A2 example: Restore Iris Relay

Teaching demonstration, not an assigned response or a claim of real student activity.

All voltage readings use the common return. The unchanged voltage trace does not prove the lamp repair; the OPEN pre-repair continuity and verified light explain it. Ari replaced only the lamp with the source isolated.

Original signed scope: exactly 22 core sections, digest I-4.

The signed source card supplies both the I-4 core identity and the I-9 bundle composition. Delta and Echo are equally defensible matches for the original request. A later transfer timestamp does not distinguish their content.

All nine resulting decisions were checked. Positive regression: maintainer service allowed. Negative regression: reader approval denied. The other seven results also matched the charter.

• Core package: South Dome / 4 / TEAL. Nia initially repeated 3. Ari corrected the count; Nia repeated all fields and Ari confirmed.
• Amended trace package: Hill Station / 2 / GOLD. Ari repeated all fields and Nia confirmed. This receipt applies to the Foxtrot bundle, not the earlier core-only instruction.

Version 1: accepted both Delta and Echo under the 22-section/I-4 core-only scope; completed the first package handoff.

Signed scope amendment: deliver the same core plus six calibration-trace sections, exactly 28 sections with bundle digest I-9.

Version 2: selected Foxtrot under that larger evidence requirement. The original mirrors remain valid for the original scope. The permission charter remains unchanged; the new delivery requires its own confirmed handoff.

Ari: I collected the powered readings, performed the isolated continuity and replacement sequence, and verified the lamp output. After rotation I checked the archive scope and policy; I later received the amended handoff. I did not claim that identical mirrors required a unique winner.

Nia: I proposed the continuity check because available supply did not prove a healthy load. I then operated the record and policy comparisons, preserved my corrected count error, and issued the amended trace-package instruction after another role exchange.

Our pair restored Iris Relay and prepared the research package authorised by its current signed request. Ari initially operated the circuit while Nia held the requirement cards and predicted what each possible fault would produce. We agreed to exchange responsibilities after verification. This is an authored example of paired reasoning: the contribution records show who measured, compared, operated and confirmed, rather than treating a shared result as proof of equal understanding.

The lamp was dark, but Ari measured twelve volts at the source, fuse output, cable output and lamp supply, all against the common return. These readings established available supply throughout the model. They did not establish a complete path through the lamp. Nia rejected an immediate fuse replacement because there was no missing supply after that component. We chose a different test: isolate the source and inspect continuity through the load, with healthy components as comparisons.

With power off, the supplied continuity record showed the fuse and cable continuous and the lamp open. That evidence supported replacing the lamp. Ari made the replacement while isolated, then restored power. All four supply readings remained twelve volts, but the lamp now lit. The voltage trace alone therefore could not explain the repair; the retained continuity result and changed output could. We preserved both the original dark condition and the verified result before rotating roles.

Nia next operated the archive console while Ari stated the first request: exactly twenty-two core sections with digest I-4. Delta and Echo had identical content satisfying both conditions. Their different transfer timestamps did not create different content identities. We recorded both as matching mirrors instead of inventing a unique winner. Foxtrot contained the same core plus six calibration-trace sections, forming a twenty-eight-section bundle with digest I-9. Its extra material was outside the original exact scope, without making that material false or corrupt.

The permission charter remained a separate question. The reader originally held excess service and approval rights, while specialist roles lacked their required actions. We repaired all nine cells: everyone could read, only the maintainer could service, and only the custodian could approve. We checked both the permitted maintainer service and the denied reader approval, alongside the other seven decisions. Denying everything would remove misuse but also prevent the relay from performing its intended work.

For the first delivery, either matching mirror supported the core package. Ari issued South Dome, four units, TEAL. Nia initially repeated three units. Ari corrected that discrepancy, and Nia repeated the entire message before confirmation. We retained the mistake because it demonstrates what the protocol caught. Confirmation applied to this package and these fields; it was not general permission for a later delivery.

The signed amendment then required the core together with six calibration-trace sections, exactly twenty-eight sections and bundle digest I-9. Foxtrot alone met that enlarged scope. Delta and Echo remained valid answers to the earlier request, so Version Two preserved the original two-match finding rather than rewriting it as failure. The amendment changed the deliverable, not the permission charter. Roles rotated again: Nia issued Hill Station, two units, GOLD, and Ari completed a fresh read-back.

Our completed evidence links diagnosis, content scope, authority and communication without treating them as interchangeable checks. Matching a supplied digest establishes reference agreement within this model; it does not establish scientific truth, permanent hardware reliability or a real-world delivery. In a new task we would reuse the reasoning method, identify its actual measurement conventions and requirements, and collect fresh evidence before reusing any conclusion.

The names, actions and records are a complete authored example. No physical measurements, external system tests, real student participation or real delivery are claimed. An assessed pair must provide its own records, reasoning, contributions and declared assistance.

Use information held by another role to choose and verify your next action.

Iris has a load fault and an amended calibration deliverable; the assigned relay requires its own source and receiver checks.

The complete example explains one defensible approach. In your own challenge, select observations and tests from the supplied rules, keep your prediction separate from the result, and justify both the checks you perform and those you omit.