Reputation as Contextual Knowledge: Incentives and External Value in Truthful Blockchain Oracles

The primary strengths of blockchain systems come from strong guarantees of immutability and reliability, and these systems can be extended with programmatic logic through Smart Contracts. Many such programmatic use cases would benefit from the use of external data. However, the method of bringing that data onto the blockchain needs to be trustworthy and secure. Otherwise, the benefits of blockchain, namely distributed trust, no single points of failure, and immutability, would be at risk. Blockchain oracles are proposed as a conceptual solution to this problem. A blockchain oracle acts as a trusted intermediary to external data. In this work we analyze existing proposals for Discernible event oracle mechanisms, which seek to address oracle queries with a broad knowledgeable answering population, to characterize problems related to the incentives imposed by external value that depends on oracle outcomes. In doing so, we focus on continued challenges to voting based oracles, including trust limitations and the need to match questions to knowledgeable answering parties. To address these difficulties, we propose an extension to existing oracle protocols to utilize reputation as a tool to measure value and as a tool to associate questions with context. By providing a method to track contextual knowledge, our proposal allows for context-based query matching and enables higher probability of correctness for a given population size as well as stronger participation incentives.