The Bright Side of Telephone Metadata

Since the exposure of the NSA's data collection tactics, telephonic metadata has become an unlikely star in the popular press. The Verizon order leaked by Edward Snowden has prompted a critical public education about how much can be learned from the information routinely recorded by our phone companies. But the same data currently playing a sinister role in national security surveillance is also fueling a wave of new research that will help international aid organizations save money and lives.

Global Pulse, a United Nations initiative, recently released a primer showing how cell phone metadata about the placement of phone calls and the location of the cell phone user can be mined during periods of crisis. A leading example comes from Haiti. When the devastating earthquake hit Port-au-Prince in 2010, cell tower geo-location data created a more accurate record of the destinations of evacuating city residents than the estimates of the Haitian government. Moreover, the destinations of the fleeing residents could have been predicted in advance: when disaster strikes, people go where they have social networks, and their social bonds are revealed through calling patterns that preceded the disaster. Thus, with some rudimentary network analysis, aid organizations can figure out where to send help before the emergency migrations are complete. Metadata can get the right amount of aid to the right places, faster.

Similar studies linking calling network data to migration patterns have been used to track the spread of Malaria in Kenya , and to validate Mexico's containment efforts during the H1N1 flu epidemic. So far, the research has been retrospective, but in the future cell phone data could be used to control the spread of disease in real time, with as few restrictions on mobility as possible.

Cell phone metadata also can be used as an indirect measure of a community's economic health. A 2010 study of UK cell phone data confirmed that areas placing and receiving calls from more diverse geographic locations are more prosperous, and have more economic opportunities. Knowing this, humanitarian groups can use aggregate calling patterns to monitor an area's financial security. They can also gauge whether development efforts in a particular region have worked.

These promising programs are a helpful reminder that mass data collection can be used in socially desirable ways, and that the NSA is not the only institution with an interest in this data. Of course, Global Pulse's data mining differs from the NSA's in one important respect: traceability. Cell phone metadata prepared for research purposes has been de-identified, and researchers are usually obliged to refrain from re-identifying any of the subjects. In contrast, the very goal of criminal and national security investigations is to identify, and possibly seize, the subjects of investigation. But the larger implication of data-driven humanitarian aid is that data collection is not inherently bad.

Global Pulse's projects are antidotes to the tired parade of New York Times articles generically criticizing modern data collection practices. Productive discussions of data policy should not attempt to reverse the ever-increasing reserves of data that accumulate in modern life. Instead, since data continually surprises us with new uses and insights, information policy should align incentives to increase socially valuable uses of data (like the ones described here) while decreasing uses that lead to adverse consequences.