‘Is Twitter Wrong?’ became central to debunking during Hurricane Sandy
Tom Phillips, an international editor at MSN based in England, started “Is Twitter Wrong?” in August to debunk misinformation coursing through the social sphere. His Tumblr rose to prominence quickly this week as he sorted real photos from fake ones during Hurricane Sandy. Earlier this week, Phillips said correct tweets on Twitter is "like putting toothpaste back in the tube, except the toothpaste is alive and didn’t like it in the tube and is dreaming of Broadway." Curious about his work, I emailed questions to Phillips on Tuesday, he responded overnight Wednesday. Our edited email exchange appears below.
Poynter: The first post I see on the 'Is Twitter Wrong?' Tumblr is from August. What was the inspiration for starting it then?
Tom Phillips: It was something I'd been mulling for a while; the need for something Snopes-like, but focused on quick-turnaround verification of stuff on social media. I'm a big fan of the rise of fact-checking as up-front journalistic content, rather than just background process -- but for some reason, the major fact-check sites tend to concentrate more on important claims made by serious people who run countries, and less on whether @topbants473 has really taken a picture of a tiger in his back garden. (There are occasional noble exceptions to this rule.)
I think the original thing that tipped me over the edge into actually doing it was a spate of pictures supposedly from the Mars Curiosity rover, which had just landed. Most of them weren't (they were largely from previous rover missions), but there were so many tweets going around that trying to track them all down to tell people they were wrong individually seemed impractical. Not to mention a bit dickish. So I set up the Tumblr instead.
Before that, I'd often find myself doing stray bits of fact-checking things on Twitter, and finding that it was a) remarkably quick and easy to debunk a lot of it, and b) really hard to stop it spreading even after it had been debunked. I figured that it might be easier if, rather than playing an endless game of chase-the-tweet, you had somewhere off-Twitter that both gave you more space to show people evidence, and also provided a permanent URL that you could easily share -- and which other people could then spread for you.
So, partly it was just to scratch an itch, but it was also deliberately an experiment to see if something like that could work, from a journalistic point of view. I'm a big fan of using Tumblr as a quick-and-dirty way of trying ideas out in your spare time. It's littered with the bones of works-in-progress and good-ideas-at-the-time that never went anywhere. Given that we've got platforms which make it so easy (and free) to set things like this up, it seems a bit weird to not use them to prototype your content before committing major resources to it -- to see if there's an audience, see if the content actually works, see if the workflow is manageable or if it gets crazy.
In Monday's post you got on a bit of a roll, debunking photo after photo that claimed to be of the current storm. Can you share a bit about the process you went through for doing that?
It happened in a pretty haphazard fashion. I certainly hadn't planned on doing rolling debunking. Initially, the post was just a response to the first notable false picture that was being passed around, plus a few usual-suspect storm pictures thrown in as a sort of pre-emptive strike. Really, it was mostly an excuse to talk about cloud formations, because I am a very interesting person who is a delight at cocktail parties.
But from there -- possibly because of the time it went up, early afternoon London time, but when a lot of [people on] the U.S. East Coast were starting their day -- it sort of snowballed. People started getting in touch asking about other images, or pointing out fakes that they'd seen, so I kept adding them to the post.
I'm not quite sure when I realised that it had become a de facto hub for all of the day's dodgy pictures, but I think once Wil Wheaton reblogged it, it became apparent that it had become sort of A Thing.
At that point -- still several hours before landfall -- almost all of them were pretty easy to disprove (mostly, a simple reverse image search was enough, or other techniques such as "remembering what happened in 'The Day After Tomorrow'"). At the same time, Alexis Madrigal had started doing his InstaSnopes thing over at The Atlantic, and somebody tweeted at us that we should team up. So we did, at which point we got sucked into a swirling vortex of Photoshopped shark pictures.
That's when the fact-checking became a lot harder. Suddenly, we had images that couldn't be reverse image searched, and where the original provenance was uncertain (and harder to track down when things are moving so fast). So we were falling back on, variously, scouring StreetView and Bing Maps for buildings that matched with pictures, reverse-engineering Photoshops by image searching their constituent parts, looking for mentions of locations on Facebook, and often just asking Twitter for help. Not long after landfall, the Instagrams were coming so quickly that it was impossible to keep up, so it was a question of picking the ones that, essentially, looked most interesting.
(By the way, one reason Alexis's format was a huge improvement on mine: the idea of putting the Real/Fake labels on each picture, which was very simple and very smart. Not only did it look rather stylish, but it encouraged people to share the fact-checked images without worrying that they could then be taken out of context again. Given that the rationale I had for starting ITW was that rumour control works best when you make it easy for people to share the results of the fact-checking, I'm kicking myself for not thinking of it.)
Given this little niche you've carved out here, what's your worldview on the reliability of social media information and how the public and the media should use it? Do you think Twitter is a net positive in spreading true information efficiently -- or have your debunking efforts soured you on it?
I think it is a net positive, although only just -- and it takes work to keep it that way. I'm still a bit bemused by how casually some people spread things that are flat out wrong, but I think most people have a general desire to not be talking absolute bollocks. Or at least, to not be seen to talk absolute bollocks.
The Guardian's great analysis of how Twitter rumours spread (and died down) during the London riots suggested how finely balanced it is. Plenty of the rumours explode rapidly, threaten to get out of control, but are eventually tamped down thanks to a combination of both journalists and the general public going through the motions of being sceptical, asking for facts, and making sure good information is shared widely enough to drown out the bad.
I know people are getting a bit meh over the phrase "citizen journalists," for all sorts of good reasons. But when you're looking at the basics of fact-checking and making sure false rumours don't multiply, it really has to be a mass participation activity.
We've got all these platforms built for rapid spreading of information, but we don't really have any architectures of verification.*
So it needs social structures to make up for that lack; standard ways of interacting that help us get to the truth, and a degree of social censure that provides a disincentive to spread misinformation.
Little functional ways of pushing ourselves in that direction can be newsrooms building this kind of verification into their workflows, journalists collaborating with each other across organisations outside of their newsroom structures, journos and non-journos working together, or the public simply getting on and doing it themselves regardless of what the hacks are up to.
As John Herrman notes, traditional newsgathering in intense or fast-moving situations can -- with the best of intentions -- be as dangerous a source of false rumour, and sometimes it can take the combined force of multiple minor acts of verification to push back against that. And that's the key point; much of this really is something everybody can do, in minimal time, with free and publicly available tools. There's no secret journalist magic to it.
*Total sidenote here, but I hope somebody, somewhere is working on some kind of platform for collaborative, real-time verification. Obviously we already have lots of platforms for collaborating, and lots of tools for verification, and lots of sites and services that specialise in collating widely-sourced information, especially around major events. But I'd really like to see something that puts the kind of ad-hoc working-together-to-check-a-firehose-of-dubious-stuff that loads of us were doing yesterday at its core -- in the same sort of way that, say, Branch is taking conversation really seriously, or Storify makes curating user-friendly. Obvious caveat: Maybe it already exists, and I just don't know about it.
Editor's Note: Craig Silverman has proposed just such a collaborative, real-time verification architecture. Details here.