{
  "display": {
    "formal_suite": "TreasureBench",
    "historical_alias": "DiscoveryBench",
    "role": "formal-instrument",
    "subtitle": "TreasureBench: a benchmark for collective search under shared and private evidence."
  },
  "metrics": [
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected union discovery or registered coverage value.",
      "metric_id": "discovery",
      "required_observables": [
        "covered_outcomes"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability-or-coverage"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected correct actions divided by actions.",
      "metric_id": "average-action-quality",
      "required_observables": [
        "actions",
        "target_outcomes"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected number of distinct actions.",
      "metric_id": "distinct-actions",
      "required_observables": [
        "actions"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "actions"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Declared weighted union coverage.",
      "metric_id": "weighted-coverage",
      "required_observables": [
        "coverage_sets",
        "weights"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "coverage-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected successful hits beyond the first.",
      "metric_id": "redundant-hits",
      "required_observables": [
        "target_hits"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "hits"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Declared concentration of action shares.",
      "metric_id": "action-concentration",
      "required_observables": [
        "actions"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "index"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Declared concentration of source exposure.",
      "metric_id": "source-concentration",
      "required_observables": [
        "source_ids"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "index"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Model-defined reciprocal source concentration.",
      "metric_id": "effective-channels",
      "required_observables": [
        "source_ids",
        "channel_definition"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "channels"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Registered frontier minus protocol discovery.",
      "metric_id": "protocol-loss",
      "required_observables": [
        "frontier",
        "discovery"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "objective-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Smallest action budget recovering the comparator.",
      "metric_id": "recovery-budget",
      "required_observables": [
        "budget_frontier",
        "comparator"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "actions"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected declared source-acquisition cost.",
      "metric_id": "information-cost",
      "required_observables": [
        "source_choices",
        "costs"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "payoff-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected number of actions used.",
      "metric_id": "expected-actions",
      "required_observables": [
        "actions"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "actions"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected number of decision rounds.",
      "metric_id": "expected-rounds",
      "required_observables": [
        "rounds"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "rounds"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected declared private payoff.",
      "metric_id": "private-payoff",
      "required_observables": [
        "rewards",
        "costs"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "payoff-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Discovery value net of declared information and transfer costs.",
      "metric_id": "social-net-value",
      "required_observables": [
        "discovery",
        "information_cost",
        "transfer_budget"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "payoff-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected external transfer budget.",
      "metric_id": "transfer-budget",
      "required_observables": [
        "transfers"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "payoff-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Whether truthful reporting satisfies the registered incentive condition.",
      "metric_id": "truthfulness",
      "required_observables": [
        "reports",
        "deviations"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "boolean-or-margin"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Whether recommended action satisfies the registered incentive condition.",
      "metric_id": "obedience",
      "required_observables": [
        "recommendations",
        "deviations"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "boolean-or-margin"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Minimum truthful/obedient payoff advantage over registered deviations.",
      "metric_id": "strict-margin",
      "required_observables": [
        "deviation_payoffs"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "payoff-units"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Number of agents using the public signal.",
      "metric_id": "attention-count",
      "required_observables": [
        "attention_choices"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "agents"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Number of agents whose action uses the public signal.",
      "metric_id": "public-signal-user-count",
      "required_observables": [
        "actions",
        "public_signal"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "agents"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Discovery gain from the first public-signal user over private-only play.",
      "metric_id": "first-use-gain",
      "required_observables": [
        "discovery_frontier"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Discovery loss from duplicate public-signal use relative to one user.",
      "metric_id": "duplicate-use-loss",
      "required_observables": [
        "discovery_frontier"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Planner optimum minus the registered voluntary-attention equilibrium.",
      "metric_id": "attention-wedge",
      "required_observables": [
        "planner_optimum",
        "equilibrium_discovery"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Number of agents given access to the public signal.",
      "metric_id": "audience-size",
      "required_observables": [
        "information_permissions"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "agents"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Discovery loss relative to the optimal audience assignment.",
      "metric_id": "publicity-cost",
      "required_observables": [
        "audience_frontier"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Registered conditional policy used on disagreement.",
      "metric_id": "conditional-attention-category",
      "required_observables": [
        "contingent_policy"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "category"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Exact deterministic minimum-team planner discovery.",
      "metric_id": "planner-discovery",
      "required_observables": [
        "planner_frontier"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "probability"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Expected count of candidates meeting the declared threshold.",
      "metric_id": "expected-viable-candidates",
      "required_observables": [
        "occupancies",
        "threshold"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "candidates"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Registered games whose worst weak pure equilibrium has zero discovery.",
      "metric_id": "zero-worst-equilibrium-games",
      "required_observables": [
        "equilibrium_registry"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "games"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Registered games with no pairwise-strict-stable weak pure equilibrium.",
      "metric_id": "pair-instability-games",
      "required_observables": [
        "pair_deviations"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "games"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Registered games with no exact-size-tau-strict-stable weak equilibrium.",
      "metric_id": "tau-instability-games",
      "required_observables": [
        "tau_deviations"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "games"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Registered games where tied-mode uniform mixing is not an equilibrium.",
      "metric_id": "tied-mode-failure-games",
      "required_observables": [
        "mixed_action_payoffs"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "games"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Exact rows where the dynamic planner strictly exceeds autonomous play.",
      "metric_id": "planner-strict-gain-rows",
      "required_observables": [
        "dynamic_values"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "rows"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Fixed-budget cells where visibility lowers discovery and dispersion.",
      "metric_id": "visibility-joint-loss-cells",
      "required_observables": [
        "visibility_control"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "cells"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Cells where stopping strictly lowers expected actions.",
      "metric_id": "stopping-action-savings-cells",
      "required_observables": [
        "fixed_and_stopping_values"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "cells"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Mechanism rows attaining the minimum-team planner portfolio.",
      "metric_id": "planner-portfolio-rows",
      "required_observables": [
        "mechanism_registry"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "rows"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Rows passing the declared strict-member pair-deviation check.",
      "metric_id": "pair-stable-rows",
      "required_observables": [
        "pair_deviations"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "rows"
    },
    {
      "aggregation": "task-level; family profiles and Pareto vectors only",
      "definition": "Exact pure-equilibrium count or fixture-ordered count vector.",
      "metric_id": "equilibrium-multiplicity",
      "required_observables": [
        "equilibrium_registry"
      ],
      "schema_version": "discoverybench-metric-v3",
      "scope": "registered task observables only",
      "status": "exact-or-estimated-per-result",
      "units": "count-or-vector"
    }
  ],
  "run_id": "20260722T054447Z_DD-010_d265e480_6930915b02",
  "schema_version": 3
}
