Could Computers Predict Political Unrest Like They Predict The Weather?

from the doubtful dept

Via betajames, we learn of some research being done on a computer system that's trying to predict political unrest by looking at press coverage.
In a recently published paper, Culturomics 2.0: Forecasting Large-scale Human Behavior Using Global News Media Tone in Time and Space, [Kalev Leetaru] shows that by feeding millions of news articles from around the world into an SGI supercomputer, you can analyze the tone of media coverage and pinpoint moments of unrest: a revolution, riots, or even when a despotic ruler will relinquish power and flee his country.
Of course, so far, the system has only been used on past events, and as it says on the box: past results are no guarantee on future performance. Separately, looking at the examples in the article, it sounds like it notices stuff... well... about the same time a human paying attention would notice stuff. It talks about how right before Pearl Harbor the Japanese press dropped all talk of peace and ramped up criticism of the US. It also discusses Mubarak leaving office and how the press turned really negative right before that. But it wasn't like you needed a computer to tell you that the protests were having an impact and likely leading to the end of Mubarak's reign.

The project sounds interesting, but it seems like Leetaru is overly optimistic about what it'll show. Until there's evidence of it picking out things that you couldn't pick out yourself just by reading the news, I'm not sure there's really any breakthrough.

Reader Comments

Subscribe: RSS

View by: Time | Thread


  1. identicon
    Anonymous Coward, 30 Sep 2011 @ 7:34am

    I have have no confidence that this effort will ever reliably forecast anything. An obvious problem being how to effectively accounting for media bias.
    For example, the mainstream media here in the US have largely ignored the Occupy Wall Street demonstrations in which thousands of people have been massed in New York's financial district in opposition to Wall Street's corruption for the last week and a half. Thousands of people a week and a half, and still the US corporate owned media has very little if any coverage. (None print or video for the first week.) Now consider the coverage given to Tea Party protesters who's messages are in alignment with corporatist goals (if not, in fact, authored by them).
    So, if a significant social trend/narrative is not reported (or is dramatically under reported) and its converse enjoys heavy coverage, how do you program to correct for the biases in the data sample without introducing yet other biases?
    The cardinal rule of computing: GIGO (garbage in garbage out).

Add Your Comment

Have a Techdirt Account? Sign in now. Want one? Register here
Get Techdirt’s Daily Email
Use markdown for basic formatting. HTML is no longer supported.
  Save me a cookie
Follow Techdirt
Techdirt Gear
Shop Now: I Invented Email
Advertisement
Report this ad  |  Hide Techdirt ads
Essential Reading
Techdirt Deals
Report this ad  |  Hide Techdirt ads
Techdirt Insider Chat
Advertisement
Report this ad  |  Hide Techdirt ads
Recent Stories
Advertisement
Report this ad  |  Hide Techdirt ads

Close

Email This

This feature is only available to registered users. Register or sign in to use it.