Resourced Authority A Mechanism-Design Model for Participatory Governance of Deployed AI Agents

Ganesh Ghalme; Praphul Chandra; Sujit Gujar

REVIEW 4 major objections 5 minor 14 references

Resourced Authority A Mechanism-Design Model for Participatory Governance of Deployed AI Agents

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper models participatory AI governance as a two-sided vote over compute budgets, proving that authorization fires exactly when breadth-weighted support clears the opposition and a quorum floor.

desk verdict Serious mechanism-design framework with a real proof gap in the main theorem; worth referee time, but don't rely on Theorem 5.7 as proven. read the letter →

arxiv 2608.06353 v1 pith:5EASQ5RS submitted 2026-08-06 cs.GT cs.AIcs.MA

Praphul Chandra , Sujit Gujar , Ganesh Ghalme This is my paper

classification cs.GTcs.AIcs.MA MSC 91B03

keywords mechanismdesignAIgovernancequadraticfundingprovision-pointmechanismspredictionmarketscomputehardware-enabledattestation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The authors are trying to establish that a deployed AI agent can be governed continuously by its human stakeholders through resource allocation rather than through training or alignment: each governance period, stakeholders contribute in a human-anchored currency to either an authorize or a halt market, and the mechanism decides whether to let the agent run by metering its compute budget. The central result is that, in an efficient equilibrium, the agent is authorized exactly when the breadth-weighted support of beneficiaries clears both a quorum floor and a start margin over the breadth-weighted opposition, so authorization tracks the effective number and intensity of backers, not their wealth. If this is right, governance and enforcement collapse into a single artifact, a signed compute license that hardware can enforce, which makes the decision self-enforcing without requiring continuous network contact. The result matters because it turns the claim that compute is a governance lever into a checkable theorem, while carving out the class of agents for which the approach fits.

What carries the argument

The load-bearing objects are the quadratic-funding aggregator $\phi(c_1,\dots,c_m)=(\sum_j \sqrt{c_j})^2$, which turns raw contributions into breadth-weighted effective supports $S^+$ and $S^-$; the two-threshold gate with hysteresis (start margin $\kappa_{\rm start}$, halt margin $\kappa_{\rm halt}$, quorum floor $H_0$); and the coupling map $\rho$ that converts net support into a released compute budget $\beta=\min(\Gamma,\rho(S^+-S^-,\mu))$, with $\Gamma$ an exogenous safety ceiling. The authorization decision is realized as a signed compute license, so the slow governance loop issues a metered license and a fast hardware loop enforces it. The theorem works by comparing each side's breadth-weighted capacity $\Phi^\pm(\theta)=\phi(\{w_i^\pm\})$ and the effective breadth $n_{\rm eff}^\pm=\Phi^\pm/\vartheta^\pm$, the participation ratio of backers; that ratio is what lets breadth overturn wealth. The securities layer, inherited from the provision-point-mechanism lineage, supplies early-commitment bonuses that make contribute-on-arrival a weakly dominant timing choice and supplies the revelation assumption on stake capacities.

What would settle it

Run a cold-start generation with complete information and known types where $\Phi^+(\theta) \geq H_0$ and $\Phi^+(\theta)-\Phi^-(\theta) \geq \kappa_{\rm start}$, and check every coalition-undominated equilibrium: if any such equilibrium withholds authorization, or any equilibrium authorizes when the condition fails, Theorem 5.7 is false. A direct test of the weakest premise would compare observed stakes to the assumed capacities $w_i^+=\theta_i-b_i$ and $w_j^-=|\theta_j|-b_j$; systematic deviation would break the theorem's support bounds.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is Theorem 5.7: in a cold-start generation, under complete information, the securities Conditions 1–7, the bounded-stake condition, the inherited securities-revelation assumption (A3), and the cold-start run-feasibility condition, the governance game has an efficient, objection-robust, coalition-undominated perfect Bayesian equilibrium, and in every such equilibrium the authorization decision is $D_g=1$ exactly when $\Phi^+(\theta) \geq H_0$ and $\Phi^+(\theta)-\Phi^-(\theta) \geq \kappa_{\rm start}$, written equivalently as $n_{\rm eff}^+\vartheta^+ \geq H_0$ and $n_{\rm eff}^+\vartheta^+ - n_{\rm eff}^-\vartheta^- \geq \kappa_{\rm start}$. Whenever authorization is granted, the released budget satisfies $\beta \geq h_0$, so the decision is physically runnable. The theorem is accompanied by a sided incentive-compatibility result with a stated asymmetry: the halt side is incentive-compatible unconditionally, while the authorize side requires each supporter's stake plus bonus not to exceed the supporter's own valuation. The authors also show that the breadth-weighted comparison can disagree with the raw valuation comparison, so broad support can authorize what concentrated wealth would reject.

