The history of technology-assisted review (TAR 1.0)—that is, the first generation of predictive coding or machine learning in e-discovery—is laden with tedious processes, expensive per gig payments and frustrated users.