Infowar and Measuring Authenticity

I had an opportunity to share my observations about “citizen propaganda” in the Russia/Georgia conflict with someone far more knowledgeable than myself – Gregory Shvedov, the editor in chief of Caucasian Knot, a leading alternative news publication focused on the Caucuses, joined the fellows at the Berkman Center for a brief discussion yesterday. Shvedov has been following the conflict closely, both through the reports of journalists associated with Caucasian Knot who are on the ground in Georgia and Ossetia, and through the panoply of blog accounts that have accompanied the war.

Shvedov confirms that there’s been deep interest in “eyewitness accounts” of events on the ground, and that the blogs he’s read seem to include people in Moscow or Tblisi misrepresenting themselves as being on the ground, usually to forward a political agenda. He worries that people are paying more attention to these bloggers – some of whom are misrepresenting themselves – than to reporters he and other publications have on the ground.

This situation presents interesting challenges for those of us who advocate for citizen media or who are involved with projects that aggregate and promote citizen media. If it becomes clear that many blogger eyewitnesses are simply amateur propagandists, it’s going to be harder and harder for journalists to reference and amplify the opinions of bloggers on the ground. It puts an interesting pressure on projects like Global Voices – in citing a blogger writing from Ossetia, are we authenticating that blogger, certifying that the person in question is really an eyewitness?

We can’t offer that authentication – that’s not what we do. We study the blogosphere closely, and we know who’s been blogging for a while, and can often offer background information on the bloggers we amplify. But we’re not a news agency, we don’t have reporters on the ground in Ossetia, and we’re discovering that there’s a real challenge presented by fast-breaking news events: news inspires people to start blogging. It’s not hard to believe that someone in Ossetia would begin blogging shortly after Georgian and then Russian troops came into the region – that blogger would be an eyewitness, and would likely have some interesting insights on the conflict. But she wouldn’t have a reputation or a track record, and it would be difficult to evaluate whether she was actually on the ground, or writing from either Tblisi or Moscow.

We all make decisions about the authenticity of information we receive, either online or offline. I tend to take a story more seriously if it comes from the New York Times than if it comes from the Montgomery County Bulletin. (Friends of mine who lean to the right may suggest that this is an error on my part.) We rely on the reputation and brand of the Times, and the fact that the paper is widely read and fact-checked by bloggers and readers. The Times gets it wrong sometimes – Jayson Blair, Judith Miller – but the reading public tends to be quite aware of those errors.

Reading blogs demands a different strategy. Reputation matters, but it’s less about brand than about longevity, links and comment threads. It’s not hard to manufacture a single, incorrect blogpost and have it survive some scrutiny – it’s harder for a blog to survive scrutiny over a long period of time. If a blog has no history, no incoming links, no active comment threads, it’s hard to value the opinions expressed there any more (or less) highly than those from a stranger on a city street. A blog that’s made controversial and questionable statements will often generate discussions, either on comment threads or on blogs linking to the blog – reading those threads often helps establish whether this blogger is viewed as reliable or with skepticism, as mainstream or fringe within a particular debate. These methods certainly don’t guarantee that a blogger is neutral or non-partisan – they simply make it easier to determine whether that person is a respected voice in an existing community of interest.

These reputation mechanisms don’t work well in the case of breaking news. We don’t check the previous blogposts or the incoming link count of the people who sent mobile phone photos from the underground during the 7/7 London bombings. Those images were believable to the recipients because they came from known contacts – they were believable in the press because the photos were consistent with other individual and professional press accounts. How do we make decisions about authenticity of observations coming from a war zone that’s not closely covered by the professional media?

My sense is that we’ve generally assumed good faith in reading citizen accounts of wars and other disasters. We may need to re-examine that assumption. As citizen media becomes a more influential and closely watched space, there’s a greater incentive to fight information wars on blogs and video-sharing sites, not just in the traditional press.

As I mentioned in my earlier post on citizen propaganda, the Ossetian conflict isn’t the first time we’re seeing propaganda efforts in digital media. We’ve seen evidence both of professional, paid posting by the “fifty cent party” and well-organized amateur efforts in the Chinese blogosphere, especially around riots in Lhasa.

We’ve covered these efforts on Global Voices because our job isn’t to report news, but to report what’s being discussed in a nation’s blogosphere. If that discussion includes amateur propaganda videos, or organized campaigns to post pro-government comments on blogs, we need to cover those posts… but we also need to provide context, helping people understand where these comments are coming from. Are these the spontaneous expression of angry individuals in China, Georgia, or Russia? Are they part of a larger, orchestrated campaigns? How do we value an opinion expressed as part of an organized campaign versus the opinion expressed by an independent individual?

My colleague Judith Donath offered a very useful framework in considering these questions. She suggests that what we’re trying to determine here is the “authenticity” of a particular voice. There are a number of factors we consider in considering authenticity… and perhaps we overvalue some and undervalue others.

In discussing the “fifty cent party” (a group of bloggers paid by Chinese authorities to publish pro-government comments on blogs and in forums), the fact that commenters are being paid makes me tend to take their contributions less seriously. It isn’t clear that we should draw this conclusion – it’s quite possible that the people who choose to work with Party authorities are generally in agreement with Party position on most matters… but somehow the presence of money causes me, personally, to consider these comments as less authentic.

Spontaneity also appears to be a characteristic that we correlate with authenticity. If I find a comment on a New York Times story comparing Vladimir Putin to Vlad the Impaler, it looks like a spontaneous, heartfelt response. If I find 100 comments all making the same analogy, it looks like an attempt to organize a campaign drawing associations between a world leader and the inspiration of vampire novels. Should I take the apparently spontaneous comment more seriously? Doesn’t the attempt to engineer a meme about Putin and vampires reflect a seriousness of sentiment that’s at least as authentic as an offhand comment?

Remember, the goal here isn’t to determine the accuracy of a statement, but the authenticity. Are people in China really angry about Jack Cafferty’s characterization of Chinese authorities as “goons and thugs” or was this simply an orchestrated campaign? Are thousands of angry Russians commenting on New York Times coverage because they were authentically pissed off? Because someone’s organizing a campaign to encourage them to do so? Does it matter if such a campaign is organized by a clever, web2.0 activist or by the Kremlin? What should we consider as we try to evaluate whether individual blogs are authentic or orchestrated? (Or both?)

Donath had another intriguing suggestion – she wonders whether large-scale text analysis could help detect blogposts that are part of orchestrated campaigns. Assume that there’s a wave of blogposts about, oh, say Sarah Palin. It might be possible to automatically identify blogposts that use similar phrases to attack or defend the VP candidate. A wave of posts all making the same three critiques or offering a similar worded defense might indicate organized campaigns to spread certain talking points. (Or they might just point to the tendency of bloggers to cut and paste arguments from people they read and like.) Even if such a tool didn’t help us figure out the authenticity of sentiment, it could help us avoid reading another half-dozen repetitive posts on the same topic.

Is authenticity the right characteristic to consider when evaluating blog accounts from conflicts? How should we evaluate authenticity when it’s difficult to verify the identity and location of an author?