I'd hoped to have "Concurrent Automata and Their Logic" fully rewritten by now as "Event Spaces and Their Linear Logic" to incorporate the many new things that have emerged meanwhile. Unfortunately too much else has been going on so ES has been on the stack for a while. I have three projects due before I leave for POPL on Saturday so it will remain on the stack till February. Meanwhile here's how the abstract looks. The original version is still available as catl.tex by anonymous ftp from boole.stanford.edu:/pub, it plus this abstract is enough to permit reconstructing everything. Incidentally Boole's ftp server has logged 646 sessions from 322 different hosts since I started this service in October. Boole is a useful little (in fact tiny) file server. Event Spaces and their Linear Logic V.R. Pratt An event space or schedule is a poset having a top as the permanently deferred event and arbitrary nonempty joins as the concurrence of the joined events, with event conflict represented by their join being top. The state space or automaton dual to an event space is obtained simply by removing the top and adjoining a bottom; this yields a poset having a bottom as the initial state and arbitrary nonempty meets, which surprisingly all exist, as decision states, with demonic state choice represented by their meet being bottom. These structures form dually isomorphic nondegenerate categories Sched and Aut determining the operations of a concurrent programming language, along with additional operations to make it a linear logic of concurrency whose exponential yields a free schedule and its dual a free automaton. The structure of event spaces may be understood in terms of that of vector spaces via three moves: first drop scalar division and complete to yield complete modules, then drop scalar subtraction to yield complete semilattices, and finally relocate the origin to the top to yield event spaces. Each stage preserves the operations of direct product, tensor product, internal hom, and dual while introducing an improvement: first removal of the finite-dimension limitation on duality, then removal of group-induced rigidity to create a logic, and finally separation of all dual pairs of operations into usefully different operations. Category theory was used to find event spaces but is not needed for their exposition, for which the language of elementary lattice theory suffices.