Garbage Can Decision Processes on a Lattice
Erik Larsen**, Alessandro Lomi
e.r.larsen@city.ac.uk, Alx@economia.unibo.it


     Two conflicting views coexist uneasily in the study of organizational decision process. According to the first, decisions within organizations are as the outcome of random encounters among problems, solution and decision makers. The second view emphasizes the stabilizing effect of routines and other procedural ñ and institutional - constraints on ordered sequences of organizational decisions.
     In an attempt to reconcile these two views, in this paper we propose a dynamic lattice model that shares many of the features of the original garbage can model of Cohen, March and Olsen. According to the model the sites of a square two-dimensional lattice are occupied by three types of objects: problems, solutions and agents. Collisions among objects are regulated by local rules of interaction and generate three kinds of organizational decisions: Decisions by resolution, by oversight and by flight. Objects are assigned to sites. The state of each site is updated synchronously by iterative application of deterministic local rules.
     In keeping with the original model, collisions generate a new class of objects called ìdecisions.î However unlike the original model, collisions also change the primitive component objects in fundamental ways. For example, decisions by resolution destroy problems and increase agents computing capacity. Decisions by oversight change solutions which then become parts of compound objects called routines. Decisions by flight generate new types of problems that ultimately destroy agents.
     We design and conduct a series of virtual experiments to explore the complex relationship between stability and change in organizational decision processes.