Reputation and Disclosure in Dynamic Networks

I. Sebastian Buhai

arxiv: 2512.22987 · v3 · pith:DSVSYSZ4new · submitted 2025-12-28 · 💰 econ.TH

Reputation and Disclosure in Dynamic Networks

I. Sebastian Buhai This is my paper

Pith reviewed 2026-05-21 17:27 UTC · model grok-4.3

classification 💰 econ.TH

keywords disclosurepublicrecordbayesiancensoringcompactcustodianreputation

0 comments

The pith

In a dynamic network disclosure model, common-record censoring from retention generates network value missed by pairwise formation under reputational discipline.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The setup involves a sealed, dated record that travels through public hands but stays private until a final disclosure time. Public review dates exist where the current holder can choose to keep the record instead of passing it along. Keeping it is not neutral: it rules out certain possibilities about what the record contains, because if the content were something the holder would have shared, they would have done so. This changes what observers can infer using probability rules. The author separates who holds the record from who audits its content, allowing a clean update of beliefs that stays Gaussian but only on the possibilities still open after retentions. Strategies are assumed to be intervals, so the possible states can be tracked with a small number of boundary points. To check whether these strategies form stable outcomes where no one wants to deviate, the paper uses a quadratic version of the Ornstein-Uhlenbeck process as a benchmark. This shows that reputation concerns lead to resolution in finite time, with bounds on how far beliefs can stray. The key twist comes when the same record can travel along two different certified routes. Retention on one route changes what is still possible on the other route, even before the second route acts. This interaction produces extra value for the overall network structure that cannot be seen by looking only at pairs of connections.

Core claim

This common-record censoring creates a network value that pairwise formation can miss.

Load-bearing premise

The model assumes interval strategies under which the support of the restricted Gaussian reference likelihood is tracked by finitely many endpoints, allowing the Ornstein-Uhlenbeck quadratic benchmark to verify Markov perfect Bayesian equilibria with finite-time resolution.

Figures

Figures reproduced from arXiv: 2512.22987 by I. Sebastian Buhai.

**Figure 2.** Figure 2: Benchmark topologies for Theorem 3. In the ordered line, information must be relayed sequentially through both intermediaries. In the reputational star, each intermediary has a direct reporting link to the decision maker. Intuition: why the bottleneck does not switch with the sign of the news. Assumption (i) restricts attention to bias-aligned paths (Definition 1), so all intermediaries tilt the decision m… view at source ↗

read the original abstract

Public delay can be informative when the existence, custodian, and review dates of hard evidence are observed. I study a disclosure protocol in which a sealed record is docketed, held by a public custodian, and revealed only at terminal disclosure. At each review, retention is not silence: it rules out the states in which the holder would have relayed or disclosed. This censoring event yields an exact Bayesian filter. Under interval strategies, the public posterior is summarized by finitely many support endpoints. A compact reputation benchmark verifies Markov perfect Bayesian equilibria with such strategies and gives finite time resolution on compact unresolved slices. With two certified routes for the same record, retention on one route changes what remains feasible on the other before it acts. Common record censoring creates network value that pairwise formation can miss.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper claims common-record censoring across routes creates network value missed by pairwise models, but the finite-endpoint tracking of belief support may not hold once routes interact.

