The myth of Twitter’s power to self-correct

Last August, lions and tigers were released from the London Zoo and let free to roam the posh Primrose Hill district. Or at least, this is how it might have seemed if you were following that summer’s London riots via Twitter, as a number of high-profile users on the social network site re-tweeted news to that effect. In fact, as the British newspaper The Guardian later showed, the story about escaped animals was only one of the subsequently debunked claims about sensational happenings during the riot that circulated in the Twitterverse that summer.

In the weeks following Hurricane Sandy, a similar story has emerged. In his Nov. 4 New York Times column “Twitter’s Uneasy Role Guarding the Truth,” Nick Bilton praised Twitter for its “self-correcting mechanism” that allowed users to quickly debunk falsehoods circulating in the aftermath of the storm. It didn’t take long, he observed, for users to identify a shot of a tidal wave engulfing the Statue of Liberty as an image stolen from the movie The Day After Tomorrow. This self-correcting virtue of Twitter, he argues, means that Twitter shouldn’t be held responsible for the truth of its content.

While Bilton and others’ faith — that with social media the truth will eventually out — is touching, the more disturbing reality is how quickly we have accepted social media, and crowdsourced truth in general, as an inevitable force for good.  In fact, the quick identification of fraudulent photos or information, however impressive and inspiring, does not prove and does not establish that Twitter is a verifiably “self correcting” social network. Or, as some optimists alarmingly suggest, that Twitter should be a self-regulating one, either.

Does Twitter indeed self-correct?

In some cases, yes: In fact, knowing that “everyone lies,” as Bilton puts it, some news outlets, including The Atlantic, set up debunking projects to test the notion. The Atlantic’s led to the identification of fraudulent photos such as the submerged Lady Liberty. But often fraud and satire aren’t easy to spot: Artist Ti Kawamoto recently created a a highly detailed 3D model of a hypothetical future Sony phone to see if he could fool Twitter sleuths into believing that images of the phone were real leaks. He did.

But the real point here is that most false claims on Twitter are much harder to identify than a phony sinking statue. For instance, where’s the self-correcting mechanism at work in Ann Coulter’s Twitter feed, when she re-tweets, “Obama won 99.5% of the vote with 90% turnout in Philly? coughBScough.”  OK….but with “a total lack of evidence that any fraud occurred” (according to David Weigel at Slate), is it responsible to let a patently false statement stand, and echo through the Twitterverse? The result is that despite seemingly incontrovertible evidence, millions of Americans are now convinced that President Obama was not the legitimate winner of the election. “Truth” is hard to establish even in the least complex situations. To say Twitter has a self-correcting system to ferret out every form of canard is just not true.

What is Twitter’s responsibility?

Another reason we shouldn’t be so quick to assume that Twitter’s users will automatically converge on the truth is the role that Twitter itself plays in this process. The structure and rules of Twitter have been designed to promote particular types of conversations, and to tip the power of truth towards particular individuals and companies.

First, the Twitter Rules, and in particular the section Spam and Abuse, show how Twitter already regulates, and has claimed the power to regulate, the type of speech it wants to promote in on its site. As the rules note:

Some of the factors that we take into account when determining what conduct is considered to be spamming are:

? If your updates consist mainly of links, and not personal updates;

? If you post misleading links;

? If a large number of people are blocking you;

? The number of spam complaints that have been filed against you;

Obviously these rules have important uses for stopping malicious porn bots or other types of spam. But, before we assume that Twitter self-regulates the truth, we should ask ourselves who defines “misleading?” For instance, in a high-profile case this year, Chris Loesch, husband of conservative media figure Dana Loesch, instigated a public brouhaha after his Twitter account was suspended for seemingly dubious reasons. Loesch’s suspension was overturned, re-instated, and then finally overturned later but the example shows how Twitter’s rules are enforced in a manner that is neither consistent nor transparent.

Supplementing these formal procedures, there are ways that Twitter informally shapes the nature of what is “true.” In addition to Twitter-suggested trusted lists, by anointing particular accounts as “verified,” Twitter endorses some users, building in a legitimizing function that operates independently of the actions of its user base who have no access to similar procedures.

This is not to say that Twitter uses these guidelines maliciously. However, it is important to acknowledge that it is managers or editors at Twitter, and Twitter alone, who have the authority to apply these rules in any way that they deem fit. And further, since Twitter is a private company, management is not beholden to the guarantees of expression that a public forum would provide.

Does this matter?

Appearances notwithstanding, Twitter and other social networks already regulate the nature of our discourse. I share Bilton’s hopes that Twitter will prove itself a tool for public good that helps bring truth to light. For this to take place, though, we cannot simply assume that its users will give fair attention to conflicting points of view, or to inconvenient evidence that supports opposite conclusions.

Public goods rely on structures (e.g. due process laws) to insure that the minority is heard. So while the popular success of social network sites, and their young, still evolving forms may lead some supporters preferring to let them develop their own systems for regulating content, it is only by making these regulations public that we can ensure that our new public forums act in the way that some believe they already do.

Monitoring all the Tweets produced on the service would obviously be infeasible. However, by making the process by which they censor and regulate content more transparent and open to public scrutiny, Twitter’s management can show that they take free speech as seriously as their supporters claim they do.

Mathias Crawford  is Stanford Graduate Fellow and Values in Design Fellow researching the architecture of digital media. Follow him on Twitter @mfcrawford.