This was a tri-event comprising:
The central attraction was HCAT on July 26, with a star-studded cast of speakers: Brzozowski, Sheila Greibach, Michael Harrison, Juris Hartmanis, John Hopcroft, Werner Kuich, Robert McNaughton, Maurice Nivat, Michael Rabin, Grzegorz Rozenberg and Arto Salomaa, in order of appearance. Greibach, Harrison and Kuich talked about the past, Hopcroft discussed the future, and the rest gave largely technical lectures. Sheila Greibach had an amusing technique of using two slide projectors concurrently, with technical accounts on one, and cartoons, quotes, etc on another. Harrison talked of the excitement of the old days, and decried the lack of rigour in refereeing these days; he also vehemently criticised the `conference culture' today. Hopcroft predicted that automata theory would make major contributions in the areas of information retrieval (in particular, from the World-Wide Web) and analysis of data in multi-media (audio/video).
Brzozowski used an example of recent research on \emph{hazards} in asynchronous circuits to illustrate the interplay between circuits, algebra and automata. Hartmanis showed how several results linking automata and complexity theory could be phrased as incompleteness of formal systems, thus providing G\"odel-type theorems (for instance, for any system $F$, there exists a context-free language L such that for all $G, Lang(G) = L$ is not provable in $F$).
McNaughton, Rabin and Rozenberg sold (very well) infinite games, randomization and DNA processing respectively. Nivat and Salomaa stuck to specific technical issues. (With a long list of speakers, the lectures ran very late and I had to miss the last few as I had to catch a bus.)
I attended CIAA and not DCAGRS. Despite its title, only a few talks in CIAA were on applications or implementations, most were `theory' talks. There were two invited lectures, one by David Harel on system development using automata-based tools, and the other by Lari Karttunen on finite state transducers in natural language processing. Harel gave a nice picture of how structured system modelling and design could incorporate what he called `play-in scenarios'. The designer says, this is the kind of scenarios we should have, and a tool modifies the (finite state) system appropriately. Karttunen's work comes from linguistics, and his account of how minimization techniques help in the analysis of phonemes was very informative.
Among the contributed talks, the one on MONA implementation by Anders M{\oe}ller was most impressive. Nils Klarlund, M{\oe}ller and Michael Schwartzbach have speeded up this tool (which automates the decision procedure for monadic second order logic) enormously, using a host of techniques including formula reduction, use of three valued logic, BDD manipulation, cache-conscious data structures, \ldots The talk illustrated how using an amalgam of techniques can really help.
The talks by Jeff Shallitt and Pighizzini stood out because they showed the intricacies involved even in the study of regular languages over one-letter alphabets: the former had results on complexity of concatenation, and the latter on intersection, with nice connections from number theory. On the other side, Oscar Ibarra showed how results on timed automata follow from results in `classical' automata theory, on machines with reversal-bounded counters (these can be incremented, decremented or compared against constants, but alternation between increasing and decreasing modes should be bounded).
CIAA had only pre-proceedings, the final proceedings will come out soon as an LNCS volume. There was talk of making HCAT lecture slides available on the net, but that has not happened yet. Sheng Yu must be commended for a well-coordinated organization of the event.