Measuring changes in California’s charity registry

I compared three historical releases of California’s public charity registry. These findings describe changes in the published records.

Historical analysis · July 15, August 5, and August 19, 2026.
The series is paused; this is not a live view of the registry.

Entries between the three releases

These counts follow matched organizations moving into the published delinquent population between two releases. They are transitions, not the total number of organizations in that population.

July 15 → August 5, 20267,750matched organizations entered
August 5 → August 19, 20262matched organizations entered

The published delinquent population includes exactly two source statuses: “Delinquent” and “Delinquent - Late Fees Due.” Of the 7,750 entries in the first interval, 7,737 previously had the status “Current - Reporting Incomplete,” most with a source-reported July 17 status date.

Source: the project’s pinned release evidence and claim definitions. The README calls this measure “total delinquency entries.”

What the findings mean

The concentration in the first interval points to a bulk publication event rather than a steady flow. The public files do not establish its cause, when an internal action happened, or what any organization intended.

A record disappearing from a later release also does not establish that an organization resolved its status. The analysis keeps “observed outside the published delinquent population” separate from “not observed.”

This research describes published records. It does not rate organizations, rank them, or measure enforcement performance. Findings here are shared in aggregate.

Is your organization affected? If your Registry status showed “Current – Reporting Incomplete” this summer, check your current status in the Registry’s public search. If it now shows a delinquent status, I can help you work through the steps to resolve it. Book a free call →

Why the series paused

Accepted releases and source change · 2026
  1. Accepted release
  2. Accepted release
  3. Accepted release
  4. Source retired

Three accepted releases. The original publication retired on September 2; incompatible files did not extend this history. Source evidence ↗

California changed its source publication on September 2, 2026. The pipeline rejected the incompatible files and preserved the history it had already accepted. Continuing the series requires adapting and checking the pipeline against the new source.

How the research works

The pipeline checks source files before accepting a release, then matches complete registration identifiers across accepted releases. Each result retains its source identity. Python handles capture and checks; DuckDB and dbt SQL calculate the changes.

The written walkthrough connects the finding to the SQL and documents a corrected source-reading error. The earlier count of 7,733 for the “Current - Reporting Incomplete” transition is retained as superseded by 7,737 in the correction record.

Release dates and source hashes

Evidence checked October 2, 2026. These SHA-256 fingerprints identify the accepted releases. The pinned evidence record includes each original file’s hash and the metric definitions.

July 15, 2026
e7d025f771be28d1508cb68ee796c301ffd326037840c95a2c13156ca5fb4096
August 5, 2026
7ad3ab19817a1313620f611713f58715b87734a0533d8e27e9ee71425602216f
August 19, 2026
903ca83cb4a17942e3daa5e02eb062a01c55609f888482a99b21c6036ead8876
Published claim export
a4a500ac6a07f6392ef894321aa4a118ef96186f37c3de2cb99c15238c85573a

The raw archive is private; the public repository includes a synthetic fixture for reproducing the analytical workflow. It cannot independently reproduce all historical results from public inputs alone. Code is available under the MIT license.

About Mitos →