The University of Tennessee's 1,024-processor Nautilus supercomputer predicted the Egyptian revolution and bin Laden's location
In the Foundation series of novels by American science fiction writer Isaac Asimov, the behavior of masses of people could be predicted with the help of “psychohistory” - a method of forecasting political and social trends using a device called the Primary Radiant. In the 1950s, there was no computing power that could make such a device a reality. It is now possible.
Supercomputers such as Nautilus, located at the University of Tennessee's Center for Remote Analysis and Data Visualization, could bring the world closer to Asimov's universe, although this is only an early stage. The key is to find patterns in a huge amount of data and be able to visualize them. This is what Kalev Litaru is doing at the Center for Digital and Text Analytics at the University of Illinois.
Litaru uses a database of 100 million news articles covering the period from 1979 to early 2011. Analyzing the text and tone of the news, be it negative or positive, Lytaru discovered the emergence of patterns that seem to line up, indicating major periods of unrest in the future. For example, in Egypt, the tone of news articles about Mubarak became increasingly negative as public discontent grew, until Mubarak eventually resigned.
Though it's not just the tone of the articles; the change in tone of articles over time also matters. According to Lytaru's calculations, the Saudi government remained in power because the tone of news there in the past was equally negative, while the tone of articles in Egypt and Tunisia reached new lows. Litaru notes that many experts on Egypt believed that Mubarak would survive the uprising safely, as has happened in the past.
This is possible because a supercomputer can look for patterns in networks with 100 trillion connections and 10 billion nodes. An ordinary computer, says Lytaru, can only grasp a small amount of data at a time, and even running monitoring in parallel in the background will not be able to solve the problem. This happens because when analyzing networks, the required amount of memory grows exponentially in accordance with the number of connections. Only a supercomputer can handle this, and it took time (140,000 hours per processor, or about a week to work through at a time) for the machine to be able to draw such conclusions based on these data.
Technology is not yet capable of predicting events. Litaru compares it to early weather forecasts - in some ways nothing more than guesswork, but now the weather forecast technology is reliable enough to make decisions based on it. Technology cannot predict the actions of individuals, but it can predict how people might react, such as the self-immolation of Mohamed Bouazizi in Tunisia (a street vendor who set himself on fire because he could not feed his family and the police confiscated fruit and vegetables he was trying to sell: approx. mixednews).
Obviously, this technology is of great interest to the security forces and the military. When asked if they had contacted him, Litaru replied, "I cannot talk about it."
Promotional video: