“Show your work” makes a triumphant return

Next year will herald the return of open source culture in newsrooms.

But first, some history: In the spring of 2008, I was working at ProPublica. It was a month or two before we launched. A colleague and I had lunch with Aron Pilhofer, head of a new interactive technology team at The New York Times. His generosity set us on a course that would become a big part of our newsroom’s early success.

The day we had lunch was the day that Hillary Clinton, then running to be the presidential nominee of the Democratic Party, had released her travel records as first lady in the 1990s. Most newsrooms wrote stories and let readers download a PDF of the release. But the team at the Times, which was a new hybrid of journalists and software engineers, published a beautiful interactive web interface that let readers read the documents right in their browsers.

Knowing documentary evidence would be central to our work at ProPublica, and feeling pretty bold, we asked Aron if the Times would give us the code behind it.

To our surprise, Aron not only gave us the code, but he and his team gave us a copy of the Amazon cloud server blueprint required to publish it.

It was the opening chapter in the story of Aron and me founding DocumentCloud with some of our colleagues. But it was a characteristic, even mundane, example of a bygone era, when newsroom technical teams happily shared their work with no strings attached.

For most of the 2010s, newsroom nerd teams around the country and world learned from each other by sharing code, teaching at conferences, and answering questions posed on NICAR-L, the email list that was required reading for data journalists around the world.

Those days, unfortunately, are gone.

Earlier this year, my friend Ben Welsh and I did an analysis of 300 newsrooms with open source repositories on GitHub. Our research found that there’s been an 85 percent drop in new repository creation, from a peak of 2,097 new repositories in 2016 to only 320 last year. The finding was robust, with both large and small newsrooms opening up less code, and contributions to existing repos also going down. And it’s not just code that’s declining. Posts to NICAR-L are also dramatically lower compared to ten years ago — down almost 90%.

As part of our research, we also interviewed more than a dozen of the most active newsroom technologists in the 2010s to investigate what happened. Our full findings, presented earlier this year at the NPA Summit in Chicago, are coming out in a post on OpenNews’ Source blog in the coming days.

But I predict a resurgence is coming. In 2026, newsrooms will reverse this decade-long decline and return to prioritizing open source collaboration, not as a luxury but as an existential imperative, incubated by financial support and driven by the changes AI will bring to how we do our work.

We’re already seeing promising signs of progress. The American Journalism Project’s Product & AI Studio has been helping its AI project grantees post their work on GitHub. Spotlight PA, founded in 2019, posts seemingly all of its code on GitHub. Newspack, the team I work on at Automattic, releases virtually all the code behind a publishing platform that powers more than 325 newsrooms around the world, under an open source license.

The primary driver for the change will be the recognition that transparency and trust are inextricably linked. With trust in our industry at historic lows, the “show your work” ethos will again demonstrate that journalism is open, verifiable, and reproducible. News organizations will realize that sharing their methods — like publishing detailed methodologies, developing projects in the open, and yes, giving away stuff for free — is essential to building trust with their communities.

Second, philanthropic pressure will force change. Institutional funders often tolerate grantees keeping software projects they build private, on the theory that the work might someday generate follow-on revenue. In practice, that path is rarely pursued in earnest and even more rarely realized. Funders will recommit to open-source requirements for code produced with grant dollars, so it benefits the whole ecosystem.

Finally, news organizations will recognize sharing as collective self-preservation. Generative AI is the biggest change in the last 25 years in how humanity consumes and produces information, and none of us is big enough to solve all the challenges it introduces by ourselves.

And with vibe coding making it possible for more English majors like me to do software development, there will be a huge new generation of coder journalists excited to show their work.