A growing demand for distributed and decentralized software, together with an increasing inherent complexity of these systems challenges traditional software engineering practices. Recently, the utilization of self-organizing processes has been proposed as a way to relieve development efforts. While Agent-Oriented Software Engineering (AOSE) provides suitable abstractions and implementation frameworks for these system dynamics, the design of phenomena that rise solely from entity interactions contradicts traditional top-down development strategies. This paper describes work in progress and outlines a doctoral thesis that addresses the analysis and design of self-organizing distributed systems in Multi-Agent Systems (MAS), i. e. collections of autonomous, pro-active entities.