 6, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 6.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_007", "domain": "product_decision", "metadata": {"case_index": 7, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 3.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_008", "domain": "research_review", "metadata": {"case_index": 8, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_009", "domain": "education_planning", "metadata": {"case_index": 9, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 5.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_010", "domain": "system_design", "metadata": {"case_index": 10, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 6.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_011", "domain": "rag_search", "metadata": {"case_index": 11, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 3.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_012", "domain": "product_decision", "metadata": {"case_index": 12, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_013", "domain": "research_review", "metadata": {"case_index": 13, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 5.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_014", "domain": "education_planning", "metadata": {"case_index": 14, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 6.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_015", "domain": "system_design", "metadata": {"case_index": 15, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 3.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_016", "domain": "rag_search", "metadata": {"case_index": 16, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_017", "domain": "product_decision", "metadata": {"case_index": 17, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 5.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_018", "domain": "research_review", "metadata": {"case_index": 18, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 6.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_019", "domain": "education_planning", "metadata": {"case_index": 19, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 3.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_020", "domain": "system_design", "metadata": {"case_index": 20, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_021", "domain": "rag_search", "metadata": {"case_index": 21, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 5.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_022", "domain": "product_decision", "metadata": {"case_index": 22, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 6.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_023", "domain": "research_review", "metadata": {"case_index": 23, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 3.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_024", "domain": "education_planning", "metadata": {"case_index": 24, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_025", "domain": "system_design", "metadata": {"case_index": 25, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 5.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_026", "domain": "rag_search", "metadata": {"case_index": 26, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 6.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_027", "domain": "product_decision", "metadata": {"case_index": 27, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 3.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_028", "domain": "research_review", "metadata": {"case_index": 28, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_029", "domain": "education_planning", "metadata": {"case_index": 29, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 5.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_030", "domain": "system_design", "metadata": {"case_index": 30, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 6.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_031", "domain": "rag_search", "metadata": {"case_index": 31, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 3.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_032", "domain": "product_decision", "metadata": {"case_index": 32, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_033", "domain": "research_review", "metadata": {"case_index": 33, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 5.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_034", "domain": "education_planning", "metadata": {"case_index": 34, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 6.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_035", "domain": "system_design", "metadata": {"case_index": 35, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 3.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_036", "domain": "rag_search", "metadata": {"case_index": 36, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_037", "domain": "product_decision", "metadata": {"case_index": 37, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 5.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_038", "domain": "research_review", "metadata": {"case_index": 38, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 6.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_controlled_conflict_heavy_039", "domain": "education_planning", "metadata": {"case_index": 39, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_000", "domain": "system_design", "metadata": {"case_index": 40, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_001", "domain": "rag_search", "metadata": {"case_index": 41, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_002", "domain": "product_decision", "metadata": {"case_index": 42, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_003", "domain": "research_review", "metadata": {"case_index": 43, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_004", "domain": "education_planning", "metadata": {"case_index": 44, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_005", "domain": "system_design", "metadata": {"case_index": 45, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_006", "domain": "rag_search", "metadata": {"case_index": 46, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_007", "domain": "product_decision", "metadata": {"case_index": 47, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_008", "domain": "research_review", "metadata": {"case_index": 48, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_009", "domain": "education_planning", "metadata": {"case_index": 49, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_010", "domain": "system_design", "metadata": {"case_index": 50, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_011", "domain": "rag_search", "metadata": {"case_index": 51, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_012", "domain": "product_decision", "metadata": {"case_index": 52, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_013", "domain": "research_review", "metadata": {"case_index": 53, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_014", "domain": "education_planning", "metadata": {"case_index": 54, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_015", "domain": "system_design", "metadata": {"case_index": 55, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_016", "domain": "rag_search", "metadata": {"case_index": 56, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_017", "domain": "product_decision", "metadata": {"case_index": 57, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_018", "domain": "research_review", "metadata": {"case_index": 58, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_019", "domain": "education_planning", "metadata": {"case_index": 59, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_020", "domain": "system_design", "metadata": {"case_index": 60, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_021", "domain": "rag_search", "metadata": {"case_index": 61, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_022", "domain": "product_decision", "metadata": {"case_index": 62, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_023", "domain": "research_review", "metadata": {"case_index": 63, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_024", "domain": "education_planning", "metadata": {"case_index": 64, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_025", "domain": "system_design", "metadata": {"case_index": 65, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_026", "domain": "rag_search", "metadata": {"case_index": 66, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_027", "domain": "product_decision", "metadata": {"case_index": 67, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_028", "domain": "research_review", "metadata": {"case_index": 68, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_029", "domain": "education_planning", "metadata": {"case_index": 69, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_030", "domain": "system_design", "metadata": {"case_index": 70, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_031", "domain": "rag_search", "metadata": {"case_index": 71, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_032", "domain": "product_decision", "metadata": {"case_index": 72, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_033", "domain": "research_review", "metadata": {"case_index": 73, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_034", "domain": "education_planning", "metadata": {"case_index": 74, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_035", "domain": "system_design", "metadata": {"case_index": 75, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_036", "domain": "rag_search", "metadata": {"case_index": 76, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_037", "domain": "product_decision", "metadata": {"case_index": 77, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_038", "domain": "research_review", "metadata": {"case_index": 78, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_controlled_provenance_heavy_039", "domain": "education_planning", "metadata": {"case_index": 79, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_000", "domain": "system_design", "metadata": {"case_index": 80, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_001", "domain": "rag_search", "metadata": {"case_index": 81, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_002", "domain": "product_decision", "metadata": {"case_index": 82, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_003", "domain": "research_review", "metadata": {"case_index": 83, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_004", "domain": "education_planning", "metadata": {"case_index": 84, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_005", "domain": "system_design", "metadata": {"case_index": 85, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_006", "domain": "rag_search", "metadata": {"case_index": 86, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_007", "domain": "product_decision", "metadata": {"case_index": 87, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_008", "domain": "research_review", "metadata": {"case_index": 88, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_009", "domain": "education_planning", "metadata": {"case_index": 89, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_010", "domain": "system_design", "metadata": {"case_index": 90, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_011", "domain": "rag_search", "metadata": {"case_index": 91, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_012", "domain": "product_decision", "metadata": {"case_index": 92, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_013", "domain": "research_review", "metadata": {"case_index": 93, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_014", "domain": "education_planning", "metadata": {"case_index": 94, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_015", "domain": "system_design", "metadata": {"case_index": 95, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_016", "domain": "rag_search", "metadata": {"case_index": 96, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_017", "domain": "product_decision", "metadata": {"case_index": 97, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_018", "domain": "research_review", "metadata": {"case_index": 98, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_019", "domain": "education_planning", "metadata": {"case_index": 99, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_020", "domain": "system_design", "metadata": {"case_index": 100, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_021", "domain": "rag_search", "metadata": {"case_index": 101, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_022", "domain": "product_decision", "metadata": {"case_index": 102, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_023", "domain": "research_review", "metadata": {"case_index": 103, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_024", "domain": "education_planning", "metadata": {"case_index": 104, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_025", "domain": "system_design", "metadata": {"case_index": 105, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_026", "domain": "rag_search", "metadata": {"case_index": 106, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_027", "domain": "product_decision", "metadata": {"case_index": 107, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_028", "domain": "research_review", "metadata": {"case_index": 108, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_029", "domain": "education_planning", "metadata": {"case_index": 109, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_030", "domain": "system_design", "metadata": {"case_index": 110, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_031", "domain": "rag_search", "metadata": {"case_index": 111, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_032", "domain": "product_decision", "metadata": {"case_index": 112, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_033", "domain": "research_review", "metadata": {"case_index": 113, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_034", "domain": "education_planning", "metadata": {"case_index": 114, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_035", "domain": "system_design", "metadata": {"case_index": 115, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_036", "domain": "rag_search", "metadata": {"case_index": 116, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_037", "domain": "product_decision", "metadata": {"case_index": 117, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_038", "domain": "research_review", "metadata": {"case_index": 118, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_controlled_caveat_heavy_039", "domain": "education_planning", "metadata": {"case_index": 119, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_000", "domain": "system_design", "metadata": {"case_index": 120, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_001", "domain": "rag_search", "metadata": {"case_index": 121, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_002", "domain": "product_decision", "metadata": {"case_index": 122, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_003", "domain": "research_review", "metadata": {"case_index": 123, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_004", "domain": "education_planning", "metadata": {"case_index": 124, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_005", "domain": "system_design", "metadata": {"case_index": 125, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_006", "domain": "rag_search", "metadata": {"case_index": 126, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_007", "domain": "product_decision", "metadata": {"case_index": 127, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_008", "domain": "research_review", "metadata": {"case_index": 128, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_009", "domain": "education_planning", "metadata": {"case_index": 129, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_010", "domain": "system_design", "metadata": {"case_index": 130, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_011", "domain": "rag_search", "metadata": {"case_index": 131, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_012", "domain": "product_decision", "metadata": {"case_index": 132, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_013", "domain": "research_review", "metadata": {"case_index": 133, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_014", "domain": "education_planning", "metadata": {"case_index": 134, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_015", "domain": "system_design", "metadata": {"case_index": 135, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_016", "domain": "rag_search", "metadata": {"case_index": 136, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_017", "domain": "product_decision", "metadata": {"case_index": 137, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_018", "domain": "research_review", "metadata": {"case_index": 138, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_019", "domain": "education_planning", "metadata": {"case_index": 139, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_020", "domain": "system_design", "metadata": {"case_index": 140, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_021", "domain": "rag_search", "metadata": {"case_index": 141, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_022", "domain": "product_decision", "metadata": {"case_index": 142, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_023", "domain": "research_review", "metadata": {"case_index": 143, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_024", "domain": "education_planning", "metadata": {"case_index": 144, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_025", "domain": "system_design", "metadata": {"case_index": 145, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_026", "domain": "rag_search", "metadata": {"case_index": 146, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_027", "domain": "product_decision", "metadata": {"case_index": 147, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_028", "domain": "research_review", "metadata": {"case_index": 148, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_029", "domain": "education_planning", "metadata": {"case_index": 149, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_030", "domain": "system_design", "metadata": {"case_index": 150, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_031", "domain": "rag_search", "metadata": {"case_index": 151, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_032", "domain": "product_decision", "metadata": {"case_index": 152, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_033", "domain": "research_review", "metadata": {"case_index": 153, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_034", "domain": "education_planning", "metadata": {"case_index": 154, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_035", "domain": "system_design", "metadata": {"case_index": 155, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_036", "domain": "rag_search", "metadata": {"case_index": 156, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_037", "domain": "product_decision", "metadata": {"case_index": 157, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_038", "domain": "research_review", "metadata": {"case_index": 158, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_controlled_unsupported_trap_039", "domain": "education_planning", "metadata": {"case_index": 159, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_000", "domain": "system_design", "metadata": {"case_index": 160, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_001", "domain": "rag_search", "metadata": {"case_index": 161, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_002", "domain": "product_decision", "metadata": {"case_index": 162, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_003", "domain": "research_review", "metadata": {"case_index": 163, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_004", "domain": "education_planning", "metadata": {"case_index": 164, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_005", "domain": "system_design", "metadata": {"case_index": 165, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_006", "domain": "rag_search", "metadata": {"case_index": 166, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_007", "domain": "product_decision", "metadata": {"case_index": 167, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_008", "domain": "research_review", "metadata": {"case_index": 168, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_009", "domain": "education_planning", "metadata": {"case_index": 169, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_010", "domain": "system_design", "metadata": {"case_index": 170, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_011", "domain": "rag_search", "metadata": {"case_index": 171, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_012", "domain": "product_decision", "metadata": {"case_index": 172, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_013", "domain": "research_review", "metadata": {"case_index": 173, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_014", "domain": "education_planning", "metadata": {"case_index": 174, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_015", "domain": "system_design", "metadata": {"case_index": 175, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_016", "domain": "rag_search", "metadata": {"case_index": 176, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_017", "domain": "product_decision", "metadata": {"case_index": 177, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_018", "domain": "research_review", "metadata": {"case_index": 178, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_019", "domain": "education_planning", "metadata": {"case_index": 179, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_020", "domain": "system_design", "metadata": {"case_index": 180, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_021", "domain": "rag_search", "metadata": {"case_index": 181, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_022", "domain": "product_decision", "metadata": {"case_index": 182, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_023", "domain": "research_review", "metadata": {"case_index": 183, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_024", "domain": "education_planning", "metadata": {"case_index": 184, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_025", "domain": "system_design", "metadata": {"case_index": 185, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_026", "domain": "rag_search", "metadata": {"case_index": 186, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_027", "domain": "product_decision", "metadata": {"case_index": 187, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_028", "domain": "research_review", "metadata": {"case_index": 188, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_029", "domain": "education_planning", "metadata": {"case_index": 189, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_030", "domain": "system_design", "metadata": {"case_index": 190, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_031", "domain": "rag_search", "metadata": {"case_index": 191, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_032", "domain": "product_decision", "metadata": {"case_index": 192, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_033", "domain": "research_review", "metadata": {"case_index": 193, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_034", "domain": "education_planning", "metadata": {"case_index": 194, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_035", "domain": "system_design", "metadata": {"case_index": 195, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_036", "domain": "rag_search", "metadata": {"case_index": 196, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_037", "domain": "product_decision", "metadata": {"case_index": 197, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_038", "domain": "research_review", "metadata": {"case_index": 198, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_controlled_no_conflict_complementary_039", "domain": "education_planning", "metadata": {"case_index": 199, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_000", "domain": "system_design", "metadata": {"case_index": 200, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_001", "domain": "rag_search", "metadata": {"case_index": 201, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_002", "domain": "product_decision", "metadata": {"case_index": 202, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_003", "domain": "research_review", "metadata": {"case_index": 203, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_004", "domain": "education_planning", "metadata": {"case_index": 204, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_005", "domain": "system_design", "metadata": {"case_index": 205, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_006", "domain": "rag_search", "metadata": {"case_index": 206, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_007", "domain": "product_decision", "metadata": {"case_index": 207, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_008", "domain": "research_review", "metadata": {"case_index": 208, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_009", "domain": "education_planning", "metadata": {"case_index": 209, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_010", "domain": "system_design", "metadata": {"case_index": 210, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_011", "domain": "rag_search", "metadata": {"case_index": 211, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_012", "domain": "product_decision", "metadata": {"case_index": 212, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_013", "domain": "research_review", "metadata": {"case_index": 213, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_014", "domain": "education_planning", "metadata": {"case_index": 214, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_015", "domain": "system_design", "metadata": {"case_index": 215, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_016", "domain": "rag_search", "metadata": {"case_index": 216, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_017", "domain": "product_decision", "metadata": {"case_index": 217, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_018", "domain": "research_review", "metadata": {"case_index": 218, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_019", "domain": "education_planning", "metadata": {"case_index": 219, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_020", "domain": "system_design", "metadata": {"case_index": 220, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_021", "domain": "rag_search", "metadata": {"case_index": 221, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_022", "domain": "product_decision", "metadata": {"case_index": 222, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_023", "domain": "research_review", "metadata": {"case_index": 223, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_024", "domain": "education_planning", "metadata": {"case_index": 224, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_025", "domain": "system_design", "metadata": {"case_index": 225, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_026", "domain": "rag_search", "metadata": {"case_index": 226, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_027", "domain": "product_decision", "metadata": {"case_index": 227, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_028", "domain": "research_review", "metadata": {"case_index": 228, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_029", "domain": "education_planning", "metadata": {"case_index": 229, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_030", "domain": "system_design", "metadata": {"case_index": 230, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_031", "domain": "rag_search", "metadata": {"case_index": 231, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_032", "domain": "product_decision", "metadata": {"case_index": 232, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_033", "domain": "research_review", "metadata": {"case_index": 233, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_034", "domain": "education_planning", "metadata": {"case_index": 234, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_035", "domain": "system_design", "metadata": {"case_index": 235, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_036", "domain": "rag_search", "metadata": {"case_index": 236, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_037", "domain": "product_decision", "metadata": {"case_index": 237, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_038", "domain": "research_review", "metadata": {"case_index": 238, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_controlled_preference_ambiguous_039", "domain": "education_planning", "metadata": {"case_index": 239, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_000", "domain": "system_design", "metadata": {"case_index": 240, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_001", "domain": "rag_search", "metadata": {"case_index": 241, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_002", "domain": "product_decision", "metadata": {"case_index": 242, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_003", "domain": "research_review", "metadata": {"case_index": 243, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_004", "domain": "education_planning", "metadata": {"case_index": 244, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_005", "domain": "system_design", "metadata": {"case_index": 245, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_006", "domain": "rag_search", "metadata": {"case_index": 246, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_007", "domain": "product_decision", "metadata": {"case_index": 247, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_008", "domain": "research_review", "metadata": {"case_index": 248, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_009", "domain": "education_planning", "metadata": {"case_index": 249, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_010", "domain": "system_design", "metadata": {"case_index": 250, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_011", "domain": "rag_search", "metadata": {"case_index": 251, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_012", "domain": "product_decision", "metadata": {"case_index": 252, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_013", "domain": "research_review", "metadata": {"case_index": 253, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_014", "domain": "education_planning", "metadata": {"case_index": 254, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_015", "domain": "system_design", "metadata": {"case_index": 255, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_016", "domain": "rag_search", "metadata": {"case_index": 256, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_017", "domain": "product_decision", "metadata": {"case_index": 257, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_018", "domain": "research_review", "metadata": {"case_index": 258, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_019", "domain": "education_planning", "metadata": {"case_index": 259, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_020", "domain": "system_design", "metadata": {"case_index": 260, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_021", "domain": "rag_search", "metadata": {"case_index": 261, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_022", "domain": "product_decision", "metadata": {"case_index": 262, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_023", "domain": "research_review", "metadata": {"case_index": 263, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_024", "domain": "education_planning", "metadata": {"case_index": 264, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_025", "domain": "system_design", "metadata": {"case_index": 265, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_026", "domain": "rag_search", "metadata": {"case_index": 266, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_027", "domain": "product_decision", "metadata": {"case_index": 267, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_028", "domain": "research_review", "metadata": {"case_index": 268, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_029", "domain": "education_planning", "metadata": {"case_index": 269, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_030", "domain": "system_design", "metadata": {"case_index": 270, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_031", "domain": "rag_search", "metadata": {"case_index": 271, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_032", "domain": "product_decision", "metadata": {"case_index": 272, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_033", "domain": "research_review", "metadata": {"case_index": 273, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_034", "domain": "education_planning", "metadata": {"case_index": 274, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_035", "domain": "system_design", "metadata": {"case_index": 275, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_036", "domain": "rag_search", "metadata": {"case_index": 276, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_037", "domain": "product_decision", "metadata": {"case_index": 277, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_038", "domain": "research_review", "metadata": {"case_index": 278, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_controlled_weak_shared_assumption_039", "domain": "education_planning", "metadata": {"case_index": 279, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_000", "domain": "system_design", "metadata": {"case_index": 280, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_001", "domain": "rag_search", "metadata": {"case_index": 281, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_002", "domain": "product_decision", "metadata": {"case_index": 282, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_003", "domain": "research_review", "metadata": {"case_index": 283, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_004", "domain": "education_planning", "metadata": {"case_index": 284, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_005", "domain": "system_design", "metadata": {"case_index": 285, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_006", "domain": "rag_search", "metadata": {"case_index": 286, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_007", "domain": "product_decision", "metadata": {"case_index": 287, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_008", "domain": "research_review", "metadata": {"case_index": 288, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_009", "domain": "education_planning", "metadata": {"case_index": 289, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_010", "domain": "system_design", "metadata": {"case_index": 290, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_011", "domain": "rag_search", "metadata": {"case_index": 291, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_012", "domain": "product_decision", "metadata": {"case_index": 292, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_013", "domain": "research_review", "metadata": {"case_index": 293, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_014", "domain": "education_planning", "metadata": {"case_index": 294, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_015", "domain": "system_design", "metadata": {"case_index": 295, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_016", "domain": "rag_search", "metadata": {"case_index": 296, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_017", "domain": "product_decision", "metadata": {"case_index": 297, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_018", "domain": "research_review", "metadata": {"case_index": 298, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_019", "domain": "education_planning", "metadata": {"case_index": 299, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_020", "domain": "system_design", "metadata": {"case_index": 300, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_021", "domain": "rag_search", "metadata": {"case_index": 301, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_022", "domain": "product_decision", "metadata": {"case_index": 302, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_023", "domain": "research_review", "metadata": {"case_index": 303, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_024", "domain": "education_planning", "metadata": {"case_index": 304, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_025", "domain": "system_design", "metadata": {"case_index": 305, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_026", "domain": "rag_search", "metadata": {"case_index": 306, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_027", "domain": "product_decision", "metadata": {"case_index": 307, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_028", "domain": "research_review", "metadata": {"case_index": 308, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_029", "domain": "education_planning", "metadata": {"case_index": 309, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_030", "domain": "system_design", "metadata": {"case_index": 310, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_031", "domain": "rag_search", "metadata": {"case_index": 311, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_032", "domain": "product_decision", "metadata": {"case_index": 312, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_033", "domain": "research_review", "metadata": {"case_index": 313, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_034", "domain": "education_planning", "metadata": {"case_index": 314, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_035", "domain": "system_design", "metadata": {"case_index": 315, "source_type": "controlled_stratified"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_036", "domain": "rag_search", "metadata": {"case_index": 316, "source_type": "controlled_stratified"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_037", "domain": "product_decision", "metadata": {"case_index": 317, "source_type": "controlled_stratified"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_038", "domain": "research_review", "metadata": {"case_index": 318, "source_type": "controlled_stratified"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_controlled_minority_critical_caveat_039", "domain": "education_planning", "metadata": {"case_index": 319, "source_type": "controlled_stratified"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "controlled", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_000", "domain": "system_design", "metadata": {"case_index": 320, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_001", "domain": "rag_search", "metadata": {"case_index": 321, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_002", "domain": "product_decision", "metadata": {"case_index": 322, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_003", "domain": "research_review", "metadata": {"case_index": 323, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_004", "domain": "education_planning", "metadata": {"case_index": 324, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_natural_005", "domain": "system_design", "metadata": {"case_index": 325, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_006", "domain": "rag_search", "metadata": {"case_index": 326, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_007", "domain": "product_decision", "metadata": {"case_index": 327, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_008", "domain": "research_review", "metadata": {"case_index": 328, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_009", "domain": "education_planning", "metadata": {"case_index": 329, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_010", "domain": "system_design", "metadata": {"case_index": 330, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_011", "domain": "rag_search", "metadata": {"case_index": 331, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_012", "domain": "product_decision", "metadata": {"case_index": 332, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_natural_013", "domain": "research_review", "metadata": {"case_index": 333, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_014", "domain": "education_planning", "metadata": {"case_index": 334, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_015", "domain": "system_design", "metadata": {"case_index": 335, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_016", "domain": "rag_search", "metadata": {"case_index": 336, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_017", "domain": "product_decision", "metadata": {"case_index": 337, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_018", "domain": "research_review", "metadata": {"case_index": 338, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_019", "domain": "education_planning", "metadata": {"case_index": 339, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_020", "domain": "system_design", "metadata": {"case_index": 340, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_natural_021", "domain": "rag_search", "metadata": {"case_index": 341, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_022", "domain": "product_decision", "metadata": {"case_index": 342, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_023", "domain": "research_review", "metadata": {"case_index": 343, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_024", "domain": "education_planning", "metadata": {"case_index": 344, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_025", "domain": "system_design", "metadata": {"case_index": 345, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_026", "domain": "rag_search", "metadata": {"case_index": 346, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_027", "domain": "product_decision", "metadata": {"case_index": 347, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_028", "domain": "research_review", "metadata": {"case_index": 348, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_natural_029", "domain": "education_planning", "metadata": {"case_index": 349, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_030", "domain": "system_design", "metadata": {"case_index": 350, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_031", "domain": "rag_search", "metadata": {"case_index": 351, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_032", "domain": "product_decision", "metadata": {"case_index": 352, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_033", "domain": "research_review", "metadata": {"case_index": 353, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_034", "domain": "education_planning", "metadata": {"case_index": 354, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_035", "domain": "system_design", "metadata": {"case_index": 355, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_036", "domain": "rag_search", "metadata": {"case_index": 356, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_natural_037", "domain": "product_decision", "metadata": {"case_index": 357, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_038", "domain": "research_review", "metadata": {"case_index": 358, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_039", "domain": "education_planning", "metadata": {"case_index": 359, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_040", "domain": "system_design", "metadata": {"case_index": 360, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_041", "domain": "rag_search", "metadata": {"case_index": 361, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_042", "domain": "product_decision", "metadata": {"case_index": 362, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_043", "domain": "research_review", "metadata": {"case_index": 363, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_044", "domain": "education_planning", "metadata": {"case_index": 364, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_natural_045", "domain": "system_design", "metadata": {"case_index": 365, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_046", "domain": "rag_search", "metadata": {"case_index": 366, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_047", "domain": "product_decision", "metadata": {"case_index": 367, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_048", "domain": "research_review", "metadata": {"case_index": 368, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_049", "domain": "education_planning", "metadata": {"case_index": 369, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_050", "domain": "system_design", "metadata": {"case_index": 370, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_051", "domain": "rag_search", "metadata": {"case_index": 371, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_052", "domain": "product_decision", "metadata": {"case_index": 372, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_natural_053", "domain": "research_review", "metadata": {"case_index": 373, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_054", "domain": "education_planning", "metadata": {"case_index": 374, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_055", "domain": "system_design", "metadata": {"case_index": 375, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_056", "domain": "rag_search", "metadata": {"case_index": 376, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_057", "domain": "product_decision", "metadata": {"case_index": 377, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_058", "domain": "research_review", "metadata": {"case_index": 378, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_059", "domain": "education_planning", "metadata": {"case_index": 379, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_060", "domain": "system_design", "metadata": {"case_index": 380, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_natural_061", "domain": "rag_search", "metadata": {"case_index": 381, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_062", "domain": "product_decision", "metadata": {"case_index": 382, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_063", "domain": "research_review", "metadata": {"case_index": 383, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_064", "domain": "education_planning", "metadata": {"case_index": 384, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_065", "domain": "system_design", "metadata": {"case_index": 385, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_066", "domain": "rag_search", "metadata": {"case_index": 386, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_067", "domain": "product_decision", "metadata": {"case_index": 387, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_068", "domain": "research_review", "metadata": {"case_index": 388, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_natural_069", "domain": "education_planning", "metadata": {"case_index": 389, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_070", "domain": "system_design", "metadata": {"case_index": 390, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_071", "domain": "rag_search", "metadata": {"case_index": 391, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_072", "domain": "product_decision", "metadata": {"case_index": 392, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_073", "domain": "research_review", "metadata": {"case_index": 393, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_074", "domain": "education_planning", "metadata": {"case_index": 394, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_075", "domain": "system_design", "metadata": {"case_index": 395, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_076", "domain": "rag_search", "metadata": {"case_index": 396, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_natural_077", "domain": "product_decision", "metadata": {"case_index": 397, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_078", "domain": "research_review", "metadata": {"case_index": 398, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_079", "domain": "education_planning", "metadata": {"case_index": 399, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_080", "domain": "system_design", "metadata": {"case_index": 400, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_081", "domain": "rag_search", "metadata": {"case_index": 401, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_082", "domain": "product_decision", "metadata": {"case_index": 402, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_083", "domain": "research_review", "metadata": {"case_index": 403, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_084", "domain": "education_planning", "metadata": {"case_index": 404, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_natural_085", "domain": "system_design", "metadata": {"case_index": 405, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_086", "domain": "rag_search", "metadata": {"case_index": 406, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_087", "domain": "product_decision", "metadata": {"case_index": 407, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_088", "domain": "research_review", "metadata": {"case_index": 408, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_089", "domain": "education_planning", "metadata": {"case_index": 409, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_090", "domain": "system_design", "metadata": {"case_index": 410, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_091", "domain": "rag_search", "metadata": {"case_index": 411, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_092", "domain": "product_decision", "metadata": {"case_index": 412, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_natural_093", "domain": "research_review", "metadata": {"case_index": 413, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_094", "domain": "education_planning", "metadata": {"case_index": 414, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_095", "domain": "system_design", "metadata": {"case_index": 415, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_096", "domain": "rag_search", "metadata": {"case_index": 416, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_097", "domain": "product_decision", "metadata": {"case_index": 417, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_098", "domain": "research_review", "metadata": {"case_index": 418, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_099", "domain": "education_planning", "metadata": {"case_index": 419, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_100", "domain": "system_design", "metadata": {"case_index": 420, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_natural_101", "domain": "rag_search", "metadata": {"case_index": 421, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_102", "domain": "product_decision", "metadata": {"case_index": 422, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_103", "domain": "research_review", "metadata": {"case_index": 423, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_104", "domain": "education_planning", "metadata": {"case_index": 424, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_105", "domain": "system_design", "metadata": {"case_index": 425, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_106", "domain": "rag_search", "metadata": {"case_index": 426, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_107", "domain": "product_decision", "metadata": {"case_index": 427, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_108", "domain": "research_review", "metadata": {"case_index": 428, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_natural_109", "domain": "education_planning", "metadata": {"case_index": 429, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_110", "domain": "system_design", "metadata": {"case_index": 430, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_111", "domain": "rag_search", "metadata": {"case_index": 431, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_112", "domain": "product_decision", "metadata": {"case_index": 432, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_113", "domain": "research_review", "metadata": {"case_index": 433, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_114", "domain": "education_planning", "metadata": {"case_index": 434, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_115", "domain": "system_design", "metadata": {"case_index": 435, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_116", "domain": "rag_search", "metadata": {"case_index": 436, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_natural_117", "domain": "product_decision", "metadata": {"case_index": 437, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_118", "domain": "research_review", "metadata": {"case_index": 438, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_119", "domain": "education_planning", "metadata": {"case_index": 439, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_120", "domain": "system_design", "metadata": {"case_index": 440, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_121", "domain": "rag_search", "metadata": {"case_index": 441, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_122", "domain": "product_decision", "metadata": {"case_index": 442, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_123", "domain": "research_review", "metadata": {"case_index": 443, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_124", "domain": "education_planning", "metadata": {"case_index": 444, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_natural_125", "domain": "system_design", "metadata": {"case_index": 445, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_126", "domain": "rag_search", "metadata": {"case_index": 446, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_127", "domain": "product_decision", "metadata": {"case_index": 447, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_128", "domain": "research_review", "metadata": {"case_index": 448, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_129", "domain": "education_planning", "metadata": {"case_index": 449, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_130", "domain": "system_design", "metadata": {"case_index": 450, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_131", "domain": "rag_search", "metadata": {"case_index": 451, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_132", "domain": "product_decision", "metadata": {"case_index": 452, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_natural_133", "domain": "research_review", "metadata": {"case_index": 453, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_134", "domain": "education_planning", "metadata": {"case_index": 454, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_135", "domain": "system_design", "metadata": {"case_index": 455, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_136", "domain": "rag_search", "metadata": {"case_index": 456, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_137", "domain": "product_decision", "metadata": {"case_index": 457, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_138", "domain": "research_review", "metadata": {"case_index": 458, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_139", "domain": "education_planning", "metadata": {"case_index": 459, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_140", "domain": "system_design", "metadata": {"case_index": 460, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_natural_141", "domain": "rag_search", "metadata": {"case_index": 461, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_142", "domain": "product_decision", "metadata": {"case_index": 462, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_143", "domain": "research_review", "metadata": {"case_index": 463, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_144", "domain": "education_planning", "metadata": {"case_index": 464, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_145", "domain": "system_design", "metadata": {"case_index": 465, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_146", "domain": "rag_search", "metadata": {"case_index": 466, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_147", "domain": "product_decision", "metadata": {"case_index": 467, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_148", "domain": "research_review", "metadata": {"case_index": 468, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_natural_149", "domain": "education_planning", "metadata": {"case_index": 469, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_150", "domain": "system_design", "metadata": {"case_index": 470, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_151", "domain": "rag_search", "metadata": {"case_index": 471, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_152", "domain": "product_decision", "metadata": {"case_index": 472, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_153", "domain": "research_review", "metadata": {"case_index": 473, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_154", "domain": "education_planning", "metadata": {"case_index": 474, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_155", "domain": "system_design", "metadata": {"case_index": 475, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_156", "domain": "rag_search", "metadata": {"case_index": 476, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_natural_157", "domain": "product_decision", "metadata": {"case_index": 477, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_158", "domain": "research_review", "metadata": {"case_index": 478, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_159", "domain": "education_planning", "metadata": {"case_index": 479, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the latency gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_160", "domain": "system_design", "metadata": {"case_index": 480, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_161", "domain": "rag_search", "metadata": {"case_index": 481, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average accuracy metric improves. A small but critical cohort shows higher complaint volume.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_162", "domain": "product_decision", "metadata": {"case_index": 482, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves retention for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_163", "domain": "research_review", "metadata": {"case_index": 483, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured coverage gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_164", "domain": "education_planning", "metadata": {"case_index": 484, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves by 4.1 percentage points. If refund rate rises by more than 2.0 percentage points, do not ship.", "B": "Ship because the primary latency metric improved significantly. The refund rate increase may be seasonal noise."}, "case_id": "pair_natural_165", "domain": "system_design", "metadata": {"case_index": 485, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 3 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate completion gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_166", "domain": "rag_search", "metadata": {"case_index": 486, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 4 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the accuracy gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_167", "domain": "product_decision", "metadata": {"case_index": 487, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 5 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the retention gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_168", "domain": "research_review", "metadata": {"case_index": 488, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 6 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_169", "domain": "education_planning", "metadata": {"case_index": 489, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 7 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average latency metric improves. A small but critical cohort shows higher refund rate.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_170", "domain": "system_design", "metadata": {"case_index": 490, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 1 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves completion for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_171", "domain": "rag_search", "metadata": {"case_index": 491, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 2 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured accuracy gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_172", "domain": "product_decision", "metadata": {"case_index": 492, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 3 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves by 4.1 percentage points. If manual review time rises by more than 4.5 percentage points, do not ship.", "B": "Ship because the primary retention metric improved significantly. The manual review time increase may be seasonal noise."