Journalism as a product method

Reporting gives us a way to investigate a proposed product. Who needs help? What do they do now? Where does it fail them? Interviews and original records can reveal a problem that an attractive idea conceals.

A model can help prepare an interview or suggest objections to a plan. It cannot supply evidence that a real person has experienced the problem. We treat those exercises as preparation for research. A prototype gives people something to respond to, and use, once we have a question worth testing.

After publication, the source record becomes part of maintaining the product. A figure in a chart should lead back to a dataset and a definition. An assertion should have a source. When either changes, we need to know which pages and comparisons are affected.

This is one reason we keep research separate from its presentation. The same source may inform a short explanation today and a reference page later. Preserving where a claim came from makes it possible to correct both. It also lets readers distinguish the recorded fact from our interpretation.

Publication introduces another set of questions. Where is the reporting cited or republished? How does an AI answer represent it? Our publisher tools examine these uses, keeping a mention, a republication, and a citation distinct. Each observation says something different about the circulation of a story.

The software deserves scrutiny too. Open source lets someone inspect an implementation, test it in another setting, and point out a flaw. That does not guarantee correctness. It makes the method available to the people who may rely on it.

Journalism is therefore both the subject of some of our products and a discipline for developing them. We investigate the need, retain the evidence, and stay responsible for corrections after release.