An American research team is taking a scientific approach to tracking potentially abusive activity on Twitter in Canada's first so-called "social media election."
The University of Indiana team tracks memes — ideas that go trend online — in an effort to find ways to automatically detect smear campaigns, the spreading of misinformation and instances of astroturfing, or giving the impression of a grassroots campaign where there is none.
Their website, named Truthy in homage to U.S. comedian Stephen Colbert's use of the term which means something that seems like the truth but has no basis in fact, began as a research project assessing the Twittersphere during the 2010 U.S. midterm elections. A Canada election section was launched on Monday.
Fil Menczer, associate professor of informatics and computer science at Indiana University and the principal investigator for the project, says his team found several suspicious examples of accounts Tweeting about the Canadian general election using spamming techniques that appear to violate Twitter rules.
One was a Twitter account called @calgary62, where an unidentified user with an unabashed Conservative slant repeatedly microblogs the same Tweet, using a different hashtag.
"This user is using a lot of different hashtags to inject his content into the streams of lots of people," said Menczer. "So he's just trying to grab attention with the same message from all these different streams hoping that somebody will pick it up and re-tweet it."
Among the recent repeat Tweets is one that states: "BLOC tried to extort $5-B list of demands to vote for fed budget. Now JACK LAYTON's socialist NDP wants to govern with BLOC." It was sent at least two dozen times, with a different hashtag on the end of each one.
The user, identified as Old Liberal, did not respond to a request for information about the account. Nor did another Twitter account identified by Menczer's group as suspicious: @mrprorogue.
That account uses another method Menczer's team first noticed during the U.S. election, which involves a user sending out identical messages directed at different users. In this case, the @mrprorogue user promotes the website, FireHarper.com, in almost every Tweet. The website is registered by an anonymous user.
"He'll find somebody who is popular, who tweets a lot or who has a lot of followers and mention that person in conjunction with this URL in the hope that one of these people will retweet it and that will generate a cascade," said Menczer.
Retweets increase believability
Both the Conservative and Liberal parties denied any connection to the @calgary62 and @mrprorogue accounts. Conservative Party spokesperson Ryan Sparrow said the party does not use any Twitter accounts not openly identified with the party.
Liberal Party spokesperson Kate Purchase said the party doesn't monitor all members' Twitter accounts, but the party's three official English language Twitter accounts — and their French counterparts — are the only Liberal-sanctioned accounts and they are controlled by the party.
Other egregious methods observed by the Indiana team in the U.S. election were supporters "donating" their accounts to an organization that then automatically generated tweets via their accounts or a single user setting up multiple accounts to retweet itself in order to appear more popular.
The two Canadian examples appear to violate Twitter's rules. The site states that an account abuses the service if it impersonates someone to deceive or mislead others, creates serial accounts for disruptive or abusive purposes or uses Twitter with the purpose of spamming. An abuse can result in permanent suspension.
But beyond breaking the social media site's rules, Menczer says the practice is troubling because the retweeting of spamming accounts lends an air of so-called 'truthiness' to a claim that increases its believability.
"When people hear the same message from lots of different sources they are inclined to believe that it is true," said Fil Menczer. "The danger is it might dupe people into having the impression of reality that is distorted in addition to propagating false news to smear someone with something that is not true."
No election-related laws in Canada deal directly with the use of social media.
Under the Canada Elections Act, it is, however, forbidden to affect election results by knowingly making or publishing false statements about candidates.
Another section specifies that any election advertising — which could include paying someone to Tweet — that is paid for by a candidate or political party or a person acting on their behalf must state that it is authorized by the candidate or party.
Elections Canada spokesman John Enright said he'd not heard of any such social media-related violations, but the tool is relatively new to the campaign trail.
But the truth is Twitter accounts and other online communication operates in a realm where anonymity is the norm, making creators difficult if not impossible to track down.
"It is easy not to get caught if you're a little bit careful," said Menczer.
Early in the federal election campaign, a Craigslist ad seeking writers to post right-wing comments to social media and news outlet made the rounds. A Conservative campaign chief denounced it as "clearly a fake" and media outlets questioned whether it was a plant by Conservative opponents.
In the 2010 Toronto mayoral campaign, a fake Twitter account created by a member of Rob Ford's campaign team came to light. The account, created with a fictional profile under the @QueensQuayKaren handle, appeared to be a supporter of opponent George Smitherman in an attempt to flush out a potentially damaging audio tape.
Instead of relying on happenstance to discover suspicious Tweeters, Menczer is hoping that his tool will provide a way to automatically weed out questionable accounts that are spamming users or using astroturfing techniques.
He notes, however, that the Indiana University system is still in its early stages and relies heavily on crowdsourcing.
The meme page currently tracks all hashtag, URL and direct mentions that use an election-related term and are tweeted more than five times an hour.
A key part of the website is a "Truthy" button on all meme pages allowing users to report misinformation or suspicious activity. Once users start to sift through and tag memes as "Truthy", those suspicious memes will populate the Truthiness category.
And then perhaps the website will become a useful tool for people trying to sift through the truth and the merely "truthy" in the Twittersphere.
"It's still the early days [of social media], so a lot of people are naïve because it's just a new thing and you are not thinking people are out there to trick you," said Menczer.