As a writer this is a painful thing to admit, but here goes: There are times when photos can tell more of a story than words could ever express. When news broke of last month’s tsunami and resulting destruction in Japan, what story could possibly convey the depth of the tragedy like Alan Taylor’s In Focus images at The Atlantic? Or the BBC’s striking “In Pictures” feature?
Part of the reason images like those and others are so powerful and telling is that there is an inherent trust that goes along with photography, particularly news photography: Seeing is believing.
And yet we also live in a time when we’re just as prone not to believe that a picture is real. We use “Photoshopped” in casual conversation and even applaud new achievements in fakery. (Don’t pretend that photo of Joe Biden working on his sweet Trans Am isn’t cool.) And if you’re a professional photojournalist or a news organization that wants — and needs — people to trust the images you publish, that’s a big problem.
“The assumption a lot of people make is, well, pictures don’t lie — you can believe what you see,” said Santiago Lyon, director of photography for the Associated Press. “But of course pictures can lie, and they do lie, and they’ve been manipulated for a long time.”
Lyon and Hany Farid, an expert in the emerging field of digital forensics, were in town last week for a talk at MIT about photojournalism in the age of Photoshop. What the two men share is an interest in establishing systems to ferret out manipulated photos, with Lyon focused on policy while Farid looks to math (the Lab had an extended conversation with him about that last year). Their differing approaches make sense: Lyon oversees about 300 photographers, and Farid is a computer science professor at Dartmouth.
While they spend a fair amount of their time trying to find ways to combat photo manipulation, they say their bigger concern is our cultural acceptance of that manipulation and the resulting erosion of trust in photojournalism. But here’s the good news: Skepticism being alive and well is not entirely a bad thing.
“The camera can see things and report things quite differently than the human eye does,” Lyon said. “I would say we should approach each photograph we look at with the same skepticism that we bring to each story we read.”
That skepticism counts inside and outside the halls of the AP, which Lyons said uploads around 3,000 images every 24 hours, totaling around 1 million a year. For photographers and photo editors, the organization has an established policy on the limits of altering photos through editing (a little dodge and burn here, some cropping there), as well as additional considerations when dealing with photos submitted by citizen journalists. Still, to a large degree, it falls on editors to look critically at what they see before them in a photo and ask questions of the photographer or person who submitted the image.
“What we’ve learned very quickly is that words, and near words, fail us because there are so many variables in play here,” Lyon said. Farid agreed. The problem with relying on a critical eye, he said, is that it, too, is imperfect.
“Let’s agree that while the visual system is remarkable and we are very good at interpreting photographs, we are not perfect and we make mistakes,” Farid said. “The scary part is we don’t often know what the mistakes are.” Which is to say that through an over-reliance on visual cues and through our analytical tendencies (perhaps informed by assumptions about Photoshop lurking around every corner), we end up having trouble discerning real photos from fakes.
This is where Farid’s work is helpful. He uses the aid of computers to break images down to their base numeric elements in order to find patterns and/or discrepancies that the eye may miss. (In fact, Lyon and Farid previously worked together to analyze a set of photos the AP knew had been altered.) That method, though — in the case of the AP set and in others — isn’t 100 percent effective. That’s one reason Farid and his team use multiple methods to test a photo’s veracity.
“There’s no magic button,” Farid said. “You have to come up with a lot of different ways to analyze photographs. At least as many different ways as there are ways of manipulating photographs.”
For the AP, it also helps to have the sort of ambient analysis that comes with having your photos appear in front of countless eyes around the world. Bloggers, Lyon said, have become sources for spotting questionable photos that wind up on the AP wire. But as helpful as they can be, reliance on them means that photo manipulation is discovered only after images are published. It also means that outside observers can quickly start chasing (in this case photographic) windmills. “We find these people can start to see things or think they see things that aren’t really there, and that feeds a kind of frenzy,” Farid said.
That kind of skepticism-run-amok, Lyon suggested, can have a detrimental effect on journalism overall: Being skeptical of a photo is a small step away from being skeptical of an entire news organization, which can ultimately undermine the mission of journalism, he said.
But as odd as it may seem, skepticism may be the new equilibrium in photojournalism. It’s hard to sell the idea of truth in photography, after all, when photo-altering products are being marketed so aggressively to consumers. Take, for example, the Windows 7 commercial where those darn kids are so fidgety and mom just wants that family photo to be perfect. Simple: Just edit in the perfection! Or, as the Windows mom puts it: “Windows gives me the family nature never could.”
The ubiquity of photo manipulation “is starting to slip into the mainstream — forget about journalism,” Lyon said. “We’re talking about consumer stuff now. And that, for me, is very disturbing, in part because I think photography has always been something reliable — even if there is a lot of subjectivity involved.”