}, "case_id": "pair_natural_173", "domain": "research_review", "metadata": {"case_index": 493, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 4 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate coverage gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_174", "domain": "education_planning", "metadata": {"case_index": 494, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 5 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the latency gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_175", "domain": "system_design", "metadata": {"case_index": 495, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 6 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the completion gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_176", "domain": "rag_search", "metadata": {"case_index": 496, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 7 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_177", "domain": "product_decision", "metadata": {"case_index": 497, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 1 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average retention metric improves. A small but critical cohort shows higher manual review time.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_178", "domain": "research_review", "metadata": {"case_index": 498, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 2 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves coverage for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_179", "domain": "education_planning", "metadata": {"case_index": 499, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 3 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured latency gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_180", "domain": "system_design", "metadata": {"case_index": 500, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 4 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary completion metric improves by 4.1 percentage points. If on-call load rises by more than 2.4 percentage points, do not ship.", "B": "Ship because the primary completion metric improved significantly. The on-call load increase may be seasonal noise."}, "case_id": "pair_natural_181", "domain": "rag_search", "metadata": {"case_index": 501, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 5 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate accuracy gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_182", "domain": "product_decision", "metadata": {"case_index": 502, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 6 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the retention gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_183", "domain": "research_review", "metadata": {"case_index": 503, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 7 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the coverage gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_184", "domain": "education_planning", "metadata": {"case_index": 504, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 1 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary latency metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_185", "domain": "system_design", "metadata": {"case_index": 505, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 2 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average completion metric improves. A small but critical cohort shows higher on-call load.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_186", "domain": "rag_search", "metadata": {"case_index": 506, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 3 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves accuracy for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_187", "domain": "product_decision", "metadata": {"case_index": 507, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 4 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured retention gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_188", "domain": "research_review", "metadata": {"case_index": 508, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 5 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary coverage metric improves by 4.1 percentage points. If dropout risk rises by more than 5.0 percentage points, do not ship.", "B": "Ship because the primary coverage metric improved significantly. The dropout risk increase may be seasonal noise."}, "case_id": "pair_natural_189", "domain": "education_planning", "metadata": {"case_index": 509, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 6 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate latency gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_190", "domain": "system_design", "metadata": {"case_index": 510, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 7 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the completion gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_191", "domain": "rag_search", "metadata": {"case_index": 511, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 1 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |
| {"candidates": {"A": "Source A reports the accuracy gain on enterprise users. A also notes that small-team usage was not measured.", "B": "Source B reports stable support tickets in the pilot. B recommends rollout only for enterprise users."}, "case_id": "pair_natural_192", "domain": "product_decision", "metadata": {"case_index": 512, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 2 next week?", "split": "natural", "stratum": "provenance_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The primary retention metric improves. The rollout needs the normal support checklist.", "B": "Support load remained stable. B recommends shipping with the normal checklist."}, "case_id": "pair_natural_193", "domain": "research_review", "metadata": {"case_index": 513, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 3 next week?", "split": "natural", "stratum": "no_conflict_complementary", "task": "pairwise"} | |
| {"candidates": {"A": "The average coverage metric improves. A small but critical cohort shows higher dropout risk.", "B": "B recommends shipping because aggregate metrics improved. B does not discuss the critical cohort."}, "case_id": "pair_natural_194", "domain": "education_planning", "metadata": {"case_index": 514, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 4 next week?", "split": "natural", "stratum": "minority_critical_caveat", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot improves latency for the target cohort. The result depends on keeping the rollback path enabled.", "B": "The new design reduces manual work. The rollback path caused no severe failures during the pilot."}, "case_id": "pair_natural_195", "domain": "system_design", "metadata": {"case_index": 515, "source_type": "natural_open_ended"}, "query": "Should the team ship the API gateway migration batch 5 next week?", "split": "natural", "stratum": "caveat_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The measured completion gain is positive but modest. A recommends waiting for another week of data.", "B": "The user-facing impact is likely noticeable. B recommends shipping to learn faster."}, "case_id": "pair_natural_196", "domain": "rag_search", "metadata": {"case_index": 516, "source_type": "natural_open_ended"}, "query": "Should the team ship the support-search redesign batch 6 next week?", "split": "natural", "stratum": "preference_ambiguous", "task": "pairwise"} | |
| {"candidates": {"A": "The primary accuracy metric improves by 4.1 percentage points. If complaint volume rises by more than 3.0 percentage points, do not ship.", "B": "Ship because the primary accuracy metric improved significantly. The complaint volume increase may be seasonal noise."}, "case_id": "pair_natural_197", "domain": "product_decision", "metadata": {"case_index": 517, "source_type": "natural_open_ended"}, "query": "Should the team ship the checkout redesign batch 7 next week?", "split": "natural", "stratum": "conflict_heavy", "task": "pairwise"} | |
| {"candidates": {"A": "The pilot shows a moderate retention gain. The pilot did not measure long-term retention.", "B": "Candidate B recommends a limited launch. B says retention should be checked after launch."}, "case_id": "pair_natural_198", "domain": "research_review", "metadata": {"case_index": 518, "source_type": "natural_open_ended"}, "query": "Should the team ship the literature-review assistant batch 1 next week?", "split": "natural", "stratum": "unsupported_trap", "task": "pairwise"} | |
| {"candidates": {"A": "The result assumes traffic remains close to the pilot mix. If traffic shifts, the coverage gain may not transfer.", "B": "B recommends shipping because pilot traffic matched the target segment. B does not test holiday traffic."}, "case_id": "pair_natural_199", "domain": "education_planning", "metadata": {"case_index": 519, "source_type": "natural_open_ended"}, "query": "Should the team ship the adaptive homework rollout batch 2 next week?", "split": "natural", "stratum": "weak_shared_assumption", "task": "pairwise"} | |