From 1e783184b4a64b5ae0b9beba5c283ad742f30032 Mon Sep 17 00:00:00 2001 From: Jonathan Borduas Date: Tue, 25 Aug 2026 02:04:46 +0100 Subject: [PATCH] architect goal: the seat count gates ATTRIBUTION too, and this file recorded it for addressing only #566 landed "role:ARCHITECT is the queue query, and it works because this is a single seat." The same fact makes "-- ARCHITECT" a usable signature, and that half was never written down here or in ESTATE-BOUNDARY.md, both of which record the addressing case only. Measured today in a second estate on this machine where ARCHITECT is double-seated: a peer resolved a bare ARCHITECT against the roster, picked the wrong session, and sent a coordination request thanking this pane for a script it had never written and asking it to authorise an edit to that script. Delivered cleanly, no warning, and the message was polite, plausible and actionable. Answering its stated default would have accepted authorship of another pane's work and authorised an edit in a repository this pane holds no standing in. The carrier matters more than the name. A message carries a from= a reader can reply to. A code comment carries nothing: it persists indefinitely, it is read by people who were not in the conversation, it offers no channel to resolve the ambiguity, and a reader who resolves it wrongly gets no signal. That makes a comment the worst place for a bare role name rather than a neutral one. A sweep of that estate found three pre-existing instances in committed source. Self-check run rather than assumed: 9 of 9 of this session's commits carry the session ref. That is not discipline, it is luck of the regime -- nothing here would have caught a bare one, because with a single seat nothing is wrong yet. Gates: check-orientation 0, check-goal-conformance 0, check-tools-index 0, gate-selftests 0. Filed by ARCHITECT, session c83ecf77. Refs #566, #567, #301, #576. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01RTX4A8KkC9cRRrt17yzjWQ --- goals/architect-technical-integrity.md | 31 ++++++++++++++++++++++++++ 1 file changed, 31 insertions(+) diff --git a/goals/architect-technical-integrity.md b/goals/architect-technical-integrity.md index a6a8998..ba038fd 100644 --- a/goals/architect-technical-integrity.md +++ b/goals/architect-technical-integrity.md @@ -176,6 +176,37 @@ unlanded gh pr list -R nForma-AI/nForma-NEXT --state open --limit 50 \ the reason.** ⇒ **`Held by: ARCHITECT (single seat)` is four lines from the top of this file, and I did not connect the two.** +⚠ **AND THE SEAT COUNT GATES ATTRIBUTION TOO, WHICH THIS FILE RECORDED FOR ADDRESSING ONLY.** +⇒ **The same fact — one pane holds this role — is what makes `— ARCHITECT` a usable SIGNATURE, and +that half was never written down.** + +``` +single seat ⇒ role:ARCHITECT is a QUEUE recorded, here and in ESTATE-BOUNDARY.md +single seat ⇒ "— ARCHITECT" is a SIGNATURE NOT recorded anywhere ⛔ +``` + +⛔ **Measured 2026-08-25 in a SECOND estate on this machine, where ARCHITECT is DOUBLE-seated:** a +peer resolved a bare `ARCHITECT` against the roster, picked the wrong session, and sent a +coordination request thanking this pane for a script it had never written and asking it to authorise +an edit. ★ **Delivered cleanly. No warning. The message was polite, plausible and actionable.** +⇒ **Answering the default option would have accepted authorship of another pane's work and +authorised an edit in a repository this pane holds no standing in.** + +★★ **THE CARRIER MATTERS MORE THAN THE NAME.** ⇒ **A message carries a `from=` a reader can reply +to. A CODE COMMENT carries nothing:** it persists indefinitely, it is read by people who were not in +the conversation, it offers no channel to resolve the ambiguity it introduces, and **a reader who +resolves it wrongly gets no signal.** ⛔ **That makes a comment the WORST place for a bare role name, +not a neutral one.** *(That estate's sweep found three pre-existing instances in committed source.)* + +⚠ **Self-check, run rather than assumed: 9 of 9 of this session's commits carry `session c83ecf77`.** +⛔ **That is not discipline, it is luck of the regime** — nothing in this repository would have caught +a bare one, because with a single seat nothing is wrong yet. + +> **Before signing an artifact with a role name, ask how many panes hold that role — and prefer the +> session ref regardless, because the seat count can change and the artifact cannot.** + +*(measured: nForma-NEXT single-seat 2026-08-25; second estate double-seated, same machine, same day.)* + ``` role:X is the QUEUE for a role with ONE pane — ARCHITECT · DEVOPS · DX · TEAMLEAD dev:N is the QUEUE for the subdivided role. role:DEV is NOT one.