Uninfluenced social systems often exhibit suboptimal performance; specially designed taxes can influence agent choices and thereby bring aggregate social behavior closer to optimal. A perfect system characterization may enable a planner to apply simple taxes to incentivize desirable behavior, but system uncertainties may necessitate highly sophisticated taxation methodologies. Using a model of network routing, we study the effect of system uncertainty on a designer's ability to influence behavior with financial incentives. We show that, in principle, it is possible to design taxes that guarantee that selfish network flows are arbitrarily close to optimal flows, despite the fact that agents' tax sensitivities and the network topology are unknown to the designer. In general, these taxes may be large; accordingly, for affine-cost parallel-network routing games, we explicitly derive the optimal bounded tolls and the best-possible performance guarantee as a function of a toll upper bound. Finally, we restrict attention to simple fixed tolls and show that they fail to provide strong performance guarantees if the designer lacks accurate information about the network topology or user sensitivities.

中文翻译：

---

拥塞博弈中鲁棒协调的最优机制


未受影响的社会系统通常表现出次优的性能；特别设计的税收可以影响代理人的选择，从而使总体社会行为更接近最优。完美的系统特征可能使规划者能够应用简单的税收来激励理想的行为，但系统的不确定性可能需要高度复杂的税收方法。使用网络路由模型，我们研究了系统不确定性对设计者通过经济激励影响行为的能力的影响。我们表明，原则上，可以设计保证自私网络流量任意接近最优流量的税收，尽管设计者不知道代理人的税收敏感性和网络拓扑。一般来说，这些税可能很大；因此，对于仿射成本并行网络路由游戏，我们明确地推导出最优有界通行费和最佳可能性能保证作为通行费上限的函数。最后，我们将注意力限制在简单的固定收费上，并表明如果设计者缺乏有关网络拓扑或用户敏感性的准确信息，它们就无法提供强有力的性能保证。