What do the trails of data that we leave in the cloud say about us? If you followed them closely, what could you predict about future events, such as riots, pandemics, or consumption habits?
An experiment set to begin in April will use publicly accessible data (search queries, tweets, Wikipedia changes, etc.) to try to discern the laws that guide social phenomena, much in the same way that scientists have studied the laws that dictate physical and chemical phenomenon, according to a report by John Markoff in The New York Times. The effort is funded by Intelligence Advanced Research Projects Activity (Iarpa), a branch of the office of the director of national intelligence. Its goal is to be able to "anticipate and/or detect significant societal events, such as political crises, humanitarian crises, mass violence, riots, mass migrations, disease outbreaks, economic instability, resource shortages, and responses to natural disasters."
The details of the project, called the Open Source Indicators Program, are still vague (IARPA officials would not respond to Markoff's inquiries), but even just the basic outline has raised concerns from privacy advocates and social scientists who don't want their work to be used for military purposes.
At a basic level, these sorts of predictions should be easy to make. If a lot of people search something like "airfare JFK --> SJU February," it's not rocket science to predict that many people will be vacationing in Puerto Rico come winter. But how much more could be known? What could the state do with that knowledge? There are countless harmless applications imaginable: with better information, governments can build more efficient transit systems, spot viruses before they spread, or find clusters of illnesses in specific geographic locations and figure out the causes. But at the core of the concerns of privacy advocates is that when the government has more information on the people (when the actions of the people are more "legible," as Yale political scientist James C. Scott would call it), the government has more power, and it cannot be counted on to use that power wisely. It's transparency in reverse.
But at a theoretical level, if you can get past the practical concerns, the project is fascinating. With more perfect knowledge, do humans become predictable computational devices? Put in one set of inputs, apply rules, receive predictable results. It's a deterministic view of the human mind. As Einstein, an avowed determinist, put it, "Everything is determined, the beginning as well as the end, by forces over which we have no control. It is determined for the insect as well as for the star. Human beings, vegetables, or cosmic dust, we all dance to a mysterious tune, intoned in the distance from an invisible player."
For the time being, any accurate prediction of human minds en masse is a long way off. Economists, political scientists, and psychologists simply do not understand enough about how the mind works and all the variables that shape our decisions. Would having better -- even perfect -- information make such predictions possible? Maybe, but how do you account for a charismatic leader in your formula? How do you estimate the impact of an idea that no one's yet had?
The IARPA program doesn't purport to try to answer those questions. It's only the first step in a line of research that could bring social scientists to that point. Reading one person's mind is the hard work of being a friend; reading the hive mind is now the hard work of being the government.
