The Ecology of Computation (EoC) is a book of metaphors. Bernardo Huberman and most of the contributors to this collection believe I that adopting a social or ecological perspective on distributed computation offers distinct advantages over the prevalent centralized individualistic viewpoint. Although this attitude is shared by many researchers in Distributed Artificial Intelligence (DAI) [2,3,8], the realization of a societal perspective rarely proceeds beyond the metaphorical. The articles collected here push further, developing analytical techniques and architectural foundations based directly on concepts from biology and the social sciences.
The computational environments under investigation are open systems, defined by Huberman as "distributed computational systems without global controls". The characteristics of open systems enumerated by Hewitt-----concurrency, asynchrony, incrementality, incompleteness and inconsistency of central * (North-Holland, Amsterdam, 1988); 342 pages, $84.25 (hardback), $39.50 (paperback). l In reviewing a compilation such as this, it seems impossible to simultaneously provide an overall sense of the book, maintain accuracy, and avoid hedging universal statements about what the authors believe. Henceforth I shall omit these tedious qualifiers, with the understanding that assertions about the volume as a whole do not always represent a unanimous consensus among its various authors. Generally speaking, these views are most reliably attributed to the "mainstream", Xerox PARC contributors, whose chapters comprise the bulk of EoC and almost all of those papers written originally for this collection.