When I give a talk about something technical, I explain the same concept three different times. The first time is the technical explanation, with all the code and math, for the people in the audience who operate on a higher level of expertise and can understand it. I’ll tell them about this program that looks at clustered data online to see what generated the most sustained attention across a certain period of time. Then I’ll explain it a second time, walking people through what that means. Of all the cat pictures viewed online last year, this program can tell you which one people looked at the most. How did we do that? We ran all cat pictures posted last year through the program, then manually weeded out everything that wasn’t actually cat—say, a picture of Catwoman the algorithm had accidentally picked up—and then checked to see which one was the most popular. And then the third time I explain the concept, I give a succinct, easily tweetable sound bite. I’ll show a picture of a kitten and say, “This is the cutest kitten of 2012.” There’s another trick in there, too: When you show people a picture of a kitten, they’ll get happy, and then they’ll think they liked your talk. —As told to Claire Suddath
• Mason is chief scientist at Bitly.