Change one rule about what counts, and about a hundred people a year appear or vanish.
The last essay said the gap is a definition. Here is what that looks like as an actual number, and it is cleanest when one project is built on another. Mapping Police Violence starts with the Washington Post's shooting cases and then adds the killings the Post's rule leaves out. So the difference between the two isn't two teams arguing. It is one counting rule, measured in people.
Those 104 people are not a typo and not a judgment call. They are, almost all of them, people the police killed in 2015 some way other than a gunshot: someone who died in custody, someone held down until they stopped breathing, someone hit by a car, or someone shot by an off-duty officer. The Post's rule can't see them. MPV's rule can. And 2015 wasn't a fluke. Pull the numbers straight from each project and MPV comes out ahead of the Post in every single year from 2015 to 2024, by 84 to 146 people, roughly 110 a year. The Guardian, using its own rule, put 2015 higher still, at 1,146.
Run that logic across all five projects and a shape appears. Some deaths land in every database. Some land in only the widest ones, a Taser death shows up in Fatal Encounters and MPV but never in the Post. And some deaths land nowhere, because they happened where no reporter, volunteer, or department was looking. That last group is the real question, and the next two essays go after it directly.