Visualizing Urban Data: A Journey Through Oakland Crimespotting

7th May 2008 @ 8:43 pm
nowhere
Liz Goodman recently invited Mike and I to speak at the UC Berkeley School of Information. We took the opportunity to give a full length talk about a single project, Oakland Crimespotting, which is something of a rarity since we normally try talk about lots of things a little bit, rather than one thing in depth.

Mike started and finished the talk with an in-depth look at the motivations, technical details and social issues surrounding the site, which you can read about on his blog. In the middle I gave a brief overview of related projects and talked about the how the site sits alongside our other mapping work at Stamen. Mike suggested I use a reverse-chronological narrative structure that he liked from a book about Polish history, so I started with the stuff we've finished recently and working back to Stamen's early work with MoveOn and Mappr.

Mike has since reprised the talk for the journalism school, and the whole hour is up on Youtube (part 2 here) if you have time to watch it. Alternatively, you can attempt to simultaneously read the full version of this post and Mike's post together to get a wordier overview of what we talked about.



[At this point Mike has briefly introduced the Oakland Crime site and flash map, and hands over to me for related projects and studio context]

Why Crime Maps? We're just getting started!

I feel it's very important to acknowledge that we're not the first or only people thinking about this stuff. Journalistopia has a directory of online crime maps and rightly credits ChicagoCrime.org as the first high profile example of regular web developers acquiring data from their city and presenting it in a more accessible, higher quality manner.

CrimeReports.com is a service that aims to help law enforcement agencies share their data more openly, like us they're convinced that their approach is better than the Big Government IT Packages that cities have commonly bought into. They also provide one central place to find the information, rather than relying on out-dated city websites to point people in the right direction.

EveryBlock.com generalises the concept a great deal, from the same team who developed ChicagoCrime.org but aiming to cover all manner of local information from restaurant inspections through to recent photos from Flickr.

Newspapers of course have a variety of offerings. The Oakland Tribune produced Not Just A Number, which deals in depth with the individual stories of homicide victims in Oakland. It offers a much more sombre interface than our crime maps, which is certainly appropriate for the more personal information it contains.

Crime maps can get political too. For whatever reason, San Francisco's own official crime maps do not currently contain information about homicides. The San Francisco Chronicle provides a map at SFGate.com that aims to compensate for this omission.

Like the Tribune's Not Just A Number, the LA Times has detailed demographics to back up their homicide map, as well as news articles and photos. (These are used in an excellent recent UC Berkeley Visualization Lab paper too).

And of course there are plenty of other kinds of tragedy to document, as these maps of U.S. military casualties in Iraq show, from the New York Times. This is also one of the best interfaces around for exploring time and maps simultaneously.

Studio Context: maps, maps, maps.

Maps make up a large proportion of our work at Stamen. Apart from maps of crime, last year we completed maps of 911 calls in Indiana, real estate building patterns across the U.S., travel times and house prices in London and several more. An updated version of Cabspotting, featuring animated maps of taxi cabs in San Francisco, was recently featured in the New York MoMA's Design and the Elastic Mind exhibition.

Trulia Hindsight is a an example of our mapping work. Trulia already had a fairly mature mapping and visualisation platform with their heatmaps at trulia.com, but they wanted us to take a look at their data from another angle...

... the maps we produced are animated, and the coloured dots show the years that residential properties were built across the whole USA. This map shows a sample of properties from the San Francisco bay area...

... and this one shows Plano, Texas. These are just two examples of the different kinds of building patters that become visible when you try to show everything.

Trulia Hindsight didn't spring into being as a fully interactive browseable website, it started with some animations of small samples of the data. This one shows the year that properties were built in San Francisco.

This one animates along another dimension: price, rather than time. It's interesting to see how the landscape is still represented in some ways, but it's different to the chronological view. High prices generally run uphill, and sometimes out to the coast.

The logical next steps for these maps don't always work out. Taking red, orange and yellow glowing dots and placing them on top of aerial imagery reminds us of all the wrong things...

... like this NASA image of fires in the Oakland hills, from 1991. I've been calling this accidental visual resonance, and it's something I picked on Yahoo's now-defunct Design Innovation team about a few months ago.

So I hope that it becomes clear that we think hard about the wider cultural context our work sits in. Trulia Hindsight is one outlet for our ongoing experimentation at the edges of what's well-understood about putting interactive maps online. This sometimes involves quite convoluted exploration, such as Mike's Faumaxion Map (updated here) which seeks to apply the work of Buckminster Fuller to make an interactive map that represents different points of view equally.

One point of view that's worth considering is your own. Where can I get to from here? How long will it take me to get there? We recently explored the idea of travel time maps with mysociety, producing interactive maps as a proof of concept that hopefully one day will be available to everyone to customise for their own circumstances.

The animation above (which is better experienced for yourself) shows how travel time and house price interact to make it aggressively expensive to live near where you work in London. del.icio.us user seldo said of these maps,

"This is one of the most amazing things I've ever seen -- in three clicks it reveals what it took us years to learn: that the only place you can buy a house for GBP300k and still live 30 minutes from your office is the Kennington/Oval/Vauxhall triangle."

When the experimentation is over and the concepts are proved, occasionally it makes sense to draw a line under what we know and give it a name. That's what Modest Maps does - it's a programming library for manipulating tile-based maps in Flash (actionscript 2 and 3), Python and Processing.

Modest Maps didn't spring out of nowhere either. It's based on what was learned during several previous Stamen projects, such as Cabspotting.org, which tracks San Francisco's yellow cabs in real time...

... and Mappr.com, which plotted photos from Flickr onto a map by guessing where they were from the tags people provided, a long time before Flickr allowed users to accurately position their photos on maps on the site itself.

Some of the earliest Stamen mapping projects were made for MoveOn.org. The map above showed the locations of everyone who was listening to a live "Town Hall" conference call, and allowed participants to see each others responses and post questions in real time. Stamen has continued to explore the theme of real-time visualisation through work on projects such as Digg Labs, but it's the collaborative civic awareness that maps can offer which is driving the Oakland Crime project, and that's what Mike is going to talk about next.

[That marks the end of the related projects/studio context section of our talk. You can read Mike's notes to find out what he had to say about Oakland Crime]