There’s a cynical saying in the news business that my generation of reporters used to snarl under our breath when we got some incomprehensibly stupid assignment: “News is what happens to an editor on his or her way back from lunch.”

The suggestion that the leaders of news organizations deploy their staff to explore the things that irritate or fascinate them is one of the many reasons ProPublica has made a practice of relying on reporters to generate story ideas. But every now and again, even editors stumble across news.

This is one such story.

Last Halloween, Sara Pekow, wife of ProPublica’s Deputy Managing Editor Eric Umansky, and their 6-year-old daughter were headed home after a night of trick-or-treating when they saw an unmarked police car with lights flashing headed the wrong way up a one-way street in pursuit of several teenage boys. The car struck one of the kids, who rolled across the hood. Miraculously, he was not seriously injured.

When Pekow got home, she told Umansky to head back out to find out what else was happening. He did and saw police haul away three boys, a 15-year-old, a 14-year-old and a 12-year-old. All were Black.

Over the next four hours, Umansky waited outside the precinct house with the kids’ parents, none of whom were allowed to see their children. They were released without explanation at 12:45 a.m.

No one was seriously injured, sent to jail for decades or otherwise harmed, the types of events we tend to look for when pursuing investigations. But Umansky assigned himself to figure out a seemingly simple question: Would anyone on the police force be held accountable for what appeared to be a dangerous, excessively forceful reaction to what was later described as a possible cellphone theft?

He started with official channels, asking a New York Police Department spokesman for the basic facts of the case. The official said the case involved an “unknown male” who had “fled the scene” by running across “the hood of a stationary police car.” Pekow and the three additional witnesses Umansky found who saw the car hit the kid were all mistaken, the spokesman insisted.

When Umansky tweeted about what he had learned, the agency that investigates civilian complaints against New York police, the Civilian Complaint Review Board, responded that it was looking into the case.

On May 25, the Minneapolis police killed George Floyd, and questions of police accountability were suddenly at the center of a historic racial reckoning. Umansky returned to his research on the civilian board and was startled to learn how seldom complaints about use of force resulted in serious discipline. In late June, he published an article about his family’s experience and what he had learned about NYPD accountability.

Among his findings: In 2018, the last year for which a full picture is available, the CCRB looked into nearly 3,000 allegations involving allegations of violence; only 73 were substantiated. The most severe punishment, loss of vacation days, was meted out to nine officers.

No details were available on any of these cases due to 50-a, a New York law that barred the public from seeing police discipline records. But that was about to change. About a week before we published Umansky’s story, New York lawmakers repealed that law, making it possible to ask for records on the Halloween case and a whole lot more.

Soon after, Umansky filed a request for those records. Rather than ask for documents on every complaint ever filed, he asked the board for the records of every police officer who had had at least one substantiated complaint. The files would include records of that case and a barebones summary of all complaints against those officers. It also meant the request could be processed relatively quickly.

Days later, we had the data. As we worked through it, the records also became engulfed in a legal battle. Police unions sued New York City to keep many of the records secret. Last week, a federal judge blocked not only the city from disclosing the records but also

the New York Civil Liberties Union, which had the data too. The police unions did not know that we also had many of the records.

Normally, preparing such data for publication in a searchable form requires weeks of code writing and checking. In this instance, we wanted to move quickly while still taking enough time to weigh some difficult ethical questions.

One immediate issue presented by the data was whether we would include in our database the cases CCRB investigators had concluded were “unfounded,” meaning the evidence suggests the allegation never happened. That brought a straightforward answer: No.

The unsubstantiated allegations in the data posed a more difficult question. Investigators had neither proved nor disproved them. In the end, we decided to include a terse description of those cases on the theory that their number and nature might help readers or other journalists see a pattern. We understood that some of these accusations may be exaggerated or even made up. But we agreed with the legislators’ view in repealing 50-a that the enormous power police officers wield can make them subject to more scrutiny than ordinary citizens.

On Saturday afternoon, we learned that the city of New York had mentioned in court papers that ProPublica had some of the data. In the hours that followed, a team of computer developers began scrambling to create an online database that could be searched by readers.

As it happened, both Ken Schwencke, the editor of our news apps team, and Derek Willis, one of our news applications developers, were at the beach. They quickly gathered up their towels and raced back to their computers.

Thankfully, we had been planning for just this possibility. Derek and Moiz Syed, another news app developer, had already written some of the code. Umansky had drafted a brief story that would accompany the database and Henri Cauvin, a senior editor and experienced cop reporter, pored over the copy as midnight approached.

We knew this work would come under intense scrutiny — from the police unions that fought to keep the information secret to advocates on all sides to our fellow journalists. Data editor Ryann Jones and Celeste LeCompte, ProPublica’s vice president for strategy and operations who oversees our data sales, did spot checks to search for errors.

We hit the publish button at 11:07 a.m., less than 24 hours after we had learned that our possession of the data was public knowledge.

The reckoning that is unfolding in 2020 is without precedent (I thought the repeal of 50-a wouldn’t occur in my lifetime). What we’ve learned at ProPublica about police and prosecutorial misconduct has mostly come through the digging of veteran police reporters like A.C. Thompson, who exposed the widespread misconduct of New Orleans police after Hurricane Katrina, or Topher Sanders, who broke the story on unjustified arrests for jaywalking in Jacksonville, Florida, in our “Walking While Black” series. The subject is so large, and important to where our society is going, that even editors can have some good ideas.