Answers are grounded in this paper and cite their sources.
Crowdsourced news platforms rely on open contribution, so their content quality is vulnerable to malicious contributors—astroturfers (“water armies”), sockpuppets, and on–off sleeper accounts—that can poison content and manipulate the agenda at low cost. We formalize contributor governance as a mechanism-design problem that jointly combines a differentiated reputation-scoring model, a reputation-driven role-promotion hierarchy (editor, reviewer, aggregator, committee), and a reputation-weighted supermajority consensus rule. Using a discounted dynamic payoff model with an explicit multi-layer detection technology and evolutionary game theory, we prove that honest contribution is a Nash equilibrium under a penalty–quality–work ordering and a detection-dependent incentive-compatibility condition. Within a mean-field replicator model, we derive an explicit phase-transition threshold beyond which content quality collapses, with the tolerable fraction increasing in the penalty and the detection probability. We show that reputation filtering extends the one-third weighted-voting bound: when a fraction s of malicious accounts are sleepers and their selection-weighted average committee weight is κ times that of honest contributors, the expected-weight benchmark tolerates a total malicious fraction below 1/(1+2sκ) with a strict margin, which reduces to 1/(2s+1) when κ=1. We characterize the promotion threshold and the cold-start/Sybil trade-off. Numerical experiments comparing against no-governance, consensus-only, and EigenTrust-style reputation baselines confirm that the joint design combines weighted-voting safety with an extended tolerance.
No takes yet. Share an insight, caveat, or question.
Ding et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: