This and other SwarmLinda papers have a useful discussion of the scalability problems of Linda. The proposed solution however seems a bit dubious to me, though it would be cool if such biologically inspired mechanisms actually worked. (tags: architecture scaling)