Garrick and Marshall’s research into street network patterns began in Davis, California. Often cited as the most bike-friendly city in America, it has the country’s highest rate – more than 16 percent – of people commuting to work on two wheels. It turns out, though, that Davis also has one of the lowest traffic fatality rates in the country, a counterintuitive discovery for traffic engineers who consider biking a riskier alternative to driving.
Inspired by Davis, Garrick and Marshall compiled data on 230,000 crashes spanning 11 years in 24 medium-sized California cities. And they began to parse and classify street patterns in a kind of taxonomy. There are networks that look like square grids and others that resemble trees, with one trunk, many branches. There are networks that have tributaries, like a river, and others that have main roads radiating out from a central hub. There are hybrids of all these, and street blocks of different lengths, and networks that have 45 intersections per square mile (like Salt Lake City) and others that have as many as 550 (Portland, Ore).
In their California study, Garrick and Marshall eventually realized the safest cities had an element in common: They were all incorporated before 1930. Something about the way they were designed made them safer. The key wasn’t necessarily that large numbers of bikers produced safer cities, but that the design elements of cities that encouraged people to bike in places like Davis were the same ones that were yielding fewer traffic fatalities.
These cities were built the old way: along those monotonous grids.
When streets are built to a human scale, rather than being built for cars, those streets are friendlier and safer. More generally, the metric of intersections per square mile is an incredibly useful idea to keep in mind. It’s correlated with density, but it’s not the same thing at all. For instance, Badger reproduces these maps from the Federal Housing Authority, back in the 1930s: we’re seeing two plans, here, with identical housing density. But the one labeled BAD has ten intersections, while the one labeled GOOD has only seven. And of course the distance you need to travel to get from any random point to a given house is much shorter, on average, in the BAD map than it is in the GOOD one.
Shorter distances mean that you’re more likely to walk or bike; they also make the neighborhood feel smaller. For many decades, suburbia was designed on the idea that Americans want to feel as though they’re far away from each other, but the fact is that we need community and linkages just as much as we need a space of our own.
A city with high density still feels inhuman if it has enormous blocks; it doesn’t matter how many people you squeeze into downtown Phoenix, it will never feel like a vibrant, high-density city. Even in New York, the distance between 5th Avenue and 6th Avenue is far too big: the shorter blocks east of 5th Avenue, between, say, 5th Avenue and Madison Avenue, are much more pleasant, as are the small blocks in the Financial District.
It’s astonishing to me that Portland and Salt Lake City can differ, in terms of intersections per square mile, by a factor of more than 12. According to Wikipedia, Salt Lake City has a density of 1,666 people per square mile, compared to Portland’s 4,288. Which means that Portland has almost five times as many intersections per person as Salt Lake City does.
Which raises another question. Do street intersections make you liberal?