Fifty-One Maps, No Territory

Abstract

Between 2021 and 2026, at least fifty-one distinct public definitions of "the agentic economy" have been published by industry analysts, research labs, platform vendors, consultancies, and academic institutions. A systematic review of these definitions reveals that they do not converge. They disagree on the central question (what kind of agent counts), on the unit of analysis (task, session, contract, or principal), on the infrastructure layer (protocol, platform, or product), and on the temporal horizon (present deployment vs. projected capability).

This paper treats the definitional landscape itself as the object of study. We classify the fifty-one definitions along five axes — agent type, economic role, settlement assumption, identity model, and time horizon — and show that the disagreement is not noise but structure: different actors are mapping different phenomena and calling them by the same name. The consequence is a field in which policy, infrastructure, and investment decisions are being made against incommensurable conceptual maps. We argue that clarifying the definitional terrain is a prerequisite for any coherent research program in agentic commerce.