Collective search under dispersed information
Browse studies of how evidence, roles, incentives, timing, and action allocation shape collective search.
Start with three results · See how studies fit into theorem families, papers, and the living synthesis.
26 studies shown
Which definitions and accounting quantities separate information quality, action assignment, protocol loss, redundancy, and incentives while retaining the Shared Discovery Paradox as a…
Definitions, canonical reproduction, and exact pooled-frontier evidence.
Is clue-following team-optimal? When can role-dependent signal policies or territories improve discovery? How do best-known and certified policies vary with \(M,N,p\), and what can be…
Exact private-team role and alignment results under zero communication.
Can Blackwell-more-informative public disclosure lower equilibrium discovery by concentrating autonomous actions?
Exact disclosure and equilibrium-selection catalogue for a bounded fixture.
Which source-network features predict discovery and action concentration beyond pairwise signal correlation?
Exact source-network censuses and a heterogeneous pairwise-moment counterexample.
How does outcome feedback change useful diversification, recovery budgets, and the allocation of parallel versus sequential actions?
Exact perfect-elimination batching baseline with distinct terminal, action, and round objectives.
Which information-frontier, protocol-loss, redundancy, and recovery concepts survive under overlapping coverage and non-binary value?
Exact weighted-union coverage frontiers with minimized ranking and greedy failures.
Which mechanisms implement informative reporting and differentiated search under budget balance, collusion, and private-action constraints?
Exact bounded mechanism censuses separating weak, strict, observable, and subsidized implementation.
Can a normalized mechanism combining an exact proper information score with coverage and obedience components strictly implement truthful reporting and a differentiated search…
Exact normalized joint-score frontier separating observable strict implementation from the hidden-action boundary.
Which observable data identify or bound protocol loss, recovery budgets, duplicate action, source concentration, effective channels, and latent copying?
Seeded synthetic estimator-recovery audit with explicit nonidentification and matching-error failures.
When two agents can buy independent evidence or use a free common source, can private source incentives retain common-source use when one independent channel raises net group discovery?
Exact synthetic common-versus-independent evidence source-choice grid.
Under equal prize splitting, how does the number of privately chosen independent sources compare with the planner-optimal number in the registered finite common-source model?
Exact bounded N-agent source-count accounting census with independent direct enumeration.
For arbitrary finite `N`, what thresholds characterize unilateral independent- source acquisition and planner acquisition, when does an all-common equilibrium coexist with a planner…
General-N source-choice threshold theorem with a bounded independent audit and exact interior counterexample.
What exact performance, cost, timing, and incentive tradeoffs appear across the coherent cells of a bounded aligned discovery-architecture registry?
Exact aligned 20-cell architecture atlas with 268 explicit validity exclusions and bounded Pareto comparisons.
Can a versioned benchmark exactly reproduce the project's validated bounded fixtures while enforcing protocol information capabilities and reporting task-level metric vectors without…
TreasureBench is the formal public suite for the versioned exact golden benchmark with enforced protocol information boundaries; frozen DiscoveryBench identifiers and all five quarantined Agents v1 pilots remain unchanged, while AO-0011 completes a wholly fresh 500-pairing v5 engineering pilot under forward-only provider-outcome policy v3 and unchanged protocol-validity policy v2 without publishing task-level performance or creating scientific evidence.
1. Do participants under-acquire costly independent evidence? 2. Does attribution or priority protection increase truthful disclosure? 3. Does public action visibility increase action…
Preregistration-ready bounded design and conditional synthetic power package; no human data.
When a shared clue is available through an ex-ante access gate, how many searchers should condition their actions on it, and how does that compare with privately optimal attention under an…
Two exact attention theorems and an independently reproduced 175-cell reward census for ex-ante shared-signal access.
Should an accurate shared signal be broadcast to everyone, delivered to a selected subset, privately assigned to one reader, or garbled?
Exact binding-audience and garbling theorems, a budget-balanced public implementation, and an independently reproduced voluntary-use census.
Does the value of deliberate non-use survive when agents can condition actions jointly on their private clue and the shared clue?
Exact conditional-policy planner theorem, independently reproduced equilibrium census, and a preserved larger-class counterexample.
When prior actions reveal likely use of a shared clue, when should later agents search against the public action and how do private and social incentives differ?
Verified full-credit equilibrium and exact dynamic census with a bounded visibility-herding negative result.
How should posterior evidence be converted into a portfolio of actions when a candidate is discovered only if at least tau agents choose it?
Verified minimum-viable-team theorem and independently reproduced exact threshold and occupancy census.
Which decentralized threshold-search occupancies are Nash equilibria, which survive small strict-improvement coalitions, and how far can their discovery value lie below the deterministic…
Verified pure and tied-mode equilibrium boundaries with independently reproduced welfare and small-coalition censuses.
Which bounded institutions implement two size-two teams on the top two posterior candidates without collapsing all four agents onto the common mode, and which implementation properties come…
Verified exact comparison of ten common-posterior institutions separates authoritative, individually obedient, pair-stable, and inefficient team formation.
Can finite channels with the same one-person Bayes action accuracy and the same direct private-portfolio discovery have different pooled top-`L` discovery profiles and recovery budgets?
Independently reproduced exact pooled top-L profiles for five finite channels expose a same-accuracy counterexample and baseline-specific recovery budgets.
For recommendation-like signals, what happens to group discovery as one more independent searcher is absorbed into a shared-answer block that takes one common action while all remaining…
Verified exact aggregation-rescue identity and point-channel monotonicity theorem, plus independently reproduced point and five-channel profiles.
Inside the declared finite target-symmetric protocol class, when does one more shared signal improve, preserve, or reduce group discovery, and how many centralized pooled actions recover…
Verified the error-contraction and full-capacity theorems and independently reproduced the complete 177-scenario exact registry, witnesses, recovery budgets, and bounded mixed-curve null.
Can sharing a realized pair of correlated private signals strictly increase both group discovery and each anonymous agent's expected equal-split payoff by allowing posterior-contingent…
Exact private and posterior-only shared symmetric equilibria yield a certified strict positive-sharing interval at p=3/5, with a preserved every-equilibrium negative result.