Load-bearing premise

The load-bearing premise is that the two-sided securities layer induces each stakeholder to commit exactly its valuation minus the early-commitment bonus to its own side, and the theorem further assumes everyone's valuations are known and the efficient equilibrium is selected rather than the all-abstain profile.

Editorial extensions

If this is right

Authorization can be self-enforcing: the gate's decision is issued as a signed compute license carrying a metered budget, so enforcement happens in hardware at inference time without reaching the governance layer.
The gate tracks breadth rather than wealth: because $\Phi^\pm = n_{\rm eff}^\pm \vartheta^\pm$, broad low-stake support can authorize what concentrated wealth would reject, and a single large backer cannot carry authorization against broad opposition.
Making the halt side incentive-compatible is unconditional, while the authorize side needs the bounded-stake condition $c_i+b_i \leq \theta_i$ to stop a supporter from betting on failure for the refund bonus.
The mechanism is scoped to club-or-commons-good agents — bounded stakeholder community, reversible and compute-scaled impact, genuine contestation, repeated operation, and attestable outcomes — with catastrophic or irreversible harms excluded by the certified safety ceiling.
The paper identifies manipulation of the governing electorate by the governed agent as the central open problem: the agent can groom the reputation signals feeding stakeholder beliefs, so the open target is a mechanism whose gate tracks the true $\theta$ rather than the induced $\tilde{\theta}$ within $\delta$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

This reader's inference: the two-domain decoupling — human-anchored vote in one currency, metered resource in another, joined only by a bounded coupling map — should transfer to any excludable, quantifiable resource a community is entitled to authorize, such as API quotas, data access, or shared compute allocations.
This reader's inference: the theorem's weakest spot can be probed directly in a lab by running the cold-start game with known types and comparing actual stakes to the assumed capacities $w_i^+=\theta_i-b_i$ and $w_j^-=|\theta_j|-b_j$; systematic deviations would demand a different revelation foundation.
This reader's inference: because the open manipulation problem works through the reputation signal feeding $\epsilon_i$, an implementable near-term version should maximize the share of transfers that resolve on attested outcomes, making sentiment less valuable to groom.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this paper is worth engaging with, but don't cite Theorem 5.7 as proven yet. The model is genuinely new — QF aggregator replacing linear sums in the PPS/Damle two-sided provision-point framework, decoupled governance currency and compute, signed-license instantiation, safe harbor, generations, and an honestly stated open manipulation problem. The provenance table is a model of scholarly hygiene; it tells you exactly what is new and what is inherited. Proposition 5.2, Proposition 5.4, and Lemma 3.10 are fine. The scope discussion (club/commons-good agents, attested outcomes, trusted V_soft) is candid and useful.\n\nThe problem is the sufficiency half of Theorem 5.7. The proof puts objectors at full reservation capacity and calls that a weak best response because of the "bounded-loss refund." Under the paper's own Definitions 3.7, 3.13, and 3.14, that refund applies only when D_g=0. On the D_g=1 branch the objector's escrow is spent: u_j=θ_j−z_j versus θ_j at z_j=0. Unless harm compensation λ_j makes up the difference, a non-pivotal objector strictly prefers zero. So the constructed profile is not a PBE unless λ_j is large and guaranteed, which the proof neither assumes nor argues. The necessity direction, via coalitional mobilization, is more plausible; the theorem may be salvageable by putting objectors at zero and beneficiaries exactly on the threshold so every positive contributor is pivotal. But as written, the central characterization is not established. Related soft spot: assumption (A3), the securities-revelation capacity bounds, is inherited from PPS/Damle rather than derived, and Theorem 5.7 leans on it. Complete information and the efficient-equilibrium selection are stated, so that is honest, but it narrows the force.\n\nI would not desk reject. The framework, the taxonomy, and the open manipulation problem are valuable; the flaw is a fixable proof gap, not a bad idea. It deserves a serious referee and a revision request. I might bring it to reading group to talk through whether the sufficiency can be repaired.

