About the same time that the Edens Expressway strangled the North Shore Line, Traffic Quarterly published Mr Downs's "The Law of Peak-Hour Expressway Congestion." In those days, the law was dismal enough: a limited-access highway capable of moving commuters more rapidly than they could move on the existing arterial streets would attract commuters until the marginal commuter was indifferent between using the limited-access highway or sticking to the existing street grid. In Stuck, the Law becomes the principle of triple convergence.
Thus even greatly widening any major commuter expressway cannot long reduce peak-hour congestion there. Additional drivers will shift onto that improved road from other routes, other times, and other modes (such as walking or public transit) until movement on it is just as slow as the movement on alternative routes. Since those other routes are less direct than the expressway, such equalization means expressway traffic is usually crawling at the peak hour. Even most wholly new roads will soon fill up during peak hours.(Stuck, pp. 327-328). At best, the expansion reduces the duration of the most congested times of the day, but even that is not a given. Meet the principle of the swamping effect.
[R]elatively small reductions in initial traffic congestion in a rapidly growing metropolitan area will be fully offset within a few years by the arrival of more people, jobs and vehicles there.(Id. at 328.) The book provides an extended inquiry into the nature and causes of traffic congestion, and a near-encyclopedic evaluation of possible policies to deal with congestion. There's a useful summary table at p. 348. Capacity expansion, whether augmenting existing expressways, with or without separate carpool or truck lanes, or of public transit service, grades out as "moderately effective, costly." The "most effective" policies are also those most difficult to sell to the public: peak-hour pricing of all expressways and arterial streets, possibly including satellite tracking of vehicles to enforce time-of-use tolls (invasions of privacy be hanged?); higher gasoline taxes (conscripting soccer moms into the Pigou Club?); expressway expansion in the form of high-occupancy toll lanes only; and surcharges on long-term parking in the morning. The policies favored by technocrats tend to be "very ineffective": concentrating jobs in clusters in new-growth areas; driving bans based on license plate numbers; zoning to put jobs closer to houses; higher automobile licensing fees; growth limitations. I am pleased to note that synchronized green lights receive a "relatively effective" grade. The existence of rapid-response crash teams to remove road obstructions is also relatively effective. The book is a 2004 revision of Stuck in Traffic: one item that escaped greater scrutiny is distracted drivers. Mr Downs notes that congestion is a consequence of a difference between individual and collective costs (the driver sees the delay created by all the vehicles in front, but is not required to compensate all the vehicles in rear for getting in their way.) Is a similar wedge at work when drivers take advantage of the delays to multi-task, thereby leading to inefficiently high levels of distraction and more crashes?