Case Zero External Testing Addendum

Evidence boundary

This addendum supplements, but does not alter, the currently published Case Zero report, The Search Result Is Not the Page. It should be read alongside the currently published Semantic Authority Network framework. The original findings, counts, controls, screenshots, and causal boundaries remain unchanged.

The tests summarized here occurred after the initial Case Zero publication. They are external attempts to observe whether comparable search-result behavior appeared in other locations and browsing contexts. They do not establish why the behavior appeared, whether a particular search or retrieval mechanism caused it, or whether any result will persist.

The protocol and four outcome labels below were confirmed by the evidence owner for Round 2. They are final classifications for this round and are not strengthened beyond the visible evidence.

Owner-confirmed testing protocol

1. Every participant searched Google for Eugene Prudchenko.
2. Search depth was limited to the first four Google result pages.
3. Participants were not shown the expected anomaly examples before testing.
4. Private or Incognito mode was used.
5. No VPN was used.
6. Testing was external to the original Case Zero environment.
7. This addendum records only owner-confirmed browser and account state. Any field not separately confirmed remains unconfirmed rather than inferred.
8. This addendum compares only visible outputs. It does not infer index state, entity-graph state, embedding behavior, ranking logic, personalization, or regulatory causation.

For a future controlled round, the protocol should also pre-register the exact query set, canonical destination URLs, timestamp, signed-in state, browser/version, language, device/viewport, network location, result depth, and adjudication rule.

Environment matrix

The evidence package contains no raw HTTP responses, search export, browser automation log, or Google-side diagnostic. IP-derived location, account state, contact details, and raw participant messages remain private.

Results

REP-001 — positive

The Kyrgyzstan run is consistent with the supplied positive classification. The screenshots show the canonical site and Eugene-linked material appearing throughout the result sequence, including third-party surfaces. This supports external observability in that environment. It does not establish that every displayed identity signal was a cross-URL attribution candidate under the original Case Zero adjudication rule.

REP-002 — mixed

The Lithuania run is consistent with a mixed result. Early views contain substantial name ambiguity and unrelated material; later views show the canonical site and Eugene-linked profiles or posts. The package therefore supports context-dependent discoverability, not a uniform outcome across environments.

One screenshot contains Google's notice that some results may have been removed under European data-protection law. The notice is part of the observed interface. It does not show that European regulation caused any difference between regions, and no such causal claim should be published.

REP-003 — positive

The Georgia run is consistent with the supplied positive classification. The canonical site appears prominently, and subsequent result pages contain multiple Eugene-linked snippets and third-party surfaces. Visible identity-linked representations were observed in that environment.

REP-004 — negative

The Germany run is retained as a negative result. A direct canonical result is visible, but the owner-supplied classification indicates that the target cross-URL behavior was not observed. This distinction prevents ordinary branded discovery from being counted as the Case Zero phenomenon.

Same-URL comparison

No Georgia-Lithuania same-URL anomaly comparison is included in Round 2. The source files do not establish an exact canonical third-party destination URL in both runs with sufficient certainty. Account-level overlap and truncated paths are not treated as URL identity.

Exploratory hypothesis: regional variability

The four outcomes are compatible with regional or contextual variability: two positive, one mixed, and one negative. That pattern is exploratory only. Region is confounded with participant, device, language, time, result depth, browser state, network, and possible index variation. The sample is too small and uncontrolled to attribute differences to geography, regulation, or any technical mechanism.

Limitations

• Four participants are not a representative geographic sample.
• The protocol was confirmed after testing rather than pre-registered.
• Exact capture times and browser states are incomplete for some runs.
• The owner-confirmed search depth was four result pages, but the package is not a machine-generated page-by-page export.
• Composite PDFs duplicate screenshots and add no independent observation.
• Screenshots show outputs, not search-engine internals.
• A visible direct result for the canonical site is not automatically a Case Zero-like observation.
• No inference can be made about Google authorship judgments, semantic embeddings, knowledge graphs, or regulation.
• The tests occurred after initial publication, so they are follow-up observations rather than part of the original Case Zero corpus.

Conclusion

Case Zero-like identity-linked search representations were observed in two external testing environments, while two others produced mixed or negative results. The evidence expands the observation surface while leaving the original Case Zero findings and causal uncertainty unchanged.