Referee Report

4 major / 5 minor

Summary. The paper proposes a mechanism-design model for participatory governance of a deployed AI agent, in which verified stakeholders contribute in a governance currency on separate provision and rejection markets, a quadratic-funding aggregator forms breadth-weighted supports, a two-threshold gate with hysteresis determines authorization, and a coupling map releases a compute budget capped by an exogenous safety ceiling and realized as a signed hardware license. The authors prove auxiliary incentive properties (sided incentive compatibility, early commitment, run-feasibility) and a central 'breadth-weighted authorization' theorem, then identify manipulation of the governing electorate by the governed agent as the central open problem.

Significance. The paper tackles a timely and important problem and makes a genuine attempt to translate mechanism-design concepts into an AI-governance setting, with an explicit hardware instantiation and a candid discussion of trust assumptions. The auxiliary lemmas (Lemma 3.10, Proposition 5.4) are correct given the assumptions, and the honest treatment of the verifier and adoption routes is a strength. However, the central theorem's sufficiency proof contains a fundamental error, and the theorem's universal claim rests on an undefined equilibrium refinement and an unproven inherited assumption. The main contribution is therefore not established.

major comments (4)

[§5.2, Theorem 5.7 (sufficiency)] The constructed profile places objectors at capacity and asserts that on the D_g=1 branch a non-pivotal objector is payoff-neutral by the bounded-loss refund of Conditions 6–7. This contradicts Definitions 3.13–3.14: escrow is spent when D_g=1, securities and refunds pay only when D_g=0, and u_j(D_g=1)=θ_j−z_j versus θ_j at z_j=0. Hence a non-pivotal objector strictly prefers z_j=0, and the same logic applies to non-pivotal beneficiaries. The profile is therefore not sequentially rational, so the existence half of the theorem is not established, and the 'iff' characterization fails in the sufficiency direction.
[§5.2, Theorem 5.7 ('in every such equilibrium')] The proof constructs a single profile with D_g=1 when the breadth conditions hold, but it does not prove that every objection-robust, coalition-undominated PBE yields D_g=1. The all-abstain and free-ride-to-failure profiles are excluded by assertion, not by a formal argument, so the universal quantification over equilibria is unsupported.
[§5.2, Assumption (A3)] The capacity bounds w_i^+ = θ_i−b_i and w_j^- = |θ_j|−b_j are stated as an inherited standing assumption rather than derived. Because the present model changes the gate (QF aggregation, hysteresis) and the transfers (escrow spending on provision, refunds only on rejection), the PPS/Damle revelation result does not automatically apply; the theorem's bounds S+≤Φ+ and S-≤Φ- and the entire characterization depend on A3, so a derivation or a verifiable citation is required.
[§5.2, equilibrium refinement] The terms 'objection-robust' and 'coalition-undominated' are used in the statement and proof of Theorem 5.7 but are never defined. Without formal definitions, the claims that the constructed profile is objection-robust and that all such equilibria agree on D_g cannot be checked.

minor comments (5)

[§3, first paragraph] 'governence currency domain' should be 'governance currency domain'.
[Figures 7 and 8] Figure 7 and Figure 8 are duplicates with identical captions and content; one should be removed or replaced.
[§5.3] 'T arget 5.9' should be 'Target 5.9'.
[Abstract] 'We also introduce several challenges addressing manipulation of governing electorate by the governed agents' is grammatically awkward; consider 'We also introduce several challenges to addressing manipulation of the governing electorate by the governed agent.'
[Definition 3.14] The harm-compensation term 1[z_i>0]·\bar H·λ_i is not multiplied by D_g; if \bar H=1 requires a run (D_g=1), the expression is consistent only if \bar H=0 whenever D_g=0, which should be stated explicitly.

