"By the time Shell tasked Hubbert with explaining how fracking worked, “we had the records of several thousand fracturing jobs, with varying degrees of reliability in their data,” he recalled. “We had to smoke out useful information.”"