DC has more stoplights than its peer cities

Image by takomabibelot licensed under Creative Commons.

Recently, I rode in two of the best-known group bike rides in urban America: WABA’s 50 States Ride, which hits all the state-named avenues in a 62-mile trip, and Transportation Alternatives’ NYC Century, a 100-mile trip through four boroughs. I was seriously surprised to discover that the much shorter ride around DC was considerably more tiring.

Sure, the much greater elevation change and less favorable weather (that year) made a huge difference, but arguably the biggest difference was in the route itself. On the DC ride, we were stopping at lights far more often.

The NYC Century primarily ran along the city’s extensive waterfronts, ensuring not just a flat water-level route but also surprisingly few stops. The 50 States Ride offers a tour of the city’s many state-named avenues, some of which are bucolic side streets (surprisingly, the vast states of California and Texas were represented by quiet residential side streets) but many of which are heavily trafficked through streets. But since almost all of them cross diagonally across the grid and thus result in confusing intersections, many have far more than their share of stoplights.

Bicyclists participating in Transportation Alternatives' NYC Century, riding along the Rockaway Beach seawall. Image by the author.

Even short inter-neighborhood rides in DC seem to involve many more stoplights than in cities elsewhere. The 2.5-mile diagonal trip across the grid between my home in Southwest and H Street NE, across mostly-residential Capitol Hill, involves two dozen stoplights even when traveling mostly on side streets — about 10 stoplights per mile. A similar trip in the higher-density Chicago neighborhood I previously lived in had only two stoplights per mile, which is typical for Chicago outside of downtown.

Sure enough, DC indeed has one of America’s highest traffic-signal density. Our traffic-signal density rivals San Francisco or Queens, which have twice the population density; the District has 50-95% more traffic signals per square mile than Chicago or Boston, which have significantly higher population densities. On a population basis alone, DC has more than twice as many traffic signals per capita than the American average, which is about one signal per 1,000 residents.

City Traffic
signals
Sq. mi.
of land
Signals per
sq. mi.
People per
sq. mi. (2010)
Signals per
1,000 people
Washington, DC 1,652 61.4 26.06 9,317 2.89
Los Angeles 4,114 468.7 8.78 7,545 1.16
New York City 12,460 302.6 41.18 26,953 1.53
Queens, NY 3,119 109.2 28.56 20,554 1.39
Chicago 3,035 227.2 13.36 11,868 1.13
Boston 845 48.4 17.46 13,321 1.31
San Francisco 1,193 46.87 25.45 17,246 1.48
Atlanta 946 131.8 7.18 3,154 2.28

This stop and go pattern has many drawbacks: it decreases roadway efficiency, encourages high speeds between lights, and costs a lot to maintain.

Most importantly for a world with limited energy resources, coming to a complete stop, and then accelerating back to speed, saps energy from every road user. This might be a minor inconvenience for a car driver, who can summon hundreds of horses’ worth of power with a flick of an ankle (with grave implications for safety and sustainability), but this energy really adds up for bicyclists and pedestrians.

Bicyclists participating in WABA's 50 States Ride, stopped at Alabama Avenue. Image by the author.

Research by UMD professor Hiro Iseki found that commuter cyclists in Maryland suburbs could have accessed 70% more destinations within the same energy budget if it weren’t for the need to constantly start and stop at major intersections. Many of the most popular bikeway designs feature relatively few stops, whether they’re trails that follow waterfronts, or bike boulevards that use yield signs and roundabouts (which, like “Idaho stops,” require cyclists to yield rather than make full stops) instead of stop signs and stoplights.

How do you think that DC ended up with so many stoplights? Are there safer and more efficient traffic control methods that might work instead?