Joining the faculty at Cornell is a homecoming for Robert Kleinberg, in more ways than one.

Kleinberg grew up in Wales Center, N.Y., and earned his bachelor’s in mathematics from Cornell in 1997. It’s no surprise Kleinberg chose mathematics, or Cornell. His father was a mathematics professor at SUNY Buffalo and his mother was also a professor, briefly, at Buffalo State College before going to work in a computing research lab. While neither of them are Cornellians, many in the family are—more than this mathematician can count—starting with his great-uncle Charles Kleinberg and grandfather Samuel Kleinberg, and including his brother, Jon Kleinberg ’93, who is also a computer science professor at Cornell. 

Robert Kleinberg graduated summa cum laude, picking up a Hertz Fellowship along the way. He said even then he wanted to be a professor. At the time his research interests were leading him toward topology, although he did experiment with computer science during his sophomore year. But after a couple of years studying for his Ph.D., Kleinberg left MIT to work for Akamai Technologies, a company whose co-founder, Tom Leighton, later became his adviser. 

You may not know what Akamai does, but chances are you’ve benefited from its work. If you’ve used the Internet for practically anything—to download music or software, check the headlines, book a flight—your clicks pointed you to a server chosen by an algorithm Kleinberg helped invent while working there. 

Several years—and several patents—later, Kleinberg returned to MIT, earning his doctorate in 2005. “It was exhilarating for me in the short term to be making the kinds of contributions that I was making at Akamai,” he says. “But the pursuit of science has always felt to me to be my long term priority in life, and I eventually decided that returning to academia was the right way for me to get there.” 

Kleinberg was soon back at Cornell, although his start was delayed by an NSF Mathematical Sciences Postdoctoral Research Fellowship at the University of California at Berkeley. The work he did there, and continues at Cornell, straddles applied mathematics and theoretical computer science to better understand the Internet. His interests include the economic aspects of algorithms, online learning and its applications, optimization problems in networks, and modeling complex networks. 

Kleinberg has recently been collaborating with colleagues at Berkeley, Carnegie Mellon, Harvard, and Microsoft Research on the question of how to best design an online auction in which people come and go because they have differing time constraints on their purchases. Beyond the obvious application, the question is an important one, says Kleinberg, because it shares two aspects that have become increasingly common in the computer systems we depend on: They operate in and make decisions about an environment that is at least partially unknown to them and, second, they’re controlled by multiple parties with conflicting interests. 

“As a mathematician who studies computer systems, the way I interact with these problems is by abstracting them and trying to prove theorems about them that have an impact beyond the original application,” says Kleinberg, taking pains to point out that “the features that make computational problems realistic are often the same ones which also make them messy. The challenge is to strip these away until you are left with a mathematical problem which is elegant, solvable, and still reflects enough of the reality of the original system that motivated the question.”