Start Here — Three Results
Three steps into Distributed Discovery
A contextual reading path from the atomic paradox to strategic attention and the sharing frontier. The order is editorial; evidence status remains in the claim ledger.
1 · Canonical upstream entry
Shared Discovery Paradox
Question: Can better pooled ranking produce worse group discovery?
Title result: in the atomic fixture, one pooled answer can compress several searches into duplication.
Strongest limitation: the separation depends on the declared signal, action, and protocol classes.
Evidence: sourced upstream exact result. Status: canonical upstream entry; no submission or peer-review status represented here.
Read the anchor · Study · Claim · Evidence · Next paper
2 · Strategic-attention family
The Incentive to Ignore
Question: Who should use a shared clue when independent private action routes remain?
Title result: one shared-clue reader maximizes discovery in the frozen role class.
Strongest limitation: unrestricted constant territorial policies can escape the evidence-responsive class.
Evidence: analytic theorems plus independently reproduced bounded classifications. Status: working paper; not submitted, not peer reviewed, no DOI.
Read the paper · Study · Claim · Evidence · Next paper
3 · Sharing-frontier family
When Does Information Sharing Improve Decentralized Discovery?
Question: When does aggregation improve faster than sharing removes independent rescue?
Title result: a residual-error criterion determines the sign of an adjacent sharing step under the registered rescue protocol.
Strongest limitation: the decentralized positive interval is selection-dependent; it is not an every-equilibrium theorem.
Evidence: analytic identities/theorems, independently reproduced bounded classifications, and a bounded null. Status: arXiv v1 preprint 2609.01814 [cs.AI; cs.GT]; not peer reviewed. The displayed DOI registration was pending at observation.
Read the paper · Study · Claim · Evidence · Next theorem-family paper
Translation layer
Related formulations
These formulations translate a canonical object without changing its scientific scope.
OR/MS
Shared Discovery Paradox
information improvement with action-allocation loss
atomic pinned model and protocol comparison
Economics / AGT
Shared Discovery Paradox
a decentralized information-action separation under changed protocol
not a violation of centralized Blackwell monotonicity
Multi-agent AI
Shared Discovery Paradox
shared context can collapse an agent ensemble onto one action
task analogy not evaluated agent evidence
Practitioner
Shared Discovery Paradox
a better shared answer can send everyone to the same wrong place
model illustration not field evidence
OR/MS
Incentive to Ignore / One-Reader theorem
optimal audience size under coverage loss
registered evidence-responsive role class
Economics / AGT
Incentive to Ignore / One-Reader theorem
welfare effect of selective public-information access
mechanical coverage and declared incentives
Multi-agent AI
Incentive to Ignore / One-Reader theorem
route a shared clue to one acting agent while others retain independent tasks
task analogy not agent evaluation
Practitioner
Incentive to Ignore / One-Reader theorem
not everyone who can read the same signal should act on it
scope-bound model statement
OR/MS
Information Sharing Frontier
error-contraction versus residual-coverage frontier
registered channels protocols and selection
Economics / AGT
Information Sharing Frontier
aggregation benefit net of decentralized action compression
selected positive result is not every-equilibrium
Multi-agent AI
Information Sharing Frontier
shared context helps only if better joint inference beats lost task diversity
theoretical mapping without model evaluation
Practitioner
Information Sharing Frontier
share when the map improves faster than the portfolio narrows
model-derived formulation
Open the complete contextual concordance