Spin detector calculates deception in politics
But experts wonder if math can replace the media
Tessa Vanderhart Staff
Image courtesy of David Skillicorn.
A new computer program that claims to quantify deception could be used to identify spin, according to researchers at Queen’s University.
David Skillicorn, a professor of computer science at Queen’s and a graduate of the U of M, developed the algorithm based on deception detectors from research conducted at the University of Texas. It was first used to analyze email correspondence between Enron employees, but has since deconstructed spin in Canadian politics.
The Queen’s research team is still trying to determine exactly what the system predicts, but Skillicorn likened it to spin — itself a term that is difficult to define.
He noted that while there are innumerable theories of how to identify spin, when tested they are “almost invariably just rubbish.”
The program allows “markers,” words that are often tied to deception, to become easily identifiable, although the markers are not always the most obvious words: “lies” ranks high but is not as clear an indication of spin as “going.” The use of exception words, such as “however,” and negative emotion and action words are also implicated in deceptive language.
Unusual syntax is more difficult to notice, but the system can pick up on meaningful correlations in speech, like the decreased use of first-person pronouns — something Skillicorn suggested might be a way to put distance between the speaker and the speech.
Campaign speeches, the most recent focus of the system, are unusual: prepared beforehand but still performed, they were an ideal test of the system.
“People are conditioned to expect politicians to be perfect. Politicians spin for a living; they don’t want to offend anybody,” said Skillicorn. “Speeches aren’t just off-the-cuff remarks; these are the words that speech-writers choose to represent politicians.”
Although Skillicorn acknowledged that the model is still crude, the research team noted a difference between speeches performed by the political parties: the Liberals had the most “spin” with a score of 124, something Skillicorn linked to being in the incumbent position. The NDP fared slightly better at 88, and the Conservative speeches, at 73, showed the least spin, though the usefulness of these comparisons is still unclear.
Still, Skillicorn said that the spin detector will have relevance.
“We’ll be able to label politicians who spin more overtly,” he said.
Mark Schneider, a professor at UBC’s School of Journalism, said that political spin is only to be expected — instead, he said, we should focus on how the media cover this spin.
Schneider said that trying to define spin with math or syntax is silly, as it is something that most people can see for themselves. He compared it to the way great jazz musicians feel about jazz: either you get it or you don’t.
“We all have an intuitive understanding of what it is, but if you try to define it, you can’t,” he said.
He added that journalists are given the task of noting spin and dealing with it appropriately.
“It’s essentially a very high level journalistic task to identify what the spin is and interpret it,” Schneider said.
Sometimes, he noted, journalists fail in this task — and as a result, people should perhaps be more concerned about the potential for bias and spin in the media than in political campaigns.
William Stairs, director of communications for Prime Minister Harper, said that he is not sure that an algorithm could, or should, attempt to take over from journalists, though the idea is nothing new.
Stairs also questioned what spin is. “Spin is not lying; spin is interpreting . . . to your advantage.”
He noted that every political campaign is premised on spin: it is the most basic and most sensible way of making a point, of campaigning.
“Is that spin?” he asked. “I’m not sure that people who spin for a living would see it that way.”
Stairs said that the idea is nothing new, and that he is not concerned about the impact the spin detector could have.
“I don’t think we’re going to have TV debates with a spin detector in the corner,” he said.
