The US government is searching for a social media monitoring system that can detect sarcasm. What exactly are they worried about?
“Americans don’t do cynicism, sarcasm or negativity generally,” the journalist Max Davidson wrote, advising Brits how to make it big in the US. “There is an innocence in their outlook that you either love or hate, and if you hate it, you won’t get past first base.”
So news that the US government is looking for a sarcasm detector has already generated cries of: “Yeah right! Good luck with that.”
Yet in a tender notice posted on the US government website, the Secret Service has requested a computer programme with the “ability to detect sarcasm and false positives” along with “sentiment analysis” and “influencer identification”.
This, Ed Donovan, a spokesperson for the Secret Service says, is an attempt “automate our social media monitoring process”. “The ability to detect sarcasm and false positives is just one of 16 or 18 things we are looking at,” he adds.
Millions of messages are exchanged every day, and figuring out how to police these threats is set to be a massive challenge for security services in the years ahead. Davis Lewin, political director of the Henry Jackson Society
Other specifications include understand the audience that is using social media, their location and the ability to access their social media history.
Much of this is a data collection exercise designed to give security officials greater clarity on prioritising messages that could be perceived as credible threats. “Millions of messages are exchanged every day, and figuring out how to police these threats is set to be a massive challenge for security services in the years ahead,” Davis Lewin, political director of the Henry Jackson Society, told Channel 4 News.
Despite what it may seem, detecting sarcasm on the internet is not new. The French media analytics company, Spotter, claims to be able to identify sarcastic comments online as part of a “sentiment analysis package” counting the Home Office and European Commission among its clients.
The program then uses algorithms created from linguistic analysis to detect sarcasm in as many as 25 different languages.
This latest tender, however, marks the first time sarcasm is being sought in an overall national security effort. Security sources tell Channel 4 News that the technology is likely to involve assessing keywords, their contextual and timely relevance as well as where they appear in a conversation and the user’s historical use of language.
Those algorithms are likely to generate reports and notify security agents to potential threats.
Of course, all this will dredge up now-familiar revelations of the National Security’s Agency’s stranglehold of digital data, not least the surveillance of emails, phone calls and Skype conversations. But the latest task involves translating the inference of messages flying across cyberspace each day.
While it may be ridiculed at the moment, Lewin predicts that it will be a vital tool for security services in the years ahead. “The technology is far further advanced than people understand and we are fast-heading toward where our ability to keep the cyberspace secure is running away from us,” he told Channel 4 News.
“The Secret Service has a responsibility to keep people safe, and one of the huge challenges of the future is going to be the ability to assess the threats to the state, many of which will have potential trails on social media.”
The cultural commentator Peter York says there is a vast disconnect between sarcasm in Britain and the concept of American irony. “Historically Britain is a very verbally nuanced culture, which has engrained sarcasm into the DNA of the language,” he told Channel 4 News.
“Whereas in America it is more confined to intellectual classes of Manhattan manifested in comedy such as Stephen Colbert or satirical news like The Onion. The national newscaster network America – that is, the official America – is not predisposed to sarcasm at all. There is still has all sinful, sneaky and old European connotations.”
Whichever technology manufacturer ends up with the contract can expect a challenge. A 2011 study by Rutgers University in New Jersey, found that at best, software programs could only sort the sarcastic from the sincere around 65 per cent of the time.
If security services are hoping that a sarcasm detector might be a potential time-saver, they might find what transpires is precisely the opposite. The irony of that will, of course, be lost on few.