Donald Bren School of Information and Computer Sciences

Frequently, proposals for automated information systems to provide more accurate, timely, and relevant information to top-level policymakers in public agencies have tended to neglect the ways in which public policymakers are known to search for, select, and use information in their deliberations. This essay analyzes the likely use of automated information systems by policymakers. First it articulates patterns by which policymakers are known to select and use information. Then distinctive features of automated information systems are analyzed to understand the extent to which these common patterns may be altered. Such analyses show that public agencies are beset with several "information processing pathologies" such as filtering and suppressing data which auomated information systems, in and of themselves, are unlikely to alter. An alternative proposal which couples selected organizational rearrangements with automated information systems ("technology-policy mixes") is presented as a more promising, but more difficult strategy for diminishing the information processing pathologies which are endemic to top-level policymaking. This analysis implies that people who build computer-based information systems such as simulations cannot claim that their artifacts will have some positive impact on policymaking simply because they contain some "policy-relevant" variables. Rather, they must also show that the organizational arrangements within which the data or analyses are not subject to the typical information processing pathologies found in public agencies.