Interesting! I have also been tickled by incremental map reduce and its possibilities.
> I believe that many of my problem cases above could, in theory, be expressed as a graph of more complex map-reduce operations.
Indeed, and if you can do that, then you are solving these problems in a completely declarative way, with all the advantages that that brings.
Personally I have been wondering whether it is possible to apply the concepts of map-reduce to more complex data structures (like trees) because I think that would open up a lot of possibilities.
