{
  "project": "personalized-user-feedback-agent-evolution",
  "scope_note": "AHEAD intentionally disabled for the current novel-scene implementation.",
  "ahead_enabled": false,
  "entries": [
    {
      "topic": "self_evolve",
      "round": 1,
      "method_name": "explicit-only baseline",
      "dominant_failure_mode": "implicit_feedback_blind_spots",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 5,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "baseline",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": null,
        "headline": "Baseline round; no previous round exists yet."
      },
      "selected_next_patch": "Add dwell/fast-swipe/continue calibration and content-feature evidence for silent reader dissatisfaction.",
      "visualization": "self_evolve/round_01.html",
      "metrics_file": "eval/self_evolve_round_01.json"
    },
    {
      "topic": "self_evolve",
      "round": 2,
      "method_name": "implicit calibrated updater",
      "dominant_failure_mode": "scope_overgeneralization",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 5,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": 0.0119,
        "headline": "future_probe_win_rate changed +0.0119 vs previous round."
      },
      "selected_next_patch": "Add verifier to distinguish current-story/current-arc updates from durable global user preferences.",
      "visualization": "self_evolve/round_02.html",
      "metrics_file": "eval/self_evolve_round_02.json"
    },
    {
      "topic": "self_evolve",
      "round": 3,
      "method_name": "state-diff target router",
      "dominant_failure_mode": "scope_overgeneralization",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 5,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "target_f1",
        "primary_delta": 0.3679,
        "headline": "target_f1 changed +0.3679 vs previous round."
      },
      "selected_next_patch": "Add verifier to distinguish current-story/current-arc updates from durable global user preferences.",
      "visualization": "self_evolve/round_03.html",
      "metrics_file": "eval/self_evolve_round_03.json"
    },
    {
      "topic": "self_evolve",
      "round": 4,
      "method_name": "scope verifier",
      "dominant_failure_mode": "framework_target_gap_after_scope_fix",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "overgeneralization_rate",
        "primary_delta": -0.8553,
        "headline": "overgeneralization_rate changed -0.8553 vs previous round."
      },
      "selected_next_patch": "Run a verifier pass that re-routes each surviving dimension to all primary framework targets.",
      "visualization": "self_evolve/round_04.html",
      "metrics_file": "eval/self_evolve_round_04.json"
    },
    {
      "topic": "self_evolve",
      "round": 5,
      "method_name": "self-debugged PUMA-lite",
      "dominant_failure_mode": "residual_future_probe_gap",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "target_f1",
        "primary_delta": 0.1106,
        "headline": "target_f1 changed +0.1106 vs previous round."
      },
      "selected_next_patch": "Use future-probe-aware reranking and hard anti-overgeneralization probes; keep AHEAD disabled.",
      "visualization": "self_evolve/round_05.html",
      "metrics_file": "eval/self_evolve_round_05.json"
    },
    {
      "topic": "self_evolve",
      "round": 6,
      "method_name": "Round 6 neutral-control evaluator",
      "dominant_failure_mode": "residual_future_probe_gap",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": 0.0083,
        "headline": "future_probe_win_rate changed +0.0083 vs previous round."
      },
      "selected_next_patch": "Use future-probe-aware reranking and hard anti-overgeneralization probes; keep AHEAD disabled.",
      "visualization": "self_evolve/round_06.html",
      "metrics_file": "eval/self_evolve_round_06.json"
    },
    {
      "topic": "self_evolve",
      "round": 7,
      "method_name": "Round 7 positive-signal gate",
      "dominant_failure_mode": "residual_future_probe_gap",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": 0.008,
        "headline": "future_probe_win_rate changed +0.0080 vs previous round."
      },
      "selected_next_patch": "Use future-probe-aware reranking and hard anti-overgeneralization probes; keep AHEAD disabled.",
      "visualization": "self_evolve/round_07.html",
      "metrics_file": "eval/self_evolve_round_07.json"
    },
    {
      "topic": "self_evolve",
      "round": 8,
      "method_name": "Round 8 confidence-calibrated router",
      "dominant_failure_mode": "residual_future_probe_gap",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": 0.008,
        "headline": "future_probe_win_rate changed +0.0080 vs previous round."
      },
      "selected_next_patch": "Use future-probe-aware reranking and hard anti-overgeneralization probes; keep AHEAD disabled.",
      "visualization": "self_evolve/round_08.html",
      "metrics_file": "eval/self_evolve_round_08.json"
    },
    {
      "topic": "self_evolve",
      "round": 9,
      "method_name": "Round 9 future-probe-aware reranker",
      "dominant_failure_mode": "residual_future_probe_gap",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": 0.0126,
        "headline": "future_probe_win_rate changed +0.0126 vs previous round."
      },
      "selected_next_patch": "Use future-probe-aware reranking and hard anti-overgeneralization probes; keep AHEAD disabled.",
      "visualization": "self_evolve/round_09.html",
      "metrics_file": "eval/self_evolve_round_09.json"
    },
    {
      "topic": "self_evolve",
      "round": 10,
      "method_name": "Round 10 consolidated PUMA-lite v2",
      "dominant_failure_mode": "residual_future_probe_gap",
      "evidence": {
        "missed_dimensions": 5,
        "missed_targets": 5,
        "overgeneralized_scope": 0,
        "extra_dimensions": 0,
        "neutral_errors": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "future_probe_win_rate",
        "primary_delta": 0.0052,
        "headline": "future_probe_win_rate changed +0.0052 vs previous round."
