Military meme tracker - Cairo uprising

The revolt that started a year ago today in Egypt was spread by Twitter and YouTube, or so the popular conception goes. But a group of Navy-backed researchers has a more controversial thesis: Egyptians were infected by the idea of overthrowing their dictator.

And now, these researchers claim, they’re getting close to developing tools that can track the spread of infections like these.

With funding from the Office of Naval Research, a team at Aptima, Inc. is developing software that’d do more than just scan Twitter for trending topics. Instead, it’d mine the web, including news stories, social networks and blogs, to extract topics and phrases that are gaining traction online. Then, the software would keep tabs on how the conversations proliferate, both geographically and over time.

The software would use epidemiological modeling to chart the discussions and their trajectory. It’s a strategy often used in public health initiatives to figure out where an illness started, and how it spread: Epidemiologists use collections of data to make educated guesses about causality; which health and environmental factors, for example, contributed to an outbreak in a given community.

Applied to online discourse, epidemiological models would essentially treat uprisings like illnesses. They’d pull apart a web conversation (the author of the post, the site where it was published, the comments that ensued) and try to figure out which parts contributed most readily to the spread of a revolutionary message. That’s a different approach to prediction than the Pentagon’s current initiatives, like the Integrated Crisis Early Warning System, which combines news reports with expert intel and big social and demographic forces to come up with a forecast.

The software’s overarching goal? Help the Pentagon determine how “the flow of ideas or ‘memes’ through electronic media can … infect and influence susceptible populations.”

Called “Epidemiological Modeling of the Evolution of Messages” — shortened to E-Meme — the program would use language recognition technology to determine what people in certain regions, of certain age groups, genders, or any number of other demographics, are discussing. From “next week’s election” to “link up and cause havoc.”

“We witnessed the profound power of ideas to replicate in what began as anti-government sentiment in Tunisia, then moved like a virus, reaching and influencing new groups in Egypt, Syria, and Libya,” McCormack said. “If we can better understand the flow of ideas through electronic channels to sway the perceptions of groups, we may be better prepared to develop appropriate strategies, such as supporting democratic movements or perhaps dissuading suicide bombers.”

And E-Meme, currently one year into a two-year development plan, will even be designed to go beyond those abilities. After tracking the proliferation of topics, E-Meme will analyze what kinds of attitudes those discussing them seem to have. And then how those attitudes influence the conversation’s spread across the web.

“A lot of tools exist to do things like look at trending topics, or how many people online are talking about X,” Dr. Robert McCormack, the project’s lead investigator, tells Danger Room. “We want to take that several steps further. We’re interested in the dynamics of those conversations.”

If it works as advertised, software like E-Meme would massively improve the predictive capabilities of federal officials, especially where long-brewing discontent is concerned.

And there’s no question U.S. officials need all the predictive help they can get. Egypt’s tech-fueled rebellion came as a surprise to American leaders, despite a massive Pentagon investment — $125 million in the past three years alone — on computer systems meant to spot signs of political unrest worldwide. Alas, despite heaps of research, even the very best American forecasting tools remain “bad,” one program designer told Danger Room last year, “some less bad than others.”

It’s unclear whether E-Meme will succeed where other programs have floundered. But McCormack and co. certainly have high hopes for the software’s abilities. Eventually, they’d like to analyze “sentiments” and “the perceptions of groups” to determine “what the online discussion will actually turn into.” Egypt’s relative peace, for example, compared to Libya’s rampant violence. Of course, McCormack notes, that kind of analysis remains “incredibly difficult to do.”

Such all-knowing, infallible web mining might be a ways off, but the evolution of software like E-Meme will no doubt remain a major Pentagon priority. In the past two years alone, we’ve seen the CIA invest in a company that scours the Internet to “predict the future,”  Iarpa consider the merits of person-finding via web pic and spotting rebel citizens via YouTube.

Of course, governments worldwide already do plenty of online monitoring. For those living in repressive, dictatorial regions, the prospect of monitoring that’s smart enough to understand the online spread of their ideas, and the subsequent real-world outcomes, could be downright devastating.

Here’s hoping that tech like E-Meme isn’t used to stop legitimate uprisings. If nothing else, maybe it’d prevent U.S. officials, at least, from seeming kinda clueless when the uprisings actually take place.