New DARPA Proposal Taps Social Media to Combat Future CrimesJuly 30, 2012 4:43 pm ·
If you remember the movie “Minority Report,” you remember how fantastical some elements of its plot were when it came out back in 2002. However, merely a decade since its release, this sci-fi thriller has turned out to be a lot more sci than fi. From vendors that recognize consumers and make personalized purchase suggestions to the philosophical shift in law enforcement from responding to crime to preempting it, many of the themes that at that time seemed so far off have become a reality.
In the latest installment of this rapid transition from fantasy to reality, the Defense Advanced Research Projects Agency, or DARPA, has revealed its latest project. Entitled “Forecasting Dynamic Group Behavior in Social Media,” this DARPA initiative, which is still in its initial stages, is meant to provide the U.S. military with an automated tool that sifts through social media chatter to predict terrorism.
In a release published in search of contractors, DARPA notes the way that social media “have evolved from a platform that provides infrastructure that supports maintaining connections between friends to a platform that supports recruiting, collaborating, organizing, and competing for resources.” As a result of this development, the agency has noticed the use of social media to create “virtual teams,” which themselves are allegedly to be susceptible to “evolve over time.”
“Among these communities and teams are terrorist and other criminal organizations,” DARPA argues. Therefore, to combat the potential misuse of these outlets, the agency introduced the following:
This topic seeks innovative research to develop automated tools that can (1) learn models of the dynamics of inter- as well as intra-group interaction in social media and (2) track the evolution of such dynamics and derive casual factors from online interaction data. The algorithms must be able to operate on large datasets of millions of nodes, generate robust and reliable group behavior and interaction models, and provide the users with factors and their relative contribution to changes in online behaviors. This technology will be used by analysts in forecasting online behaviors and identifying competition and possible cyber terrorism events.
Much like the “Pre-Cogs” in the futuristic Spielberg film—the humans gifted with the ability to foresee criminal acts—DARPA is clearly interested in crafting a more realistic alternative means to the same ends. To that point, the agency also published a detailed outline of the project’s phases:
* Task 1: Design and improve the feasibility of a system that can track groups and their state changes in social media.
* Task 2: Research key indicators of group interactions, including competition, recruitment activities, and effects of events and topics on group structure changes.
* Task 1: Design and develop a system that learns dynamics of group behavior and inter- and intra-group interactions in an unsupervised manner based upon design and innovation developed in Phase I.
* Task 2: Demonstrate the system on a social media dataset containing >1K groups, >100K postings/day, and >1M members. Achieve high accuracy (90%) of detecting group state changes, activities, conflicts, and competitions.
PHASE III DUAL USE APPLICATIONS: Successful development of the prototype capability would be of great interest to industrial espionage prevention specialists, law enforcement, market analysts, and polling organizations. This capability would be applicable to a broad range of tactical as well as strategic military operations.
Granted, the DARPA project is still in its proposal stages; however, its potential impact is difficult to ignore as it awaits the needed approval and resourcing. In addition to the clear due process and privacy implications of using such a system to punish “criminals” that are yet to commit the crime for which they’re being punished, perhaps an even greater cause for concern lies in its non-criminal, non-terrorism use.
As I bolded above, the potential uses of this program extend far beyond the defense or even law enforcement purposes. The inclusion of “industrial espionage prevention specialists…market analysts, and polling organizations” reveals that commercial interests also stand to make considerable gains from this project— a project which is not only being proposed under the pretext of being a national security matter, but, if approved, would also be publicly funded.
As Zack Walton of WebProNews accurately observes, “Companies like Google already track certain behavior and tailor ads towards you. Marketers would pay millions for technology that could detect even the slightest change of behavior in your social media or other online activity.” Thinking better of the proposal, however, Walton concludes, “All-encompassing knowledge of your behavior in social media should be between you and your friends—not a lifeless computer that converts human interaction into a spreadsheet.”
Perhaps most disconcerting, however, is the DARPA document’s mention of potentially using the system for “industrial espionage prevention.” Although protecting valid trade secrets (and other intellectual property) is a legitimate concern among industry leaders, this tool could also prove problematic for industry whistleblowers attempting to hold companies to minimal legal and ethical standards.
Given that employees are legally protected from employer retaliation when they attempt to ensure industrial accountability, the consequences of allowing commercial interests to utilize this technology could potentially mean the further undermining of this already embattled, yet vitally important protection—making the potential problem of increasingly invasive advertising seem more like a minor inconvenience.