Thinking like a Computer Scientist

• In Event: Economic Sociology Section Roundtables
  In Refereed Roundtable (60 minute): Roundtable 10: Theorizing Organizations

Abstract

How do academic experts influence the governance of markets when those markets are built in code? Research on the political influence of economics has shown that the discipline shaped market governance less through direct expert advice than through a "style of reasoning" — centered on efficiency and market mechanisms — that became institutionalized in the administrative state through curricula, regulatory practice, and knowledge infrastructure. Today, however, much of market governance happens inside private technology firms, where engineers write the code that allocates rides, prices advertisements, matches workers to tasks, and ranks search results. A field at the intersection of computer science, economics, and operations research (EconCS) has developed a sophisticated apparatus for designing such markets, and institutional pipelines connect top research universities to platform companies through internships, consulting, and "chief economist" positions. Given that market design is framed as technical rather than political, conditions appear favorable for direct expert influence.
Drawing on a preliminary sample of twenty-five interviews with prominent EconCS researchers, we offer two contributions. First, we characterize a computational style of reasoning as a distinct epistemic orientation in which models are treated as literal specifications rather than stylized representations, results take the form of approximation guarantees rather than equilibrium characterizations, and robustness to worst cases takes priority over analytic elegance. Second, we show that this style, like the economic style before it, enters market governance in attenuated form. A "thin" computational thinking — metric optimization, A/B testing, heuristic iteration — shapes engineering practice, but substantive EconCS knowledge is marginalized. The mechanism differs from the economics case: rather than political displacement, experts are preempted by practice. Engineers independently build working systems whose embedded assumptions and self-reinforcing metrics create path-dependent knowledge infrastructures that formal expertise arrives too late to reshape. Academic experts are relegated to peripheral advisory roles, lacking the relational authority built through sustained engagement with routine organizational work that research has identified as essential for expert influence.

Authors

• Georg Rilinger, MIT Sloan School of Management

• Arvind Karunakaran, Stanford University

• Elif Birced, MIT

• Julie Seager, MIT Sloan