      },
      "selected_next_patch": "Use future-probe-aware reranking and hard anti-overgeneralization probes; keep AHEAD disabled.",
      "visualization": "self_evolve/round_10.html",
      "metrics_file": "eval/self_evolve_round_10.json"
    },
    {
      "topic": "long_horizon",
      "round": 1,
      "method_name": "final-turn blame baseline",
      "dominant_failure_mode": "temporal_credit_blame_errors",
      "evidence": {
        "step_miss": 5,
        "dimension_miss": 5,
        "low_rank_gold": 3
      },
      "comparison_to_previous": {
        "status": "baseline",
        "primary_metric": "culprit_step_accuracy",
        "primary_delta": null,
        "headline": "Baseline round; no previous round exists yet."
      },
      "selected_next_patch": "Move beyond final-turn blame by scanning trajectory-wide evidence and delayed-feedback text.",
      "visualization": "long_horizon/round_01.html",
      "metrics_file": "eval/long_horizon_round_01.json"
    },
    {
      "topic": "long_horizon",
      "round": 2,
      "method_name": "dimension evidence scan",
      "dominant_failure_mode": "hard_distractor_temporal_ambiguity",
      "evidence": {
        "step_miss": 5,
        "dimension_miss": 5,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "culprit_step_accuracy",
        "primary_delta": 0.65,
        "headline": "culprit_step_accuracy changed +0.6500 vs previous round."
      },
      "selected_next_patch": "Add temporal utility-drop windows so early weak distractors do not steal credit from the real culprit.",
      "visualization": "long_horizon/round_02.html",
      "metrics_file": "eval/long_horizon_round_02.json"
    },
    {
      "topic": "long_horizon",
      "round": 3,
      "method_name": "temporal utility window",
      "dominant_failure_mode": "culprit_dimension_evidence_gap",
      "evidence": {
        "step_miss": 5,
        "dimension_miss": 5,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "credit_calibration",
        "primary_delta": 0.2803,
        "headline": "credit_calibration changed +0.2803 vs previous round."
      },
      "selected_next_patch": "Fuse delayed-feedback dimension tokens with per-turn feature tags, severity, and repairability.",
      "visualization": "long_horizon/round_03.html",
      "metrics_file": "eval/long_horizon_round_03.json"
    },
    {
      "topic": "long_horizon",
      "round": 4,
      "method_name": "causal candidate scorer",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 5,
        "dimension_miss": 5,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "culprit_step_accuracy",
        "primary_delta": 0.025,
        "headline": "culprit_step_accuracy changed +0.0250 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_04.html",
      "metrics_file": "eval/long_horizon_round_04.json"
    },
    {
      "topic": "long_horizon",
      "round": 5,
      "method_name": "trace verifier + repair planner",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 5,
        "dimension_miss": 5,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "repair_gain",
        "primary_delta": 0.0137,
        "headline": "repair_gain changed +0.0137 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_05.html",
      "metrics_file": "eval/long_horizon_round_05.json"
    },
    {
      "topic": "long_horizon",
      "round": 6,
      "method_name": "Round 6 tie-aware trace logger",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 5,
        "dimension_miss": 5,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "repair_gain",
        "primary_delta": 0.0001,
        "headline": "repair_gain changed +0.0001 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_06.html",
      "metrics_file": "eval/long_horizon_round_06.json"
    },
    {
      "topic": "long_horizon",
      "round": 7,
      "method_name": "Round 7 verifier fallback",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 4,
        "dimension_miss": 4,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "culprit_step_accuracy",
        "primary_delta": 0.0084,
        "headline": "culprit_step_accuracy changed +0.0084 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_07.html",
      "metrics_file": "eval/long_horizon_round_07.json"
    },
    {
      "topic": "long_horizon",
      "round": 8,
      "method_name": "Round 8 margin-calibrated confidence",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 4,
        "dimension_miss": 4,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "repair_gain",
        "primary_delta": 0.0001,
        "headline": "repair_gain changed +0.0001 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_08.html",
      "metrics_file": "eval/long_horizon_round_08.json"
    },
    {
      "topic": "long_horizon",
      "round": 9,
      "method_name": "Round 9 partial-credit multi-cause scorer",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 4,
        "dimension_miss": 4,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "repair_gain",
        "primary_delta": 0.0001,
        "headline": "repair_gain changed +0.0001 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_09.html",
      "metrics_file": "eval/long_horizon_round_09.json"
    },
    {
      "topic": "long_horizon",
      "round": 10,
      "method_name": "Round 10 CREDIT-TRACE v2",
      "dominant_failure_mode": "multi_cause_trace_ambiguity",
      "evidence": {
        "step_miss": 3,
        "dimension_miss": 3,
        "low_rank_gold": 0
      },
      "comparison_to_previous": {
        "status": "improved",
        "primary_metric": "culprit_step_accuracy",
        "primary_delta": 0.0083,
        "headline": "culprit_step_accuracy changed +0.0083 vs previous round."
      },
      "selected_next_patch": "Keep CREDIT-TRACE verifier; next expansion should add multi-cause labels from real reader logs, not AHEAD.",
      "visualization": "long_horizon/round_10.html",
      "metrics_file": "eval/long_horizon_round_10.json"
    }
  ]
}