Research Visualization • Game Theory • Mechanism Design

The Cost of Rationality in Agentic Systems

Three-Layer Multiplicative Crisis Model — 95% AI failure rate as Nash equilibrium

Current System Effectiveness
25.4%
Mechanism Design Interventions
β — Coordination Bonus threshold: β ≥ T−R
0
γ — Defection Penalty threshold: γ ≥ P−S
0
α — Welfare Weighting α shifts dominant strategy
0.00
Enterprise AI Spending
$632B
Projected by 2028 — against ~5-10% empirical success rate
Build vs Buy Success
67%
Build-first orgs — vs 33% buy-first success rate
Nash Equilibrium Failure
95%
AI project failure is the rational dominant strategy without mechanism design
Chart Legend
Baseline effectiveness
Layer erosion (loss)
Recovered via intervention
Multiplicative Model
L1 (Org): × 0.60
L2 (Trust): × 0.65
L3 (Scale): × 0.65
= 25.4% effective