Supporting methodology and editorial history

Article: OpenSanctions in 8DB: which conclusions survive a source change?

Retained October 9, 2026 from the article at source c04aea4. These passages preserve the original scope, methodology and revision history behind the commercial summary. They are not new test results. Original source references remain in the article.

The practical distinction is visible in the explorer: removing a source does not force every conclusion into one shared valid-or-invalid state. Each conclusion is checked against the evidence it actually needs. That is the behavior an evaluation should compare with an existing application, alongside implementation effort and update cost. We have not run that comparative evaluation yet.

Revision note, September 17, 2026: this replaces the earlier broad entity-resolution account with a bounded downstream-evidence experiment. The available run artifacts do not substantiate the earlier claims of a full provenance chain for every merge or byte-for-byte reproduction of every finding. Those claims have been removed. See the ownership case and corrected chronology for the analytical context.

These checks demonstrate consistent updates for the tested changes. We did not measure a speed advantage, test every possible correction, or run an end-to-end production service. The useful result is that the implementation preserved the difference between a supported link, a supported dated transition and a qualifying multi-company pattern.