Scale and Complexity Multi-layered, multi-scalar agent networks in time-based urban design

Urban design, perhaps even more than architecture, is a time-dependent discipline. With its multi-layered complexities, from individual buildings to entire regions, decisions made at one level, that may not show effect immediately, may prove to have disastrous consequences further down the line. The need to incorporate time-based simulations in urban modeling, and the demand for a means of evaluating the changes have led to explorations with multi-agent systems in computation that allow for decisions to be decentralized. From the first basic rule-based system of Conway’s Game of Life [1] to recent urban simulations developed at institutions like the ETH Zurich [2], or UCL CASA [3], these programs synthesize the various exigencies into complex simulations so that the designer may make informed decisions. It is however not enough to simply use parametrics in urban design. Rules or desires implemented at one scale may not apply to another, while isolating each scalar layer for independent study reverts to the disjunctive and shortsighted practices of past planning decisions. Central to current parametric research in urban design is the need to deal with multiple scales of urbanism with specific intelligence that can then feed back into the collective system: a networked parametric environment. This paper will present the results from a city-generator, developed in Processing by Dino Rossi, Dominik Zausinger and Jeannette Kuo, using multi-agent systems that operate interactively at various scales.