read the letter

This paper's key claim is that common-record censoring across two certified routes for the same record generates network value that pairwise formation misses. Retention on one route alters what is feasible on the other, creating interdependence through the shared information. The new element is the disclosure protocol where a dated sealed record moves through public custody with public review dates but terminal disclosure only. Retention censors by ruling out states where the holder would relay it. Bayes' rule then gives a restricted Gaussian likelihood, and under interval strategies the support is tracked by finitely many endpoints. The Ornstein-Uhlenbeck quadratic benchmark verifies Markov perfect Bayesian equilibria with finite-time resolution under reputational discipline. This approach does well by grounding the model in familiar tools from continuous-time stochastic processes and perfect Bayesian equilibrium concepts. It extends reputation models to dynamic networks in a structured way, and the tail bounds beyond compact support add some rigor to the analysis. The main soft spot is the assumption that interval strategies keep the support trackable by finitely many endpoints even with cross-route interactions. If retention on route A splits or shifts the support for route B, new endpoints could appear, and their number might grow with routes or dates. That would undermine the finite-time resolution and the benchmark's usefulness for verifying equilibria. The abstract does not detail how this is resolved, so the network-value result may depend on the strategy class holding without additional complications. This work is aimed at theorists in organizational economics and contract theory who model reputation and information flows in networks. Readers comfortable with stochastic differential equations and equilibrium verification in dynamic games will find the technical setup useful. It deserves a serious referee because the framing is fresh and the methods are standard enough to check, though the paper would benefit from explicit handling of the support evolution under interactions. I recommend putting it through peer review. The author can clarify the robustness of the finite-endpoint tracking in the full model.

Referee Report

2 major / 2 minor

Summary. The manuscript models a disclosure protocol in dynamic networks where a dated sealed record moves through public custody with public review dates and terminal disclosure only. Retention at review dates functions as a censoring event that restricts the support of a Gaussian reference likelihood obtained from Bayes' rule. Under interval strategies the support is tracked by finitely many endpoints. An Ornstein-Uhlenbeck quadratic benchmark is used to verify Markov perfect Bayesian equilibria within this strategy class and to obtain finite-time resolution under reputational discipline with tail bounds. The central result is that common-record censoring across two certified routes for the same record generates a network value that pairwise formation misses.

Significance. If the technical claims hold, particularly the preservation of finite-endpoint tracking and Markov perfection under multi-route interactions, the paper identifies a novel mechanism for network value arising from interdependent censoring and reputation. The Ornstein-Uhlenbeck benchmark supplies a concrete, tractable way to handle continuous-time belief dynamics and tail bounds, which is a methodological contribution that could extend to other models of dynamic information transmission and certification.

major comments (2)

[Model and Equilibrium Verification] The claim that interval strategies allow the restricted Gaussian reference likelihood to be tracked by finitely many endpoints (stated in the abstract and used for equilibrium verification) is load-bearing for finite-time resolution and the network-value comparison. Retention on one route altering feasible continuations on the other creates interdependent information sets; this can split or shift the support in ways that add endpoints, potentially violating the finite-tracking property and rendering the Ornstein-Uhlenbeck quadratic benchmark inapplicable. A formal argument showing that the number of endpoints remains finite despite cross-route events is required.
[Multi-Route Extension] The demonstration that common-record censoring creates value missed by pairwise formation (abstract) rests on equilibria remaining verifiable inside the maintained interval-strategy class. If cross-route censoring interactions cause the belief support to cease being a simple interval whose boundaries evolve continuously, the Markov-perfect Bayesian equilibrium characterization and the resulting network-value ranking would need re-derivation.

minor comments (2)

The abstract is information-dense; separating the single-record benchmark from the two-route extension into distinct sentences would improve readability.
[Model Description] Notation for the endpoints tracking the restricted support should be introduced with an explicit recursive definition or diagram to clarify how new endpoints are (or are not) generated at each review date.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive report. The concerns raised about the robustness of the finite-endpoint tracking property and the extension to multi-route interactions are well-taken. Below we provide point-by-point responses and indicate the revisions we plan to implement to address these issues.

read point-by-point responses

Referee: The claim that interval strategies allow the restricted Gaussian reference likelihood to be tracked by finitely many endpoints (stated in the abstract and used for equilibrium verification) is load-bearing for finite-time resolution and the network-value comparison. Retention on one route altering feasible continuations on the other creates interdependent information sets; this can split or shift the support in ways that add endpoints, potentially violating the finite-tracking property and rendering the Ornstein-Uhlenbeck quadratic benchmark inapplicable. A formal argument showing that the number of endpoints remains finite despite cross-route events is required.

