Co.Design

MIT's Place Pulse: A "Hot Or Not" For Cities, To Fix Broken Blocks

A website uses crowdsourcing to convert people’s perceptions of streets into a quantitative benchmark for measuring fuzzy qualities like how safe, rich, and unique a city feels.

When we stumble down an an unfamiliar street, we often determine quickly — instinctively — if it feels unsafe. We make a snap judgment about our surroundings, and it can have tremendous implications, influencing where we travel, shop, work, and live.

Trouble is, nobody can measure it very well. We might suspect that graffiti contributes to a sense of danger and endless cherry trees to a sense of affluence, but we don’t know with empirical authority. In this sense, policy guiding urban design is, at best, educated guesswork; at worst, flat wrong. At least it was until the whizzes at MIT Media Lab came along.

Place Pulse, by the Macro Connections group, is a website that puts the full force of science behind fuzzy things like how safe or rich or unusual a city seems, and it does it in the least likely way: by crowdsourcing people’s ratings of streets, using geotagged images, and turning those answers into hard, eminently crunchable numbers. "Cities all over the world have different aspects that affect how happy we feel, how safe we feel, and how worthy we feel," says César Hidalgo, an assistant professor at MIT Media Lab. "What we’ve done is measure something we had not been able to measure before: aesthetic capital."

Place Pulse goes beyond a basic beauty contest.

The potential applications of this so-called aesthetic capital are vast. The data could aid cities in deciding which urban-design features to invest in. It could be matched with other data sets to show the relationship between aesthetics and, say, crime. It could even be used to create a website that helps us choose where to live based on the ratings of like-minded people.

Think of the site as "hot or not" for cities, says designer and Media Lab grad student Anthony DeVincenzi. Users glance at a pair of photographs side by side, then click which one’s hotter or, in the official terms of the site, which one’s safer or more unique, or more upper-class. As each image is compared, a ranking of safety (or uniqueness or affluence) emerges. The Macro Connections group can then analyze and visualize those rankings to ask all sorts of fascinating questions: Are certain cities seen as wealthier than others and if so, why? What are the ramifications of segregated public housing? Does graffiti actually impact our perception of safety? Or are cities wasting their cash scrubbing buildings, when they could be investing in truly valuable public amenities, like trees?

The site went live a couple weeks ago, so Hidalgo and his team don’t have the answers at their fingertips yet. The goal is to generate a million responses to roughly 3,000 photos of five cities: Vienna, Linz, Salzburg, Boston, and New York. They’re already about a third of the way there (that's the beauty of crowdsourcing).

A few tidbits have emerged: Of the 10 "safest"-seeming images, all 10 were snapped in Austria. Of the 10 perceived as least safe, seven were taken in Boston and three in New York. It’s easy to see why. The Austria photos feature narrow streets, foliage, and densely packed buildings with lots of bright, welcoming windows — they’re practically the Jane Jacobs ideal. The Boston and New York photos, on the other hand, show broad, desolate streets and fortress-like buildings.

Okay, great, so Austrian cities look better than New York and Boston. What else is new? Hidalgo is quick to point out that Place Pulse goes beyond a basic beauty contest. He can conduct controlled experiments, for instance, by digitally altering images. So say he removes graffiti from a picture in Boston or adds graffiti to a photo in Vienna. Then he can compare people’s ratings of the original picture to ratings of the altered image to get a feel for whether graffiti — or even what type of graffiti (street art versus tagging) — is seen as a sign of a bad neighborhood. That, in turn, could be precious information for cities that spend untold dollars removing street art from buildings. It’d reveal whether the effort’s worth it or not.

"The long-term vision is to be able to create a site that’d allow anyone to develop a study," Hidalgo says. "They’d decide the cities they want to compare and ask the questions they want to ask." Phil Salesses, a technologist and a grad student at the Media Lab, adds: "So if you’re a 28-year-old female with kids and you’re trying to figure out where to live, you could filter the data to include only 28-year-old females with kids."

Hidalgo can conduct experiments by digitally altering images.

There are limitations. It’s possible that the photos — some lifted from the web, others provided by Salesses who rode around Austria on a scooter, zapping pictures for three weeks — don’t actually show what they’re supposed to show. Photography is a tricky thing, and it can make a lively street look like a wasteland and vice versa. The researchers’ antidote: "For every image we have, we have an image of the same location rotated 180 degrees," Hidalgo says. "What we can do later when we have enough data is we can ask: ‘Do images from the same location get the same ratings?’ If they do we’re really capturing something about the location."

There is also the snag of photographs being snapshots in time that don’t capture every last nuance of our perception. They can’t tell you how people move around a neighborhood or how fast cars zoom down the street. Anecdotally, I can tell you that I find parts of downtown L.A. charming in pictures and outright terrifying in real life because people drive like it’s the last lap at Daytona. That’s something Pulse Place doesn’t necessarily pick up on.

Perhaps the biggest risk, though, is how aesthetic capital is used. You could see a city, desperate to project an image of safety and equality between neighborhoods, sinking cash into trees and broad sidewalks and whatever other amenities the data suggests instead of tackling underlying problems like poverty and crime. As for individuals: Yeah, it’d be great to be able to run a study that tells me what neighborhoods in New York other 30-year-old female bloggers in domestic partnerships favor. But do I really want to live with a bunch of people exactly like me? And won’t that exacerbate the very segregation that the city is trying to combat? Where public aesthetics are concerned, sometimes ignorance is bliss.

Not that it’s the job of Hidalgo, Salesses, DeVincenzi, and Mauro Martino, the project’s fourth member, to regulate how people use — or abuse — the information. Ultimately, it’s a way to put a number on something we all know to be important but have never had the tools to analyze in any meaningful way. More data is always hot.

[Top image shows which areas of Salzburg people perceive as safe (in green) and unsafe (in red); courtesy of Anthony DeVincenzi]

Add New Comment

5 Comments

  • Djamila

    Nice application. When I was voting for witch place is safe I noticed that some pictures are always darker than others. I wonder why is that.
    I prefer a vote based on  real pictures/not manipulated because I hate to feel that the sun is the only factor that makes a place feel safer.

  • Paul Bagne

    True crowd sourcing this is not. A better experiment would have Tweeps reporting on what makes them feel safe or unsafe as they make their way through cities in their daily lives and to illustrate such responses to urban environs by sending along photos.

  • JanBergman

    Well written article on the pros and cons of "smart/intelligent city" techniques promissing a lot of benefits in the planning process through crowd sourcing, inexpensive sensors and other data capturing.

    I worked in the IT/Telecom sector waiting 10-15 years for third party applications to catch on so perhaps it will take some time yet until municipalities in my native Sweden start using these new planning tools. Then again, with the strong focus on sustainability and big players such as IBM, Siemens, Cisco, HP et. al making this their main future market it might speed up the uptake.

    Jan Bergman