News
An Epistemic Approach to Conflict-Free Replicated Data Types
Abstract
Optimistic replication enables highly available geo-distributed systems by allowing replicas to diverge temporarily and reconcile asynchronously. While celebrated conflict-free replicated data types (CRDTs) guarantee strong eventual convergence, their built-in conflict resolution mechanisms are often implicit and inflexible, limiting their applicability in scenarios requiring semantic transparency and user involvement. In this paper, we investigate the epistemic conflict model, a framework that enables explicit, flexible, and coordination-free conflict resolution while preserving convergence guarantees. Firstly, we show that state-based, operation-based and delta-CRDTs are naturally expressed in the epistemic model and provide a framework for their conflict-free representation. Secondly, we consider sets as a fundamental replicated data type that empowers more advanced replicated objects such as lists, graphs and maps, and define the respective conflict resolution procedures in the epistemic model, which are known as add-wins and remove-wins semantics in CRDTs. Thus, we demonstrate that the epistemic conflict model is sufficiently expressive to capture semantics of existing CRDT designs, while preserving interpretability.
Keywords
Edition
Proceedings of the Institute for System Programming, vol. 38, issue 5, 2026, pp. 37-50
ISSN 2220-6426 (Online), ISSN 2079-8156 (Print).
DOI: 10.15514/ISPRAS-2026-38(5)-3
For citation
Full text of the paper in pdf
Back to the contents of the volume