free-memory(never-idle refinement): proper order > smallest-thing-first; depends_on graph as ordering substrate (Aaron 2026-05-02)

[AceHack]

Summary

Aaron 2026-05-02 corrected my real-time application of the never-idle discipline after I defaulted to "Loop continues" minimal-acknowledgment for many ticks during an extended Aaron-pause. The correction landed as a refinement appended to the existing feedback_never_idle_speculative_work_over_waiting.md substrate (which CLAUDE.md already points at).

Aaron's framings preserved verbatim

"i would not call all backlog items speculative, many are but some are concered get this small thing done but proper order matters more than doing the smallest thing you started adding like depend on so the order could be tracked too i don't know how far you got."

"that backlog depends on never idle refinement priority is something that future ai should remember too"

"we can slowly figure it out over time too, no rush we can do best guesses as you but out proper infrstructure for this"

Five corrective clauses

1. NOT all backlog items are speculative — many are concrete-and-need-to-get-done
2. Proper ORDER matters more than smallest-thing-first — picking the smallest hygiene-only closure is NOT the right priority lever; dependency order is
3. depends_on: is the ordering substrate — currently 11/148 coverage; backfill incrementally on-demand
4. Order signal sequence: depends_on: graph > tier (P0>P1>P2>P3) > tier: frontmatter > effort/blast-radius > composes_with: clusters > date-recency
5. Backfill discipline (Aaron's no-rush framing): best-guesses-with-time + slowly figure it out + build out proper infrastructure + no rush

Why this lands as a refinement to existing memo, not a new file

CLAUDE.md already points at feedback_never_idle_speculative_work_over_waiting.md. Future-Otto's wake-time scan picks up the refinement automatically via that existing pointer. No separate CLAUDE.md edit needed.

Composes with

• memory/feedback_action_hierarchy_evidence_over_speculation_friction_reducing_over_neutral_aaron_2026_05_02.md (action-pick lens; same priority-order discipline at action-pick scope)
• memory/feedback_amortized_speed_superfluid_phase_transition_inverts_per_action_optimization_aaron_2026_05_02.md (system-level lens; backlog dependency-order respect IS one instance of the inversion against per-action local-optimization)
• memory/feedback_recurrence_after_correction_needs_operational_enforcement_otto_2026_05_02.md (the no-op-cadence pattern this correction surfaces; recurrence pattern from earlier today repeated AGAIN during evening Aaron-pause, prompting this refinement)

Test plan

• Refinement appended to existing memo (CLAUDE.md pointer already covers it)
• All five corrective clauses present
• Composes-with section
• Carved sentence at end
• CI green

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR updates project memory and backlog tracking around task selection and code-scanning cleanup. It appends a dependency-aware refinement to the existing never-idle memory, closes backlog row B-0073 based on current Code Scanning state, and syncs the generated backlog index to reflect that closure.

Changes:

• Appends a 2026-05-02 refinement to the never-idle memory, prioritizing dependency order over “smallest thing first.”
• Marks backlog item B-0073 as closed and adds a closure note describing the current Code Scanning state.
• Updates docs/BACKLOG.md so the generated index reflects B-0073 as closed.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 5 comments.

File	Description
memory/feedback_never_idle_speculative_work_over_waiting.md	Adds new refinement text describing ordering heuristics for backlog work under the never-idle discipline.
docs/backlog/P0/B-0073-lfg-csharp-code-scanning-cleanup-13-alerts-blocking-ruleset-2026-04-28.md	Closes the P0 backlog row and records closure rationale/evidence.
docs/BACKLOG.md	Updates the generated backlog index checkbox for B-0073 to match the row’s closed status.