The purpose of this document is to identify the major recurring themes in the position statements of the Concurrency Working Group. It is hoped that such a summary will serve as the starting point for the formation of the group's position statement.

In reading through the statements, I identified the following themes:

• Debugging vs. Verification (Alur,Baeten,Henzinger): Is the former all we can reasonably expect out of concurrency research (as opposed to completely verified, provably correct systems)? A related distinction: model-checking vs. theorem proving (and their integration).
• Education (Alur,Best,Groote): It seems unlikely that the fruits of concurrency research will gain wide-spread acceptance and usage until the user community is educated in these techniques. Such education should start in the undergraduate curriculum.
• Integration of Formalisms (Alur,Cleaveland,Groote,Henzinger,Hoare,Luginbuhl,Misra,Moller,Montanari,Pratt,Sifakis): Integration or unification (Hoare) of concurrency theories could illuminate the relationship between theories, and help identify selection criteria for choosing among theories when confronted with a particular application. Cleaveland points out that a modular approach to semantics could facilitate integration. Montanari proposes causal computing and Pratt proposes Chu spaces as unifying frameworks. Groote raises the related issue of standardization.
• Integration with other Disciplines (Alur,Baeten,Cleaveland,Luginbuhl,Misra,Montanari,Steffen): The proliferation of embedded systems is a compelling reason for us to increase our interaction with disciplines such as electrical, mechanical, aerospace engineering, and control theory. Integration with other computer science disciplines--e.g., databases, object-oriented programming, constraint and logic programming, term and graph rewriting--should pay dividends as well. Also, integration of concurrency-based formal methods with software engineering is likely to yield a healthy synergy (Baeten and Cleaveland).
• Correctness vis-a-vis Efficiency (Best,Gouda,Montanari): Previous concurrency research has tended to treat these two fundamental issues in a mutually exclusive fashion. Can we bridge this gap?
• Non-functional requirements (Baeten,Luginbuhl,Prasad,Sifakis,Smolka,Thomsen): Concurrent systems, unlike their sequential counterparts, must contend with the following non-functional requirements: real-time, fault-tolerance, security, reliability, probability, availability, mobility, continuous behavior (hybrid systems), etc. Mobility, and the related area of high-order process calculi (e.g. the pi-calculus), are specifically addressed by Montanari, Prasad, and Thomsen.
• Programming language development (Best,Cleaveland,Miller,Misra,Sifakis,Thomsen): The development of programming languages with strong foundations in concurrency theory should greatly aid the design, coding, and verification of concurrent systems. As pointed out by Thomsen, such an approach could eliminate the gap between the model and implementation of a system. A well-designed language could also promote portability without limiting parallelism (Best).
• Algorithmic challenges (Alur,Moller,Smolka,Wolper): Algorithmic advances are needed in order to better cope with the state space explosion problem inherent in concurrency specifications.
• Tool development (Groote,Miller,Moller,Smolka,Steffen,Thomsen,Wolper): As safety critical systems become more and more complex, the need for automated support for verification becomes even more important. Also, we need to seriously confront the problems facing many existing tools: bugs, lack of scalability, lack of portability, etc. As pointed out by Wolper, "algorithmic support" is needed for successful tool development.
• Real applications (Groote,Luginbuhl,Smolka,Steffen,Thomsen): Taking on real-world applications will allow us to evaluate the utility of our methods and at the same time improve the quality of the applications we target.
• Other topics: Other topics that stood out are processes and data, testing, infinite-state systems, and the role of linear logic in modeling concurrency.