Towards an epistemic analysis of causality in distributed systems

We propose a new formal knowledge-theoretic framework to model knowledge state of distributed agents in a distributed system. This framework is a generalization of our CTOL approach of modelling a reasoning about knowledge and causality [Moses Y, et al., (1988)]. Within this framework we model communication as a causal dependence between the agents 'actions on impermanent timelines. The agent's local actions are performed on local timelines, where global timelines are used to capture group knowledge such as common knowledge. This yields a more pragmatic definition to the notion of external knowledge. Moreover, we argue that epistemic causality plays a key role in distributed systems, further we show that causality tracking is much more involved for epistemic causality than for Lamport 's (1978) potential causality.