What Flickr Can Teach Us About the Way We Photograph Cities
What do people photograph when they visit Fort Mason, an army base-turned-cultural center on the San Francisco waterfront?
Instead of heading down with a clipboard to do interviews, UC-Berkeley researcher Alexander Dunkel analyzed data from Flickr. Using geotags, which relay the exact location of the photographer, he was able to place over 125,000 photos on a map of the area [PDF], with expanding colored disks indicating the popularity of a certain viewpoints.
Using tags, with which users describe the content of the photos, he presented popular subjects as word clouds, located at the weighted center of frequency. We can see below, for example, that visitors are photographing the Golden Gate Bridge mostly from two places: the three Fort Mason piers, and halfway out on the Van Ness pier. For photographs of Alcatraz, the legendary island-prison, one viewpoint, at the end of the jetty, is predominant. The red-to-blue color scale of the circles indicates what Dunkel calls "second-level clustering" -- whether a viewpoint is marginally more or less popular than the area mean.
Courtesy Alexander Dunkel [PDF]
Between the various postcard attractions -- the Bridge, the Bay, the Marina, Fisherman's Wharf and Ghirardelli Square -- we are treated, in smaller type, to game of word association with the waterfront, a found poem of the hive-mind: "shadows," "friends," "seagull," "statue," "summer," "dusk," "ocean," "boat," "fog," and so on. Also present are the tags of popular events, like the Fleet Week Air Show, and some Instagram filters, notably between Pier 2 and Pier 3: "Hefe," "Sutro," "Amaro," "Walden."
Dunkel isn't a programmer by training; he's a landscape architect, pursuing a doctorate in engineering sciences at the Dresden University of Technology in Germany. But during the year he spent as an International Research Scholar at Berkeley's Department of Landscape Architecture and Environmental Planning, he began to experiment with mapping Flickr's easily accessible metadata (with help from mapmaker-artist-programmer Eric Fischer).
His findings, which won him an Honor Award at the American Society of Landscape Architects 2012 Student Awards, deliver a new, precise understanding of how people see what's around them. "This might help you, as a landscape architect, to better design a spot," Dunkel says. "You can say, here's the perfect spot, I want to provide access. You can decide how you want to react in your design."
In Yosemite National Park, Dunkel ran a similar program, using "sightlines" to connect popular subjects with popular viewpoints. "There are just a few single important aspects in the landscape," says Dunkel, which allowed him to create a much simpler map, looking at tags for only four locations: Glacier Point, Yosemite Falls, El Capitan, and Half Dome.
Here, gray indicates the frequency of a location as subject, as measured by tags, and red indicates the frequency of a location as a viewpoint. The towering granite cliff face of El Capitan is the subject of many photographs -- but not, being hard to reach, a popular viewpoint. Valley View, the opening viewpoint of the park (the red disk at left) has the opposite ratio. (On his website, Dunkel has a similar map showing sightlines for the Golden Gate and Bay Bridges in San Francisco.)
For planners, a few hours of data mapping can demonstrate which viewpoints are popular, and what views they provide. The Park Service could allot parking spaces, maps, or restrooms accordingly. Conversely, mapping the data demonstrates which areas are being underutilized, and forces planners to figure out why that might be. Sorting by date would reveal whether people prefer different viewpoints -- or different views -- depending on whether it's winter or summer.
The software can also track the influence of a project. Dunkel ran two iterations of the program on Manhattan's West Side, before and after the opening of the High Line. The contrast offers strong evidence for how a park changes the way people perceive a neighborhood. Here's Chelsea pre-2009:
Of course, the proliferation of geotagging, smart phones, and photo-sharing in the past three years plays a large role in how dense the second map looks compared to the first. But the concentration of growth is clear: the path of the High Line bending gently up 11th Avenue.
Dunkel has created similar maps of Dresden, Saxony, and of large swaths of the Bay Area. "It gets easier and easier to automatically use this information," he says. He thinks the growing interaction of social media and place -- and the data it produces -- has manifold potential uses.
But he has concerns about privacy, too. "A lot of people don't know their data is being used," he worries. "If you want to use this data in a public project -- from an ethical point of view, you want to let people know when their data is being used."
All images courtesy of Alexander Dunkel.