Circularity Check

1 steps flagged · score 4.0 of 10

No fitted-data circularity; results are conditional on stated assumptions, but the central authorization theorem imports its load-bearing securities-revelation premise (A3) from the authors' own PPS/Damle lineage without re-deriving it.

self citation load bearing [§5.2, Assumption (A3) and Theorem 5.7]
"(A3) Securities revelation (inherited). In its selected undominated equilibrium, the two-sided securities layer (Definition 3.7, Conditions 1–7) induces each stakeholder to back its own side up to a reservation capacity equal to its gross type net of the early-commitment bonus: a beneficiary i∈P will stake up to w_i^+ = θ_i − b_i on the authorize side, and a harmed j∈N up to w_j^- = |θ_j| − b_j on the halt side, and no more... This is the two-sided PPS/Damle revelation property adapted to our reduced payoff; §5.1 supplies its sided and timing content."

Theorem 5.7's proof begins: 'a profile is summarized by magnitudes x_i ∈ [0, w_i^+] ... where the caps are (A3).' A3 is not proved in the present model; it is attributed to the authors' own PPS/Damle prior work ('This is the two-sided PPS/Damle revelation property adapted to our reduced payoff; §5.1 supplies its sided and timing content'). The cited §5.1 propositions establish only which side an agent chooses and when it contributes, not the capacity magnitude w_i^±. Because Definition 5.5 defines Φ± as the QF aggregate of exactly these A3 capacities, and Definition 3.8 already gates on S+≥H0 and S+−S-≥κstart, the theorem's authorization condition reads the gate at the capacity bounds imported from the self-cited lineage.

full rationale

The paper contains no fitted-input-called-prediction circularity: no parameter is fitted to a subset of outcomes and then 'predicted,' and the other lemmas (monotonicity, sided incentive compatibility, early commitment, run-feasibility) follow from the stated model definitions and explicit assumptions. The sole circularity-adjacent issue is Assumption (A3), labeled 'inherited' and attributed to the authors' own PPS/Damle lineage, which supplies the capacity bounds that make Theorem 5.7's Φ± condition coincide with the gate's S± condition; this is a load-bearing self-citation with independent game-theoretic content around it, hence score 4 rather than 6+. Separately, the sufficiency direction's claim that a non-pivotal objector at capacity is 'payoff-neutral by the bounded-loss refund of Conditions 6–7' appears inconsistent with Definition 3.14's u_j(D_g=1)=θ_j−z_j, but that is a proof-soundness concern, not a circularity, and is not scored here.

Assumptions & free parameters 0 free parameters · 5 assumptions · 2 invented entities

No numeric values are fitted to data; H0, κstart, κhalt, µ, Γ, h0 are symbolic design/exogenous parameters. The theorems are parameter-free in the sense that they hold for ranges, not fitted constants. The load-bearing axioms are the Sybil-resistance substrate, the unproven securities-revelation behavior (A3), complete information, a trusted semantic verifier, and the availability of hardware compute enforcement.

assumptions (5)

domain assumption Verified human distinctness is an assumed Sybil-resistant substrate, not modeled.
Stated before Definition 3.1: 'Individual human distinctness is an assumed Sybil-resistant substrate and not modeled as part of the mechanism.' The QF aggregator is vulnerable to Sybils without it.
domain assumption Securities revelation (A3): the two-sided securities layer induces each stakeholder to back its own side up to reservation capacity w_i = θ_i - b_i or |θ_j| - b_j in its selected undominated equilibrium.
Stated in §5.2: 'we inherit from the provision-point lineage and state as a standing assumption rather than re-derive.' Theorem 5.7 rests on these capacity caps.
domain assumption Complete information for the characterization: types are commonly known in Theorem 5.7.
Stated in the Remarks after Theorem 5.7: 'Complete information (commonly-known types) is assumed for the characterization, exactly as in the core-implementation results...'.
domain assumption A credible semantic verifier V_soft exists and resolves harm findings through a challengeable process.
Section 4.4: 'the model's accountability half rests on a trusted semantic verifier' and V_soft is 'genuinely trusted'. Liability, securities settlement, and decision-market resolution depend on it.
domain assumption Compute is a governable resource: detectable, excludable, quantifiable; hardware licenses and attestation work as described.
Section 1 and §4.1 map the model to offline licensing, flexHEG, and workload attestation; the self-enforcement claim is conditional on these technologies.

