As much key information is unknown—for instance, what the time lag would be from extracting oil to an earthquake occurring—the team used a Bayesian machine learning tool that randomly varied ...