Authors: We acknowledge that the manuscript would benefit from an explicit formal argument demonstrating that the finite number of endpoints is preserved under cross-route censoring. While the single-route case is handled by tracking the boundaries of the restricted support under interval strategies, the multi-route setting introduces interdependence through shared records. We argue that because retention on one route only eliminates states consistent with non-retention on that route, and the strategies are interval-based, the overall support remains a single interval whose endpoints evolve continuously. To make this rigorous, we will add a new lemma in the revised version that proves the preservation of the finite-endpoint property by showing that cross-route events correspond to additional restrictions that do not increase the number of boundary points beyond a finite set. This will also confirm the applicability of the Ornstein-Uhlenbeck quadratic benchmark for verifying Markov perfect Bayesian equilibria in the network setting. revision: yes
Referee: The demonstration that common-record censoring creates value missed by pairwise formation (abstract) rests on equilibria remaining verifiable inside the maintained interval-strategy class. If cross-route censoring interactions cause the belief support to cease being a simple interval whose boundaries evolve continuously, the Markov-perfect Bayesian equilibrium characterization and the resulting network-value ranking would need re-derivation.

Authors: The network-value result relies on the fact that common-record censoring generates additional reputational discipline that pairwise formation cannot replicate, even while keeping the belief dynamics within the interval-strategy class. We do not believe that cross-route interactions cause the support to cease being a simple interval; rather, the interdependent restrictions maintain the interval structure because the Gaussian reference likelihood is updated conditionally on the joint retention events. In the revision, we will include an extension of the equilibrium characterization proposition to explicitly cover the multi-route case, showing that the Markov perfection and the value comparison hold without re-derivation of the core tail bounds or finite-time resolution. This addresses the concern while preserving the central contribution. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation rests on explicit modeling assumptions and external benchmark

full rationale

The paper states that Bayes' rule yields a Gaussian reference likelihood restricted to remaining support, and that under interval strategies this support is tracked by finitely many endpoints. It then invokes an Ornstein-Uhlenbeck quadratic benchmark to verify Markov perfect Bayesian equilibria with finite-time resolution. These are presented as maintained assumptions that enable the subsequent comparison of network value under common-record censoring versus pairwise formation. No equation is shown to equal its own input by construction, no parameter is fitted to a subset and then relabeled as a prediction, and no load-bearing step reduces to a self-citation whose content is itself unverified. The finite-endpoint claim is an explicit restriction on strategy class rather than a derived result, and the network-value conclusion is asserted to follow directly from the cross-route feasibility interaction once the benchmark is applied. The derivation chain therefore remains self-contained against the stated assumptions.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 1 invented entities

The model rests on standard assumptions from dynamic Bayesian games and stochastic processes, with specific strategy restrictions and benchmark choices extracted from the abstract; no explicit free parameters or new invented entities with independent evidence are stated.

axioms (3)

domain assumption Players employ interval strategies in the disclosure game.
Abstract states that under interval strategies the support is tracked by finitely many endpoints.
domain assumption The Ornstein-Uhlenbeck quadratic benchmark verifies the Markov perfect Bayesian equilibria.
Used to establish finite-time resolution under reputational discipline.
domain assumption Custody is separated from audit information, allowing Bayes' rule to produce a Gaussian reference likelihood on remaining support.
Central to the updating mechanism described in the abstract.

invented entities (1)

Dated sealed record moving through public custody with terminal disclosure no independent evidence
purpose: To formalize a disclosure protocol where retention at review dates functions as a censoring event.
Core modeling device introduced to study reputational effects in networks.

pith-pipeline@v0.9.0 · 5668 in / 1672 out tokens · 150047 ms · 2026-05-21T17:27:03.465251+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

ladder representation: disclosure and clock choices are pinned down by finitely many posterior cutoffs... Ornstein-Uhlenbeck quadratic benchmark verifies Markov perfect Bayesian equilibria

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.