Suspecting with Data

By MARY D. FAN. Full Text.

Our pooled consumer big data, such as the pictures we post or the location history and keyword search trails we leave, are generating new ways to solve crimes. Much of the commentary on big data search strategies such as keyword, geofence, and facial recognition searches fixate on Fourth Amendment search and seizure issues rather than evidentiary safeguards. This Article breaks new ground by framing evidentiary guardrails for big data searches to reduce the harms of erroneous arrests, redress secrecy, and counteract the mystique of machine infallibility.

Advancing beyond over-reliance on Fourth Amendment doctrine, this Article illuminates how evidence law and procedures are better suited to address concerns over inaccuracy, overbreadth, and opacity surrounding big data search strategies. The Article offers three proposals. First is requiring corroboration before big data search strategies can be the basis to arrest or convict a person, thus updating the concept of probable cause for changing technologies of proof. Second are pretrial notice, disclosure, and reliability hearings to pierce the secrecy surrounding the use of big data search strategies and to permit effective defense challenges. Third is deploying expert witnesses on the reliability concerns surrounding the evidentiary fruits of big data analytical techniques to correct the mystique of machine infallibility and the risk of factfinders overweighing match evidence.