These three papers by Mark S. Miller and K. Eric Drexler appeared in The Ecology of Computation, Bernardo Huberman (ed.) Elsevier Science Publishers/North-Holland, 1988.
Like all systems involving goals, resources, and actions, computation can be viewed in economic terms. This paper examines markets as a model for computation and proposes a framework--agoric systems--for applying the power of market mechanisms to the software domain. It then explores the consequences of this model and outlines initial market strategies.
Agoric computation will require market-compatible mechanisms for the allocation of processor time and storage space. Recasting processor scheduling as an auction process yields a flexible priority system. Recasting storage management as a system of decentralized market negotiations yields a distributed garbage collection algorithm able to collect unreferenced loops that cross trust boundaries. Algorithms that manage processor time and storage in ways that enable both conventional computation and market-based decision making will be useful in establishing agoric systems: they lie at the boundary between design and evolution. Algorithms are described in detail.
A long-standing dream in the field of artificial intelligence has been to use evolutionary processes to produce systems of greater competence than those we can directly design. This paper compares different evolutionary models--such as biological ecosystems, markets, and EURISKO--with respect to this goal. This comparison suggests that a form of ecosystem here termed a direct market (as opposed to the indirect market of human society) is a promising basis for computational ecosystems.
[ Home] [Tech Library]
Last updated: 19 July 2001