Abstract

Economics often starts by stipulating what agents are trying to maximize, then explaining their choices as means to that end. This paper argues that the move quietly overcommits: when ends are uncertain or shifting, labeling behavior “irrational” or “inconsistent” mostly reflects the analyst’s mis specified objective rather than a genuine failure of reason. I propose a “primitive shift” from utility (as an often-contentless placeholder) to desire: we do not assume we know what people want—only that they want something, and act under constraints and limited knowledge to pursue it. This minimal primitive supports a correspondence-style project: familiar pillars of economics reappear as limiting cases once welfare is recast as the value of the reachable set and “cost” as a broad barrier bundle (money, time, risk, stigma, institutional friction). On this basis the paper re-derives canonical phenomena—Invisible Hand coordination, comparative advantage, subjective value, Keynesian animal spirits, Marxian “political prices,” praxeological action, Samuelsonian utility as a local compression, and Sen-style capability reasoning—without positing a universal end. Finally, the framework implies an empirical shift: when behavior is regime-conditional, econometrics should focus on decision-landscape reconstruction rather than global parametric surfaces, illustrated by benchmark gains of a regime-sensitive random forest over logit baselines on the UCI Bank Marketing task.