Pensieve: Decentralised Social Consensus for Community-Governed Knowledge Bases

Abstract

Decentralised communities increasingly need ways to generate, contest, and maintain shared knowledge without relying on central editorial authorities. Existing collaborative systems such as wikis support large-scale knowledge production, but typically depend on centralised moderation and governance, while blockchain consensus mechanisms are primarily designed for transactional state rather than the validation and revision of shared claims. This paper introduces Pensieve, a framework for decentralised epistemic governance in collaboratively governed knowledge bases. Pensieve combines structured submissions, contribution-weighted support, and stake-backed dispute escalation to govern the publication, revision, and contestation of knowledge entries over time. The paper makes three contributions. First, it presents Pensieve as a framework for decentralised epistemic governance designed specifically for collaboratively governed knowledge bases rather than for transactional ledgers alone. Second, it formalises a contribution-weighted model of item governance and dispute resolution, showing how reputational participation and economic stake may be combined to support knowledge revision that remains both dynamic and increasingly stable over time. Third, it examines the Pensieve instance on ECF.network as an early live implementation context through which the framework is being applied to the collaborative documentation and evaluation of Web3 projects. The case study provides descriptive evidence that the framework can support structured publication, community contribution, dispute handling, and AI-assisted but human-accountable input in practice, while also revealing unresolved challenges around referencing, sustained participation, postpublication revision, and long-term governance design. Pensieve is therefore offered not as a completed solution to decentralised knowledge governance, but as a research framework and live design experiment for coordinating collective knowledge production in human-AI systems.