Milo B. Beckman is really, really good at dabbling. In the half hour we spend talking, he mentions roughly half a dozen pursuits he’s spent his time on in the past few years. Like the phase he spent creating crossword puzzles for The New York Times and then spending several days working for Will Shortz personally, or the one in high school during which he created over 100 quizzes for Sporcle, spreading them out over six accounts, or that time he collaborated with Nate Silver—“I have his number if you want to hit him up”—for FiveThirtyEight.

“My problem is that I won’t know what I’ll be passionate about and then I’ll find it and go ham with it,” he admits.

Beckman, a New Yorker whose only constant commitment at Harvard has been to a youth civic education program at the IOP, speaks simply and cheerfully about himself and his unusual array of quirks and talents.

“I just find something and am like, ‘Well that’s cool. I’m going to spend my entire life on that for a little while,’” Beckman says.

Here’s another thing Beckman has done: find the secret to SAT success.

When he was a senior in high school, Beckman drew on his own test-taking experience while considering a project for his AP Statistics class. Curious whether graders paid attention to the content of SAT essays or simply graded with a quick cursory glance, he crowd-sourced roughly 115 essay lengths and corresponding scores from his friends.

His results were tough to argue with—a certain statistical correlation that longer essays lead to higher scores—and they landed him on breakfast television before he even applied to college.

“My teacher encouraged me to send it to this dude who knew a guy who knew a guy, and then Good Morning America happened, which was very silly,” he explains.

Beckman’s project got the attention of Les Perelman, then a director of writing at MIT, who had a similar hypothesis based upon previous research using sample essays provided by the College Board.

The two have worked together for the past four years, researching factors and methods in test grading.
Recently, their efforts have focused on robo-graders, demonstrating the disadvantage they pose versus actual human graders as machines prefer word length over actual content. “Let’s fight robots with robots,” Beckman jokes.

In response to automated graders, Perelman, Beckman, and the rest of the team created the Babel Generator. The software responds to a prompt and spits out a gibberish essay with enough flowery language to garner a six out of six from a robo-grader.

Beckman’s next project on essays will look at the role that font selection by the writer plays in grade outcome, after noticing that papers he turned in using Georgia, rather than Times New Roman, achieved better grades.

Still, the 18-year old, who effectively skipped middle school and began his freshman year at age 15, isn’t at all sure where all this will take him.

“I honestly don’t really know what I’m going to be doing half an hour from now,” Beckman says. “I feel like I’m going to be alive for a really long time and I’m interested in a lot of different things and I can do a bunch of different things for a short period of time.”