The week of October 19, 2014

The rise of the automated watchdogs

By Rob Price

The online realm is a political battleground. From politicians’ Ask Me Anything (AMA) interviews on Reddit to bitter edit wars fought across the pages of Wikipedia, the Internet is where campaigns are launched and the never-ending jostling for public perception takes place.

As we transition from an age of too little information to too much, traditional fact-checkers, nongovernmental organizations, journalists, and other watchdogs have struggled to keep up with the pace and reach of the Internet. It’s one thing to keep tabs on a politician’s voting record and bankroll—it’s quite another to monitor all of the information being dispersed across Twitter, Facebook, and YouTube, to say nothing of their countless, tangentially related Wikipedia pages, where even the slightest tweaks and changes can help steer a larger conversation.

That’s why a new breed of digital watchdog is fast emerging—and they need your help.

The hunting grounds for many of these watchdogs is Wikipedia. The online encyclopedia has become the Internet’s de facto public record and, according to founder Jimmy Wales, is trusted more highly than the BBC or the New York Times, coming second in credibility only to Encyclopedia Britannica.

The fact that anyone can edit Wikipedia makes it ripe for vandalism and abuse by those with an axe to grind.

The fact that anyone can edit Wikipedia makes it ripe for vandalism and abuse by those with an axe to grind, which, combined with its credibility, makes it essentially the digital equivalent of a swing state—it’s where the real attacks happen. Wikipedia, however, is also structured in such a way that every edit is resolvable to an IP address or user account. This traceability has thrown light on examples of politicians trying to subtly alter their public image. For example, the page for former British MP Joan Ryan was edited numerous times by “someone,” both in the House of Commons and in Ryan’s home constituency, to remove references to her implication in the Expenses Scandal of 2009.

There have been numerous other edit scandals in Britain, some far more serious than Ryan’s breach of Wiki etiquette, including vandalism and abuse directed at the victims of the 1989 Hillsborough football disaster in which 96 people died. The edits were eventually traced back to a now-fired government employee. Another incident caught by digital vigilantes was the use of government computers to alter information on the killing of Jean Charles de Menezes—a 27-year-old Brazilian wrongfully shot to death by police who suspected he was a terrorist, in the days after the July 2005 London Underground bombing.

The Hillsborough case was discovered in an investigation by the Telegraph, while Britain’s Channel 4 News brought the de Menezes incident to light. In August, Channel 4 did something unprecedented: It created its own automated Twitter bot, @WhitehallEdits, designed to automatically monitor the online encyclopedia for edits made by known government IP addresses, and tweet out any that it finds.

Channel 4’s actions marked a turning point: A major news organization adopting the mantle of Internet watchdog, patrolling the digital sphere for bad behavior—and doing it in an entirely automated fashion. @WhitehallEdits is by no means the first algorithm to keep tabs on critical changes to the Web’s canon of knowledge, however.

For starters, there’s @ParliamentEdits, a Twitter bot created by the British tinkerer Tom Scott and acknowledged as the direct inspiration for @WhitehallEdits. It monitors Wikipedia edits made from the Houses of Parliament itself.

The automated watchdog revolution isn’t restricted to the British Isles. Scott’s actions inspired an American, Ed Summers, to take a stab at creating a similar bot in the U.S., @CongressEdits, and there’s a Canadian version, @gccaedits, which aims to inject a degree of transparency into the political process in the Great White North.

@CongressEdits has provoked a flurry of media attention and speculation over rogue edits from the heart of American democracy. The Twitter bot has enjoyed numerous scoops (of varying seriousness), including someone with access to Congress’s network—not necessarily a congressman, we must stress—describing whistleblower Edward Snowden as a “traitor,” and the labeling of former Secretary of Defense Donald Rumsfeld as an “alien lizard.” The bot’s record-keeping eventually forced a Wikipedia admin to slap Congress IP addresses with a 10-day ban to stem the flow of vandalistic edits highlighted by the account.

Surprisingly, Wikipedia founder and champion of open government Jimmy Wales spoke out against the bot. “There is belief from some of the community that it only provoked someone—some prankster in the office—to have an audience now for the pranks,” he said, “and actually encouraged them rather than discouraged them.”

There is, of course, one obvious limitation to all these Wikipedia-monitoring bots: Edits are traceable back only to an IP address, rather than an individual. At best you might be able to pinpoint a particular politician’s office; more likely you’ll just know it was someone in the building.

Luckily, Wikipedia is by no means the primary public platform through which the political debate is filtered. Twitter stands supreme as the entire world’s office watercooler, where politicians, trolls, journalists, activists, and everyone else rub shoulders uneasily together.

No matter how clever these bots become, how sophisticated their code, how ardently they monitor the digital record, they will always be limited.

Enter Politwoops. Originally an initiative of a Danish nonprofit, the Open State Foundation, Politwoops trawls Twitter, keeping tabs on politicians by firing out alerts whenever one deletes a tweet. This could be because the tweets were only intended as temporary—or because the politicians in question made an inconvenient statement they hoped would disappear. But the Internet, of course, never forgets.

Politwoops is now available in dozens of countries and languages, from Turkey to the Vatican; Nicko Margolies heads up the American arm of the operation in concert with the Sunlight Foundation, which aims to “empower journalists and citizens to keep public officials accountable,” Margolies told The Kernel. The U.S. version of Politwoops has become a vital part of that since its launch in 2012, complimenting the Foundation’s existing suite of watchdog tools, which include press-release-rewrite trackers, campaign finance monitors and lobbyist registration trackers.

Politwoops U.S. has now logged more than 13,000 deleted tweets and hopes to be “a powerful tool in any modern watchdog’s toolbox,” Margolies said. “It allows you to see the full picture of a politicians’ public statements, rather than just the current version of their talking points.”

Politwoops U.S. has now logged more than 13,000 deleted tweets and hopes to be “a powerful tool in any modern watchdog’s toolbox.”

Margolies cites Rep. Paul Cook from California as “a politician attempting to hide a prior statement. His bot caught Cook “[deleting] a tweet and press release about Attorney General Eric Holder’s resignation.” Subsequently, Margolies said, Cook’s office “repeatedly ignored requests for comment, and if you got to his website you won’t see any mention of this release, but Politwoops has the entire statement archived.”

Margolies points out another high-profile incident Politwoops was able to highlight: a half-dozen politicians welcoming U.S. soldier Bowe Bergdahl back to the U.S., only to subsequently delete their messages when Bergdahl’s case became a political football.

With a little technical know-how, Politwoops data is exportable into automated Twitter feeds for easy access—as it the case with the British iteration @deletebyMPs. One of its most popular tweets—a deleted tweet from Tory MP Robert Halfon memorializing the late entertainer Jimmy Savile, since revealed as a serial pedophile—perfectly highlights how politicians seek to quietly distance themselves from unfortunate statements.

In the spirit of openness, the source code for the Wikipedia bots is available to the general public, allowing anyone to take it, customize it, or set up their own. It’s in keeping with the empowering nature of this new breed of digital watchdog—automating tasks that would take prohibitively long to even contemplate doing manually.

No matter how clever these bots become, how sophisticated their code, how ardently they monitor the digital record, however, they will always be limited. Without real people following them, taking action when issues raise a red flag, they’re little more than canaries expiring in coal mines unheeded.

Our digital watchdog overlords can only be as powerful as the people who give them life.

And that’s where you come in.


Photos via ttarsiuk/Flickr (CC BY 2.0), Shuichi Aizawa/Flickr (CC BY 2.0), Cindy Funk/Flickr (CC BY 2.0) | Remix by Rob Price