EVERY SILICON VALLEY company wants more data. But today, tech firms are increasingly taking a paradoxical approach to filling that endless appetite. Thanks to an emerging branch of data science called "differential privacy," they can analyze mountains of user info without breaching the privacy of any individual user. And of all the companies eager to use that science to rehabilitate their reputation for controversial privacy violations, perhaps none has more at stake than Uber.

On Thursday, the embattled ride-sharing startup announced a new and well-timed advance in that privacy engineering field, releasing an open-source tool designed to give the ride-sharing firm—and any other company that adopts its technique—a new method of letting engineers gather statistical results from massive datasets while still remaining blindfolded to the personal details of any single user.