invented entities (2)

Governance currency
purpose: Human-anchored contribution unit, distinct from compute, used on provision and rejection markets to express support or opposition.
Introduced in Definition 3.2 and the decoupling invariant (Invariant 1). It is a model-internal numeraire with no falsifiable handle outside the paper.
Semantic verifier V_soft
purpose: Adjudicates harm and outcome semantics that hardware cannot attest; required for liability, securities settlement, and belief rewards.
Defined in Definition 3.12, discussed in §4.4 as the 'genuinely trusted component'. No independent evidence of such an oracle is given; the paper treats it as a scope condition.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Resourced Authority A Mechanism-Design Model for Participatory Governance of Deployed AI Agents." pith.science (2026). https://pith.science/paper/5EASQ5RS

@misc{pith2026260806353,
  author       = {Pith},
  title        = {Pith review of: Resourced Authority A Mechanism-Design Model for Participatory Governance of Deployed AI Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5EASQ5RS}},
  note         = {Machine review of arXiv:2608.06353}
}

read the original abstract

We give a formal mechanism design model for the continuous participatory governance of a deployed AI agent. The mechanism is built on the principle that governance should control an AI agent through resource allocation so as to make authorization self enforcing via compute budgets. The mechanism seeks to establish the Safe AI paradigm that compute is an effective governance lever. We situate our work as a compliance or commons overlay on a deployer. One governance period is an extensive form game in which verified human stakeholders arrive sequentially and contribute, on a provision or a rejection market, in a governance currency that is deliberately distinct from the agents compute. A funding aggregator turns raw contributions into breadth weighted effective supports - a two threshold gate with hysteresis converts net support into a binary authorization that, through a coupling map bounded by an exogenously certified safety ceiling, releases a metered compute budget - realized in hardware as a signed compute license so that the decision is self-enforcing. We characterize the class of agents the mechanism can govern and isolate manipulation of the governing electorate by the governed agent as the central open problem. We also introduce several challenges addressing manipulation of governing electorate by the governed agents.

Figures

Figures reproduced from arXiv: 2608.06353 by the authors.

**Figure 2.** The two-sided QF gate (worked example). Five supporters contributing 4 each yield [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 3.** The coupling ρ and capability non-amplification. Released compute β is zero below the start margin, rises with net support (steeper for a larger compute subsidy µ), and saturates at the exogenous ceiling Γ. The region above Γ is unreachable by the market (Invariant 2). Proof. If Dg = 1 then S + − S − ≥ κhalt (in the start branch S + − S − ≥ κstart ≥ κhalt), so by monotonicity ρ(S + − S −, µ) ≥ ρ(κhalt, µ) ≥ h 0 , an… view at source ↗

Figures from the paper (6 more)

**Figure 4.** Figure 4: The layered stack, mapped to the variables. Four governance layers — preferences, [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

**Figure 5.** Figure 5: One generation as an extensive-form game. Stakeholders arrive sequentially, report a [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

**Figure 6.** Figure 6: Hardware instantiation and the two timescales. A slow, online governance loop (per [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

**Figure 7.** Figure 7: Breadth-weighted vs. valuation-weighted authorization. The gate fires on the breadth [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]

**Figure 8.** Figure 8: Breadth-weighted vs. valuation-weighted authorization. The gate fires on the breadth [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]

**Figure 9.** Figure 9: The central open problem. The governed agent can induce [PITH_FULL_IMAGE:figures/full_fig_p021_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

14 extracted references · 9 canonical work pages

[1]

