Wirth's Law as a KKT constraint

Wirth's Law ("software gets slower faster than hardware gets faster") isn't a moral failure — it's an optimization equilibrium. Producers maximize U(features, ease-of-building); quality is a constraint, not an objective: Q(hardware R, complexity C) ≥ Q_min. As R falls exponentially, C expands until the constraint re-binds at Q_min — a binding KKT constraint (multiplier μ>0, complementary slackness). Speed never improves because nobody was maximizing it — they were satisficing it.

Three conditions make it fire: (i) quality is a constraint; (ii) marginal utility of complexity stays positive; (iii) cost is externalized (users pay runtime, devs save dev-time).

Same structure shows up everywhere quality is a floor and a cheaper resource invites more consumption → verification-is-the-new-latency. Jevons paradox, Peltzman effect, induced demand, Parkinson's law are all instances.