Advertisement

But these guys had even more fun with the data. They overlaid the data with several other maps, like crime maps and maps of how people voted on Proposition 8, which (temporarily) banned gay marriage. And Groß and Lee created a hypothetical pool-hopping route. Say you want to go for a dip in a stranger's pool: now you'd know exactly where the next pool was located when you got kicked out.

The project's scope is incredibly ambitious, and so well-documented that just examining their methodology is highly entertaining. For an easy way to demonstrate how they did it, this flowchart gives an overview of their process:

Advertisement

Using two different maps from the National Agriculture Imagery Program—one true color composite and one false color composite—Groß and Lee sent the information to a "clipping factory," an outsourced labor center in India that specialize in Photoshopping products onto white backgrounds for e-commerce sites.

Groß and Lee examined the selections, deleting any pool-like objects that were not actually pools (like lakes) and highlighting other probable pools that had been missed, and sent it all back to the clipping factory. Each pool was then cross-referenced geographically against high-res Bing maps. Finally, they submitted the high-res images to Amazon Mechanical Turk, an online crowdsourced labor force, and asked the workers to make sure they weren't counting, say, a blue kidney-shaped roof. Groß and Lee then human-checked everything again.

Advertisement

I know, it sounds exhausting. The physicality of the book is what borders on ridiculous: The Big Atlas of LA Pools has only been printed once, at the cost of $3,700. But it only took a year (during which we know they were working on other stuff) and about $650 of outsourced labor, which is not really all that expensive or time-intensive. Groß and Lee told the Los Angeles Times that they weren't sure what exactly the value of their work would be. But I see tremendous opportunity here for the way that we take stock of our cities.

Advertisement

L.A. is currently embarking upon an equally ambitious project called Survey LA, where they're attempting to catalog every parcel in the City of L.A. This is namely to find out if they're missing any architectural or cultural gems which need to be preserved, but it will also create the city's first dataset of the built environment. Since 2010, Survey LA has been doing this the old-fashioned way—with volunteers, on foot, walking every street in every neighborhood, and talking to residents—but it will result in a highly nuanced information. When it's finished, you'll be able to see the location of very specific types of architecture, say, all the Victorian-style single-family houses built before 1910.

Pools would be tough to count this way, of course—it would require knocking on every door, for one—but I'd have to say that Groß and Lee's pool count is likely far more accurate than anything the city could have officially tabulated. If you asked a city to tell you how many pools it had, I would guess that they would tell you a figure based on the number of permits pulled to build the pools. The permitting process has likely changed over the years, and, besides, those records are not likely digitized, so you'd have to go through that whole exercise first. And, even then, you probably wouldn't have the same qualitative information that Groß and Lee were able to glean from their methods. Groß and Lee's approach works for the same way that Survey LA works: It's ultimately based on human observation.

Advertisement

That's what makes me think that Survey LA's goals—or any other comprehensive mapping project—might be able to be achieved by the imagery + mapping + outsourced labor process that Groß and Lee came up with. Studying something as frivolous as swimming pools today might turn into a way to find the most ideal locations for solar panels to generate renewable energy, or help transportation planners find and fix all the broken or missing sidewalks.

Oh, and the number of pools, in case you were wondering, is 43,123. [Los Angeles Times]