Inquiry and Change

In the AI Story

The book's argument is deliberately uncomfortable for its intended audience. Social scientists had long accepted Lindblom's critique of comprehensive planning while assuming that their own analytical work could achieve a comprehensiveness that policy-making could not. Inquiry and Change rejected the assumption. The structural constraints that defeat comprehensive policy analysis — cognitive limits, value conflicts, distributed knowledge — defeat comprehensive social science in the same way and for the same reasons.

The implication is that social-scientific knowledge, like policy, is produced through distributed incremental investigation by many researchers with different perspectives and partial access to the phenomena they study. The accumulation of partial studies produces collective understanding that no single research program could generate. The understanding is messy, inconsistent across perspectives, and perpetually revised — but it is more adequate to the actual complexity of social phenomena than any comprehensive research program could be.

Applied to the AI transition, the book's thesis suggests that no social-scientific discipline will produce the comprehensive understanding of AI's social consequences that would justify comprehensive policy responses. The research is necessarily distributed: economists studying labor-market effects, sociologists studying organizational dynamics, psychologists studying cognitive consequences, political scientists studying governance implications. Each discipline produces partial understanding. No meta-discipline synthesizes the parts into a comprehensive whole. The policy response must work with distributed partial understanding, not wait for a comprehensive account that will never arrive.

The book also anticipated aspects of contemporary AI that its author could not have foreseen. Lindblom's argument about the mutual shaping of inquiry and action — research questions arise from practical problems, research findings shape practical responses, the shaping is iterative and reciprocal — describes with some precision how AI research has actually developed, with applications driving research agendas and research findings reshaping what applications are attempted.

Origin

Lindblom wrote the book in the late 1980s, drawing on four decades of work at the intersection of political theory, policy analysis, and philosophy of social science. It was received as a provocation by the social-scientific establishment and as a vindication by the post-positivist tradition that had been arguing for reflexive awareness of research's entanglement with its subject matter.

Key Ideas

Reflexive incrementalism. The incrementalist critique of comprehensive planning applies to comprehensive social science as well.

Distributed inquiry. Social-scientific knowledge is produced through the interaction of many researchers with partial perspectives.

Mutual shaping. Research and action shape each other iteratively; the distinction between pure and applied inquiry is less sharp than the research tradition claims.

Usable knowledge. The test of social science is its contribution to practical problem-solving, not its approximation of comprehensive theoretical understanding.