You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

Why I Am Leaving Goldman Sachs...To Build A Ping-Pong Playing Robot

This morning, the Internet was abuzz over a scathing resignation letter—in the form of a New York Times op-ed—from a Goldman Sachs official named Greg Smith. Smith claims that over the last several years, the moral culture at the firm has soured. Today, Smith says, Goldman Sachs is all about “shortcuts,” but his proudest moments “have all come through hard work, with no shortcuts.” And he’s got quite a list of proud moments: A full scholarship to Stanford, selection as a national finalist for the Rhodes, and “winning a bronze medal for table tennis at the Maccabiah Games in Israel, known as the Jewish Olympics.” The letter raises one of the great questions of our time: Is it possible to take a shortcut to a ping-pong championship bronze medal?

Well, not for robots, according to recent research. In fact, table tennis is a task that has long fascinated (and bedeviled) experts in robotics. That’s because, as two experts noted in a 2009 paper, the game “has all basic components of a complex task: it requires accurate control, it is based on several elemental movements or motor primitives (e.g., different forehands and backhands), the goal functions based on perceptual context determine the behavior and even higher level strategies using opponent models can play a role.” A successful ping-pong robot has to await its opponent’s shot; position itself to return the volley; hit the ball; and reposition itself for the return shot. That’s a rather complex series of tasks, and in tests against “a robot ping pong launcher [firing] at random times and to randomly chosen positions,” the authors’ robot could only return 73 percent of the balls served. That’s nothing to sneeze at, but it’s unlikely to win you a bronze medal.