USP514: Thinking about density

In our studio we will study how to add more housing to Silicon Valley communities by evaluating the Shoreline West neighborhood in Mountain View. This will mean an overall increase in density. On the ground it will also mean more variation in density, including more open-space in some areas, as well as mixed land-uses. Here I want to lay out some parameters and questions about density that we should consider.

What is the range of density we should aim for?

First, what density is too low? I grew up in a suburb in New England: quarter-acre parcels created a net density of 4 dwelling units per acre (4 DU/ac in planning shorthand). At such a low density, mass-transit cannot be supported. Buses went by once every 40 minutes. I will never forget the snowy winter day when I could not ride my bike to the town center. I waited half an hour for the bus; and when one arrived, the driver could not see me over the snow-bank and did not stop. Practically speaking, it is infeasible to live in this sort of sprawl without a car. You are compelled to burn fossil fuels for most trips. I also found it to be a lonely, isolating environment.

Second, what is high density? The tenement neighborhoods of the Lower East Side of Manhattan, circa 1910, were relatively high density. That is where my grandmother grew up. Here is a rough estimate: 500 square foot apartments, 4 per floor, in 6-storey walk-up buildings; so 24 units per building. Parcels were 25′ x 100′ (2500 s.f.), which is 17.5 lots/acre; 1 acre = 43,560 square feet. That was a net density of 420 DU/ac. However the gross density is less, because the total surface area of a neighborhood includes streets and parks as well. In Manhattan, streets comprise 35% of the land area, and parks are 15% of the land area. Therefore the gross density was half of the net density: 210 DU/ac. When opponents to upzoning refer to “Manhattanization” or “tenements,” it is useful to know these two numbers because generally, upzoning means increasing allowable net development densities to about one tenth of this density.

Although I am strongly convinced that 4 DU/ac is too low, is 420 DU/ac (net) too high? My grandmother seemed to enjoy it. Furthermore, 210 DU/ac overall density also happens to be the density of central Paris. One famous neighborhood with an even higher density was the North End of Boston, at 275 DU/ac gross, that was demolished in the late 1950s. Both Jane Jacobs and Herbert Gans argued compellingly that this ‘urban renewal’ demolition was a terrible mistake, because the North End was a healthy, vibrant community even though planners were certain that it “shouldn’t” be healthy because of hearsay-based presumptions about “appropriate” and the supposed hazards of “over”crowding.

Should we even limit density through zoning? No. Zoning density-limits are redundant because Building codes already limit density. Furthermore, building codes are not based on hearsay and anti-urban myths; they are based on compelling evidence for the promotion of safety and health. The current building code used by most American jurisdictions is the International Building Code (IBC). The IBC specifies window-orientations and building spacing to minimize the spread of fire; stairwells and passages for exiting; and minimum operable windows for natural-air ventilation. These regulations are based on demonstrable evidence from actual fires and respiratory diseases linked to poor ventilation. The IBC does effectively limit development-densities, preventing the poorly-ventilated firetraps of the late 19th century–but only for specific and demonstrable reasons. In contrast to these regulations, the justification for zoning density-limits are extremely weak, and generally used for race/class exclusion, as demonstrated in the Mount Laurel litigation in New Jersey.

Therefore: we need to know about density, to understand its urban implications. In future posts I will get into the details of building and housing typologies to show what specific densities actually look like on the ground. When you see how building codes actually limit development densities (for good reasons), it may become more clear why I am arguing here that we do not need to directly limit densities through zoning.

So, what is a reasonable density? Over the last few days I have been analyzing two of the most fashionable neighborhoods in the United States: the Mission District in San Francisco and Park Slope in Brooklyn. Using Census Tract Maps and data from the American Community Survey (ACS) from 2013, I found that these neighborhoods ranged from 18 to 33 DU/ac. Here is a table of four tracts from each city:

City Tract DU/ac Pop/ac med.value Neighborhood
SF 106 28.94 53.19 $1,168 $744,600 Telegraph Hill
SF 202 25.46 43.30 $1,333 $625,000 Mission @ Market
SF 207 20.66 41.75 $1,832 $876,800 Mission @ 16th
SF 210 17.99 33.50 $1,735 $876,400 Mission @ 24th
Brooklyn 165 32.65 66.05 $1,989 $925,800 Prospect Park West
Brooklyn 155 25.50 55.33 $1,963 $988,600 Park Slope
Brooklyn 157 28.25 59.22 $2,000+ $938,30 Park Slope
Brooklyn 133 30.93 72.25 $2,000+ $1,000,000+ Lower Park Slope

[Source: U.S. Census, ACS 2013. Note: the Census does not measure median monthly rent above $2000 nor median property values over $1 million because they are so rare, nationwide]

The tracts shown in the table above are not the densest tracts in either city, nor the most valuable. But in recent years they have been the most fashionable. Tech-workers from Silicon Valley are flocking to SF Tract 207 in particular, so I suggest that we treat this as a reasonable target density for development in Mountain View. Like New York, San Francisco was laid out with about 35% of the surface area as public rights-of-way (streets), and about 15% open-space; so to get a 21 DU/ac gross density requires an overall 42 DU/ac net density. This is neither a minimum nor a maximum (note that the hip tracts in Brooklyn are considerably denser). Rather, it is an informed target we can use as a guideline.

What about population density?

Thus far I have only discussed unit densities. Population-density is equally as important and quite different. In the table above note that the people/acre in San Francisco hovers at about double the density of units, meaning that the residential density is about 2 people/unit in these tracts. Per-unit densities in Brooklyn are higher. According to the ACS, average household size in the Brooklyn tracts ranged from 2.03 in Tract 165 to 2.63 in Tract 133, resulting in 72 people/acre even though the DU/ac is barely higher than in San Francisco.

The disparity between population-densities at the extreme cases I cited at the beginning of this page is even more extreme. In 1910 New York my kin dwelled as 6 people per unit, which meant 1260 ppl/ac gross density. On the other end of the scale, modern household size–both in suburbs and in cities–is about 2.3. That means a gross density of 4.6 ppl/ac. So even though the built unit-densities of these two cases vary by a ratio of 100:1, the population-densities vary by a much larger ratio of 274:1.

Population-densities are critical for many parts of planning, like projecting traffic demands and numbers of students in public schools.