Constraint–Flow Lever Law

This paper proposes a domain-independent selection law for optimization under constraint. The Constraint–Flow Lever Law states that, in any persistent energy-based system, the dominant hard constraint—identified by marginal loss sensitivity—deterministically selects the dominant optimization lever. Interventions outside the selected lever family yield diminishing returns. The law is predictive and falsifiable. A minimal test protocol and explicit falsifier are provided: if a non-predicted lever produces greater marginal improvement than the predicted lever under controlled conditions, the law fails. The contribution is methodological rather than domain-specific. The law does not replace existing theories; it selects which class of intervention matters first, collapsing the optimization search space across physical, biological, informational, and social systems. Replication attempts and counterexamples are explicitly invited. Keywords: optimization, systems theory, constraints, leverage, cybernetics