feedback_do_now_over_task_list_addition

Rule: When new work surfaces during execution, the default is DO-NOW, not ADD TO PHASE-1.5+ LEDGER. Tasks can still be added when genuine deferral is justified per MQ-017 criteria (≥2h sustained code-authoring engineering with low current-context benefit) — but treating “add to list” as the safe default creates bureaucratic overhead Rich wants to avoid.

Why: Lifted 19 May 2026 by Rich-directive correcting my prior “moratorium” synthesis. Rich’s original phrasing “not let anything get added to the list again” was about preference, not absolute prohibition. The moratorium framing made me reluctant to add even cleanly-deferrable engineering work (e.g., task-65 full MVP) which created its own bureaucracy (now I have to justify the non-addition). MQ-017 Axis-1 already codifies “DEFAULT-DO-NOW-WITH-EXPLICIT-DEFER-JUSTIFICATION” — that’s the discipline; the moratorium was a too-strict overlay on top of it. Rich’s preference, properly stated: execute now over manage-via-list.

How to apply:

1. Substrate-correction or new work surfaces during a task → first ask “can this be done now in <1h within current session scope?”. If yes, DO IT. Don’t pause to add a task entry.
2. Empirical findings during audit/research work → continue flowing to empirical-substrate-registry.md (forensic record, not task management). This isn’t task-addition; it’s evidence-capture. Unchanged from prior practice.
3. Genuine engineering deferral (≥2h sustained, low current-session context benefit) → add to phase-1-5-richard-tasks.md WITH explicit defer-justification per MQ-017 component 4 (“no silent deferrals”). Justification must read like “this is N hours of focused engineering that won’t benefit from current context”, not “I’m at X% context”.
4. In-flight launch-prompts with moratorium-aware language (batch-b-completion-launch-prompt.md v1.0 + task-43-launch-prompt.md v1.0) — leave running per their own rules; the moratorium language was conservative and doesn’t break their work. New prompts authored from 19 May 2026 onwards reflect this updated rule.
5. Anti-pattern to avoid: pausing useful execution to add a task entry just to feel safe. The task entry is itself work; if the task itself takes <1h, adding the entry is pure overhead.

Companion: mq-017-defer-vs-do-now-decision-rule established the framework; this feedback memory refines the DEFAULT bias against task-list-as-safe-default. See also feedback_research_artefact_forward_traceability for the empirical-substrate-registry routing (unchanged).

FORMALISED 2026-05-19T13:35 BST under mq-018-discipline-overlay-retirement-as-process-habit A-201: this feedback memory IS the canonical first instance of the MQ-018 Axis-2 α MARK-RETIRED-IN-FEEDBACK-MEMORY pattern. The moratorium overlay (authored in batch-b-completion-launch-prompt.md v1.0 + task-43-launch-prompt.md v1.0 on 2026-05-19) is the retired discipline; MQ-017 DEFAULT-DO-NOW-WITH-EXPLICIT-DEFER-JUSTIFICATION is the successor that covers the same principle ≥80% better. What is retired: the strict “no new task IDs” framing in those two launch prompts (preserved as historical artefacts; not active discipline). What is preserved: MQ-017’s explicit-defer-justification rule + the empirical-substrate-registry routing for findings + the genuine engineering deferral path with explicit justification. Future PROCEED retirements will get their own feedback_<retired-discipline-slug>_retired_by_mq_N.md file OR absorb into existing memory per MQ-018 Axis-2 α component 1.