The space between

How the visual language of our modern maps are destroying our sense of place

Take a good look at this background image.

This photo (which I just ripped off the Internet; sorry) was taken at a street fair on Haight Street in San Francisco. I’d like to draw a comparison between that and a map of it. Here’s what you see if you zoomed in on Google Maps.

I grabbed this screenshot last week, and since then Google Maps have made their street lines much thicker.

Putting aside the useful store icons for now, it’s otherwise a bit anticlimactic. A lot of that potential wall-to-wall energy on the street has been reduced to a single line.

I’ll give one more example. Here’s Times Square in New York City.

Imagine what it’s like to be in the middle of it, surrounded by the bright lights and tall buildings and people just milling about. Then take a look at the map. It’s about four lines cutting through what looks like dead background space. If it wasn’t labeled Times Square, how would you know if there was anything interesting there?

Something is lacking in our ability to map public space. How did we get here?


Lines have been with us for a very long time. Ever since we’ve had the need to get from point A to point B, we’ve rubbed desire lines in grass, cut trade routes through mountainous vegetation, and built the Interstate Highway System.

We draw our maps with lines. We did it with hand-drawn paper maps, and we kept using lines when we turned everything into digital GIS data. It’s a useful symbolic representation, but by accidentally enforcing this abstraction on our maps, we start to assume that a road network’s only value is to be a connection between points.

The road itself starts to have less inherent value on its own.

We’ve been subconsciously training ourselves to believe that, like the Internet, a road network is nothing more than a series of tubes.

So we completely over-optimize our streets for just one single use case — moving cars around as fast as possible. For many decades, that was the entire point of traffic engineering.

All that does is reinforce the streets’ tube nature. So my argument is this: the way we draw maps cause us to think a certain way, and that way causes us to emphasize one use over others. It’s making it really hard for us to have good urban design.

Whenever someone looks at their public space or street and wants to propose something different to do with it, it threatens the generic mental model of what a street is supposed to be.

If you’ve ever been at a public meeting, you can see some people get really mad about that!

So I’d like to propose that we shift the visual language of our maps.

Lines are for pipes, paths, and we use it for streets. At a certain zoom level, the resolution is good enough for us to start representing streets as volumes, like rooms.

Why should we care about the volume of a street?

Because you might tell a large vehicle directions that could get it stuck in a narrow street. For example.

Here’s a scenario you’re more likely to face. Let’s say you wanted to get walking directions from the bus stop to the fountain at Justin Herman plaza. You wouldn’t really do this, but I just wanted to highlight one example of what happens when we’ve also trained our wayfinding algorithms to think of lines instead of space.

The result is a set of directions that uses the paths that are drawn on the map. But in real life, Justin Herman Plaza is very much an open public space and it’s really easy to walk across it directly to the fountain.

Aside: the mapping of the paths are a fiction. The circular steps aren’t a roundabout, as it’s drawn on the map. There are a couple of alternate routes that propose crossing the red-paved portion of the plaza, but they use an invented path at an arbitrary point that has no meaningful analog in real life. And some paths that could be drawn as lines are missing entirely.

That is why we need to start visualizing public spaces as volumes instead of lines. And one medium that does this really well is video games.

Here’s a screenshot from Deus Ex: Human Revolution, which takes place in a fictional cyberpunk-ish future Detroit.

On the street corner, there’s a wayfinding pylon. If you walked up to it and took a closer look, you’d see a variation of the in-game level map, which looks like this:

That white chevron is you, the character, standing in the middle of the street. The bright yellow boxes are buildings, and the “public space” (or areas your character can traverse) are shown as outlined areas. This game isn’t using lines to indicate acceptable paths of travel. Characters can move freely inside a volume of space, and so the in-game maps illustrate our conceptual mental model accurately.

Urban designers have been drawing maps like this for a very long time, too.

This is part of a map drawn in 1748 by the Italian architect and surveyor Giambattista Nolli. It’s one of the most well-known maps of Rome from that era, not just for its exemplary accuracy and attention to detail, but for its use of a “figure-ground” style to show the relationship between buildings (the “figure”) and the space between (the “ground”). Details like trees, columns, and staircases are visible.

And here’s an example of a more recent urban design plan I found. It shows a lot more detail of what’s actually going on in public space. Notice how the curbs have a shape that create parking spots and pedestrian areas. There are trees and plazas and crosswalks and areas where plants go.

This style is similar to an illustrative architectural plan, of the sort used to sell condos to prospective buyers, showing a proposed layout of furniture.

Thinking of our public spaces and streets as “rooms” is a good metaphor that helps adjust our mental model of visualizing space.

If you’re with me so far, you might be wondering, then, how do we tweak our technology and the visual language of public space? Wouldn’t that be really, really hard to do?

Well, not really. It’s already being done in some parts of the world. If you zoom in on Tokyo, Japan in Google Maps, it will actually render the spaces between buildings more accurately. So we can build this! We have the technology!

But do we have the data?

Washington, DC’s vector property geodatabase.

Curblines, which delineate where the sidewalk meets asphalt, are a great source of data we can use to start showing more detail. Chicago, Pittsburgh, Philadelphia, San Francisco and Washington, D.C. are just a few cities that I know of that publish these curbline GIS files openly. (The District’s base map, shown above, even contains curb cuts, driveways, and planted areas in separate GIS files.) This isn’t an exhaustive list, and many more are probably hidden inside the walled gardens of municipal GIS departments.

For places where data doesn’t exist or isn’t publicly available, we could even start tracing. There is a three-year old proposal on OpenStreetMap to map street areas in addition to lines. The lines are for routing purposes; the areas are for rendering purposes.

The visual language we use to represent reality can take on meanings we didn’t intend. When we record public spaces like paths and streets as lines connecting points, we have a tendency to optimize for the speed of moving along that path, rather than for the experience of being in that space.

We start thinking of public space as connecting pipes. Maps of areas turn shapes into awkward lines, and public squares become traffic sewers.

(left) Herald Square, New York City, pre-pipe (middle) Paris intersection in OpenStreetMap (right) Google’s series of tubes.

But if we tweak our mental model just a bit — and start thinking about public space as volumes, and as rooms that we can inhabit and decorate and live in, then we can start getting ourselves to think of other possibilities and uses for that space.

(left) Herald Square, New York City, post-pipe (middle) Paris intersection in OpenStreetMap proposal (right) A room.

The end.


This post is an adaptation of my notes and slides for an Ignite-style talk I presented at the Code for America Summit on September 24, 2014. Here is a video of that talk. (Thanks, Becky Boone!) Because of its short format, there were concepts I wished I could have included in the talk — for instance, the urban street wall. I may follow up with these ideas again later.

Also, many thanks to Andrew Eland of Google, who approached me after the talk to let me know that Google does collect spatial information with their Street View cars, so they do have quite a bit of this data. But rendering performance on mobile devices played a big part in making volume representation infeasible, at least for now. Apparently, that data is ready to go when the technology catches up.

Update, October 2: Thanks to Tom MacWright, who clued me in on where to get the GIS files for Washington, D.C. I’ve updated the article to reflect that.