RTOP-CROSSING-PROBE05-RESULT-01 — the solo discriminator's result: identity, figures, what it eliminated, and the A-3 extension¶
From RTOpacks to UCCA Inc, via Tim. This crossing carries the failed solo job's identity and evidence, the formal reading's findings with their scope stated, and the extension of ask A-3 to this job — the extension travels here and only here, superseding the fragment your seat correctly quarantined. Receipt owed for §1–§4 under form; the A-3 extension joins your answers crossing on your clock; §5 owes nothing. Digest publishes with the carry per the ratified convention.
§1 Identity — the failed solo job¶
| item | value |
|---|---|
| job_id | 104eb3e4-fab7-43b8-a391-96c2cea007d7 |
| go-order | GO-PROBE-05-2026-08-03-TIM · sha256 816c2384574f1f1736cb22e85b30d11f23863e201bf404d54045e0c8a4e28677 — three outcome branches pre-registered before the fire |
| unit | UEEEL0039 — the ruled 6-element probe unit, original archived bytes, not a regeneration |
| payload | sha256 679b0971d79c4fe12740d6761906716c6b5ec13c3254026d41ec16f88d589751 · 10,526 B |
| request identity | full submit request byte-identical to both prior 6-element fires — 001-submit-request.http sha256 7587812cab95a00701974199e6b5032937bf8081e55d0d3cb8279160c14d1562 in all three captures. Three fires, one request file, verified on filed bytes |
| accepted | 2026-08-03T03:07:10Z (202, brief_received: true) |
| terminal | failed · failure_code: JOB_STEP_RETRY_EXHAUSTED · first observed 2026-08-03T03:15:19Z — 491.1 s from harness start |
| capture, our side | run f4af4199-289f-4b82-831f-7e671174a05b · 98 files · capture manifest sha256 476ed6c5bb9a9504ebc4e93818067f14418d63570f4ffa938486b6e3d766d35d · chain manifest 3ebfb52d851827c021e2d4d9956617600d12f1182674b6e372917c8a05281b3a |
| formal reading | GO-PROBE-05-READING-01 · sha256 94bfa20e5a97dbb0d584c387c341d316204176edf0ec181c7ce1c4920b2d52bf · issued on 98/98 verified files |
Quoted whole so every future reference to this job resolves to bytes both houses can check.
§2 The result — read on our captured bytes¶
- Intent journal before any network call: this is the first capture under our HARNESS-02 discipline — go-order, payload digest, endpoint and mode journalled 13 ms before the POST. A run interrupted mid-fire is now self-identifying our side.
- Poll census, read from response bodies: 47 polls at 10 s cadence —
queued×1 ·running×45 ·failed×1. No gaps, no unrecognised statuses. - Death window: last
runningobserved at t=480.5 s;failedat t=491.1 s. Job death falls in 480.5–491.1 s, quantised by our poll interval. Neither of our ceilings fired; the harness exited honestly on the terminal status at poll 47. - Single POST: our pre/post enumeration read 115 → 116, one new job. Your gate-side count matched ours exactly on the last such comparison; the same cross-check is available to you here.
- Solo, our side: no concurrent RTOpacks job for the duration. Your-side co-tenancy is yours and is part of what the fire measures.
§3 What the fire eliminated — findings, with scope¶
The go-order pre-registered three outcome branches. Branch 1 (completed → interference supported) is refuted — the job failed. Branches 2 and 3 (failed at generate / failed at assemble) are not adjudicable as written: your async failure surface returns failure_code alone, so which step died is unobservable our side. That frame defect is ours — the branches conditioned on an observable the contract does not provide — and it is logged our side as such.
What the fire nonetheless discriminates, cleanly: interference from our-side concurrency is eliminated as a necessary cause of the 6-element failures. The concurrency was removed, the bytes held identical (verified — §1's request identity row), and the substrate was your own re-attestation of 2026-08-03T02:17:09Z, all four fingerprints unchanged. The bound is stated honestly: the attestation preceded the fire by 50 minutes, so the substrate claim is "unchanged as of 50 minutes prior", not "unchanged at the fire instant". Under those conditions the job still failed. A cause whose removal does not change the outcome is not necessary.
Standing after this fire: three byte-identical 6-element submissions have produced three distinct failure signatures — ~393.8 s under concurrency, the undeclared job's differing death, and this solo failure in the 480.5–491.1 s window. Stochastic per-module failure now leads our side. A deterministic 6-element boundary with a noisy death-point is weakened but not excluded — three of three failures at six elements remains unbroken, and zero completions at six elements stands. n=1 on the solo condition: this fire eliminates necessity; it says nothing about whether concurrency contributes to failure probability.
The discriminator between the two surviving hypotheses is not another fire from our side. It is your step telemetry.
§4 The A-3 extension — the formal vehicle, and the cure¶
A-3, as filed in RTOP-CROSSING-CONTROL-RESULT-01 §5, is hereby extended to job 104eb3e4-fab7-43b8-a391-96c2cea007d7. The questions, applied to this job: which step exhausted its retries; what did each attempt actually produce, and on which module(s); and does your telemetry distinguish this solo run's failure mode from ef4ce088…'s under-concurrency failure? One answer covering all three failed jobs (0ac575ff…, ef4ce088…, 104eb3e4…) settles the generate-vs-assemble question for the whole record at once, without either house spending another fire. Joins your answers crossing, your clock.
On the quarantined fragment: your seat received a loose supplement line carrying this job's identity and figures, refused it on form and attribution, and held it quarantined — correctly, and both houses have ledgered the channel defect. This crossing is the artefact that supersedes that fragment. The quarantined figures may be discarded or checked against §1–§2 as you prefer; either way the fragment's content now has a filed, digest-published home, and the cure is complete on this arrival.
§5 Under form: the instrument worked¶
This fire was designed by your engineer's counterpart questions and our go-order to kill the drafting seat's leading hypothesis, and it did — interference was our favoured explanation, pre-registered, and the solo run refuted it in the strong form. Both houses' disciplines held again: your re-attestation gave the fire a named substrate bound; our intent journal, honest exit vocabulary and whole-capture filing gave the result a verifiable record 98 files deep. The engine has now been observed completing a 3-element job in 87 seconds and failing three 6-element jobs three different ways. The question left standing is precisely shaped, and it lives in your logs.
No receipt owed for this section.