Aarne, O., Fist, T., & Withers, C. (2024). Secure, Governable Chips. Center for a New American Security

work page 2024
[2]

Bagnoli, M., & Lipman, B. L. (1989). Provision of public goods: fully implementing the core through private contributions.Review of Economic Studies, 56(4), 583–601

work page 1989
[3]

Buterin, V., Hitzig, Z., & Weyl, E. G. (2019). A flexible design for funding public goods. Management Science, 65(11), 5171–5187

work page 2019
[4]

Chandra, P., Gujar, S., & Narahari, Y. (2016). Crowdfunding public projects with provision point: a prediction market approach.ECAI, 778–786

work page 2016
[5]

Civic Crowdfunding for Agents with Negative Valuations and Agents with Asymmetric Beliefs

Damle, S., Moti, M. H., Chandra, P., & Gujar, S. (2019). Civic crowdfunding for agents with negative valuations and agents with asymmetric beliefs. arXiv:1905.11324

work page Pith review arXiv 2019
[6]

Greenblatt, R., Shlegeris, B., Sachan, K., & Roger, F. (2024). AI control: improving safety despite intentional subversion.ICML. arXiv:2312.06942

arXiv 2024
[7]

Heim, L., et al. (2025). Hardware-enabled mechanisms for verifying responsible AI devel- opment. arXiv:2505.03742

arXiv 2025
[8]

G., & Erichsen, L

Miller, J., Weyl, E. G., & Erichsen, L. (2022). Beyond collusion resistance: leveraging social information for plural funding and voting. SSRN 4311507

work page 2022

Show all 14 references

[9]

Petrie, J. (2025). Embedded off-switches for AI compute. arXiv:2509.07637

2025 arXiv
[10]

Petrie, J., Aarne, O., Ammann, N., & Dalrymple, D. (2025). Flexible hardware-enabled guarantees for AI compute (flexHEG). arXiv:2506.15093

2025 arXiv
[11]

Prelec, D. (2004). A Bayesian truth serum for subjective data.Science, 306(5695), 462–466

2004
[12]

Sastry, G., Heim, L., Belfield, H., Anderljung, M., Brundage, M., et al. (2024). Computing power and the governance of artificial intelligence. arXiv:2402.08797

2024 arXiv
[13]

Witkowski, J., & Parkes, D. C. (2012). A robust Bayesian truth serum for small populations. AAAI

2012
[14]

Zubrickas, R. (2014). The provision point mechanism with refund bonuses.Journal of Public Economics, 120, 231–234. 22

2014

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Aarne, O., Fist, T., & Withers, C. (2024). Secure, Governable Chips. Center for a New American Security

work page 2024

[2] [2]

Bagnoli, M., & Lipman, B. L. (1989). Provision of public goods: fully implementing the core through private contributions.Review of Economic Studies, 56(4), 583–601

work page 1989

[3] [3]

Buterin, V., Hitzig, Z., & Weyl, E. G. (2019). A flexible design for funding public goods. Management Science, 65(11), 5171–5187

work page 2019

[4] [4]

Chandra, P., Gujar, S., & Narahari, Y. (2016). Crowdfunding public projects with provision point: a prediction market approach.ECAI, 778–786

work page 2016

[5] [5]

Civic Crowdfunding for Agents with Negative Valuations and Agents with Asymmetric Beliefs

Damle, S., Moti, M. H., Chandra, P., & Gujar, S. (2019). Civic crowdfunding for agents with negative valuations and agents with asymmetric beliefs. arXiv:1905.11324

work page Pith review arXiv 2019

[6] [6]

Greenblatt, R., Shlegeris, B., Sachan, K., & Roger, F. (2024). AI control: improving safety despite intentional subversion.ICML. arXiv:2312.06942

arXiv 2024

[7] [7]

Heim, L., et al. (2025). Hardware-enabled mechanisms for verifying responsible AI devel- opment. arXiv:2505.03742

arXiv 2025

[8] [8]

G., & Erichsen, L

Miller, J., Weyl, E. G., & Erichsen, L. (2022). Beyond collusion resistance: leveraging social information for plural funding and voting. SSRN 4311507

work page 2022