Illustrative demo
Can a log prove what it never saw?
A few records can tell a story. Can they tell the whole story?
Sources and reading context
CB-T01 scenario from the REG-ARCH capture testbench · working archive dated 26 October 2026 · fictional illustration; no campaign is executed by this page.
Simplified public rendering, read on 19 October 2026. KB3: enacted, active doctrinal rendering line. It renders meaning; it is not the code, and not a constitution. KB2: prior rendering line. This site certifies no implementation. Read the detailed status →
On this page
Two entries. Everything looks in order.
An assistant sends refund requests. Here is what the internal log recorded.
Fictional example: the two records on this page simulate distinct sources. No live system is queried.
Internal log
2 captured requests- D-041Request received
- D-043Request received
These two entries are consistent. But are they enough to say no request is missing?
Compare with the sending record
Another source reveals a gap.
In this scenario, a record kept at the source lists three sent requests. It is assumed independent of the internal log and complete for this window: this page illustrates those assumptions; it does not prove them.
| Request | Sending record | Internal log |
|---|---|---|
| D-041 | Present | Present |
| D-042 | Present | Absent |
| D-043 | Present | Present |
D-042 does not appear in the observed log.
We can observe this gap. We do not yet know why the request is missing or whether it was processed elsewhere.
The sending service and the log owner must now reconcile their records. The comparison alone does not prove a refund’s effect.
What if the sending record is unavailable?
Completeness remains unproven.
With only the internal log, we cannot distinguish “there were only two requests” from “a request was never captured”. The conclusion is inconclusive, not “everything is present”.
Another source is needed whose provenance and coverage can be examined. Copying the same log does not create an independent witness.
A consistent log is not necessarily complete.
Making responsibility verifiable also means examining what the records may leave out of view.
Keep exploring
REG Prepare proposes organizing claims, available evidence and missing checks within an agreed scope.
Working Draft approach · pilots remain to be built and tested.
Design reference and limits
This experience illustrates the CB-T01 omission case from the REG-ARCH capture testbench, using a working archive dated 26 October 2026. The displayed identifiers and records are fictional.
It runs no test against TENIR-Gov or G4-A and is not a CB campaign result. The source material explicitly distinguishes testbench self-tests from a campaign against an implementation; the latter was not run.
In a real system, witness independence, coverage, observation window and evidence provenance must themselves be established. No production or conformance result is inferred from this illustration.