Counting the Killings · No. 05

The Open Data

A series about counting owes you a count you can open yourself. Here is what is in it, and what is not.

The colophon·data release v0.2.0

What you can download

Every figure in this series comes from one small, public bundle of files, free for anyone to use. Open it and you get six plain tables, as spreadsheets or as data: the five counters and their rules, how many people each one recorded per year, the FBI's coverage numbers, and, new in this version, a table that matches the databases up against each other case by case.

None of it lives in a private database you have to trust me about. It is all written into the site's own code in the open, and updating it means editing a file and dating the change, so the history of every number is right there in public.

What it leaves out, on purpose

This release only publishes totals and summaries. It does not republish the individual records: no names, no ages, no home addresses of the people who died. Those belong to the original projects, under their own rules, and I link to them instead of copying them here.

That is a deliberate line, and it matters. The subject of this work is the counting, not the dead. Summary numbers answer the question that drives the series, who can see how many, without turning a person's death into a row on someone else's website. Each original project keeps its own terms.

How you can check it

You do not have to take my word for any of it. Three separate things keep the numbers honest:

On the page. The main hub re-adds the important numbers from the raw data every time it loads, right in your browser. If one ever drifted, a little check on the page would turn red.
On your own computer. Download the bundle and it comes with a fingerprint for each file. Run a one-line check and your copy either matches, exactly, or it doesn't. No trust required.
Behind the scenes. Every time the site updates, an automatic test rebuilds this data from scratch and refuses to publish if a single number moved without being changed on purpose. The essays and the download can't quietly drift apart.

If you want to cite it:

Lawrence, J. (2026). Counting the Killings: how the United States counts people killed by police, as open data (v0.2.0). jakelawrence.xyz/research/police-shootings.

The point, one more time

If a society wants to deal with something, it first has to be able to count it. A count that nobody can check is barely better than no count at all. The five projects in this series each made their own count checkable. This bundle does the same for the comparison between them: open it, test it, argue with it, on the record. The last essay uses it to answer the hardest question in the whole series.

Next: No. 06, How Many Did Nobody Count →

Download the data The reconciliation hub ← the series