The Price of Anarchy

Introduces the Price of Anarchy (PoA) as the ratio of the total system cost at user equilibrium to the total system cost at the system optimum, computes it from given costs and from latency functions, and discusses the 4/3 bound for affine latencies in routing games.

Step 1 of 157%

Tutorial

Introduction

The Price of Anarchy (PoA) measures how inefficient selfish routing is compared to centralized routing. For a given congestion game, let C(UE)C(\mathrm{UE}) denote the total system cost at the user equilibrium and C(SO)C(\mathrm{SO}) denote the total system cost at the system optimum. The Price of Anarchy is

PoA=C(UE)C(SO).\mathrm{PoA} = \dfrac{C(\mathrm{UE})}{C(\mathrm{SO})}.

Since the system optimum minimizes total cost by definition, C(UE)C(SO)C(\mathrm{UE}) \geq C(\mathrm{SO}), so PoA1\mathrm{PoA} \geq 1. A value of PoA=1\mathrm{PoA} = 1 means selfish routing happens to be optimal; a larger PoA means more system-wide efficiency is lost to selfish behavior.

For example, suppose the total delay across all flights on a city pair is 540540 aircraft-minutes under user-equilibrium routing, while a centralized flow manager could achieve 450450 aircraft-minutes. Then

PoA=540450=65=1.2,\mathrm{PoA} = \dfrac{540}{450} = \dfrac{6}{5} = 1.2,

so selfish routing produces 20%20\% more delay than is strictly necessary.

navigate · Enter open · Esc close · ⌘K/Ctrl K toggle