Joe Doyle was still a grad student at the University of Chicago in the late 1990s when he went to watch the proceedings in Cook County's juvenile court.
He sat there while inexperienced lawyers argued over the fate of young offenders, mostly young black men. He witnessed judges who had to instruct those inexperienced lawyers on procedure at the same time that they, the judges, had to render life-altering decisions.
To incarcerate or not to incarcerate?
Sitting in the busy court, watching judges face that question, Doyle wondered what would happen to those kids in 10 years.
Would being locked up hurt or help?
Now an economics professor at the MIT Sloan School of Management, Doyle recently released research that suggests a partial answer to the question.
His paper has an eye-boggling, academic title: "Juvenile Incarceration, Human Capital, and Future Crime: Evidence From Randomly-Assigned Judges."
When I called him Tuesday, I asked him to put his findings about offenders into layperson's terms.
"If you're sentenced to juvenile incarceration," he said, "you're twice as likely to go into adult prison by 25."
The average person might think, well, yeah, what's the news there?
Bad dude at 15, worse dude at 25. Pass the sports section.
Doyle's research suggests it's not that simple, that juvenile incarceration can actually increase crime, a belief that has many proponents but which has been hard to quantify.
Juvenile incarceration. Juvy. The school-to-prison pipeline.
Whatever terms you use for the problem, the United States locks up young people at a far higher rate than any other country in the world.
And whatever terms you use, Doyle's research puts some numbers to the claim that that is counterproductive.
Doyle and his co-author, Anna Aizer of Brown University, collected data from 35,000 adolescents who committed crimes in Chicago in the 1990s. They studied which offenders were incarcerated and which weren't. They then compared which ones wound up in prison as adults.
They did their best to compare teenagers who were legitimately comparable, those of the same age, same race, same crime and same neighborhood.
"Apples to apples," Doyle said.
Their complex study — filled with terms like "variables," "regression" and "confounding factors" — comes down to a few basic conclusions.
One conclusion: The judge makes a difference.