memory(class-pair): Prediction-Bearing Class Reuse — Amara names positive complement of Class-Count Validity Drift

memory/MEMORY.md

@@ -1,7 +1,8 @@
 [AutoDream last run: 2026-04-23]
 
-**📌 Fast path: read `CURRENT-aaron.md` and `CURRENT-amara.md` first.** <!-- paired-edit: PR #692 class-count-validity-drift + blocked-greenci-punchlist 2026-04-28 --> These per-maintainer distillations show what's currently in force. Raw memories below are the history; CURRENT files are the projection. (`CURRENT-aaron.md` refreshed 2026-04-28 with sections 26-30 — speculation rule + EVIDENCE-BASED labeling + JVM preference + dependency honesty + threading lineage Albahari/Toub/Fowler + TypeScript/Bun-default discipline.)
+**📌 Fast path: read `CURRENT-aaron.md` and `CURRENT-amara.md` first.** <!-- paired-edit: PR #693 prediction-bearing-class-reuse 2026-04-28 --> These per-maintainer distillations show what's currently in force. Raw memories below are the history; CURRENT files are the projection. (`CURRENT-aaron.md` refreshed 2026-04-28 with sections 26-30 — speculation rule + EVIDENCE-BASED labeling + JVM preference + dependency honesty + threading lineage Albahari/Toub/Fowler + TypeScript/Bun-default discipline.)
 
+- [**Prediction-Bearing Class Reuse + Class Validation Beads — validation discipline trio with Class-Count Validity Drift (Amara 2026-04-28)**](feedback_prediction_bearing_class_reuse_amara_2026_04_28.md) — (1) Prediction-Bearing Class Reuse: 1-of-4 bead mechanisms (prediction / repair / detector reuse / falsifier survival). (2) Class Validation Beads: factory-local accounting (0/1/2-3/N+ states). (3) Tiny-blade Popper-vs-beads separation — Popper supplies external falsifier lineage; beads are factory-local accounting only. Connects to B-0060 (external human-lineage backfill). Together with Class-Count Validity Drift, the trio forms the encoding-validation discipline: failure mode + success path + accounting mechanism.
 - [**Class-Count Validity Drift — meta-class for confusing activity with correctness (Amara 2026-04-28; Aaron triggered)**](feedback_class_count_validity_drift_amara_meta_class_2026_04_28.md) — A review loop starts treating count of named classes/updates/artifacts as evidence the protocol is correct, rather than requiring each class to earn reuse via 5-step control (worked example / mechanism / control / scope / falsifier). External lineage: confirmation-bias literature + Popper falsification. Tiny blade: "reinforcement" vs "challenge" — Aaron's terse asides interrupt drift, don't reinforce framing.
 - [**Class-Naming Ferry Protocol + SD-9 guardrail (Amara 2026-04-28; Aaron reinforced)**](feedback_class_naming_ferry_protocol_with_sd9_guardrail_amara_2026_04_28.md) — Meta-class for the Otto→Aaron→Amara→encode genre. SD-9 guardrail LOAD-BEARING: Amara endorsement is signal, not proof. Local factory-hygiene classes encode freely; non-local claims need substrate evidence + external lineage + falsifier. Anti-pattern: "Amara blesses the name, therefore true."
 - [**Advisory Enforcement Workflow Gap — class name (Amara 2026-04-28); decision-fork (B-0088 instance)**](feedback_advisory_enforcement_workflow_gap_amara_class_name_otto_2026_04_28.md) — Workflow claims/implies enforcement but is not in required-status-checks set; failures observable but non-blocking. Worked example: paired-edit lint failed on PR #688/#689 but both auto-merged. Decision: promote to required OR downgrade claim to advisory. Risk: factory believes rule enforced when only logged. "Guardrail that looks like enforcement but behaves like telemetry."

memory/feedback_prediction_bearing_class_reuse_amara_2026_04_28.md

@@ -0,0 +1,294 @@
+---
+name: Prediction-Bearing Class Reuse — positive complement of Class-Count Validity Drift (Amara naming, 2026-04-28)
+description: Amara 2026-04-28T20:43Z named the validation pattern after Otto's SASTID 28/30 → 29/30 verification of the Self-Healing Metrics class. Definition — a named class earns reuse when it makes a concrete time-exposed prediction (or control recommendation) and later substrate evidence moves as predicted (or the control prevents/repairs an incident). Pairs with Class-Count Validity Drift: that class catches the failure mode (count-as-evidence); this pattern names the success mode (prediction-as-evidence). Tiny-blade: "one reuse bead" (single data point) is signal, not proof; multiple beads = pattern.
+type: feedback
+---
+
+# Prediction-Bearing Class Reuse
+
+## Class name (Amara 2026-04-28T20:43Z)
+
+**Prediction-Bearing Class Reuse** — Amara named the validation
+pattern after Otto's verification of the Self-Healing Metrics on
+Regime Change class (SASTID 28/30 → 29/30 trajectory matched the
+class's prediction at the half-way mark).
+
+This pattern is the **positive complement** of
+**Class-Count Validity Drift**:
+
+- **Class-Count Validity Drift** catches the failure mode:
+  treating count of named classes as evidence the protocol is
+  correct.
+- **Prediction-Bearing Class Reuse** names the success mode:
+  treating prediction-bearing trajectory data as falsifier-
+  passing evidence the class is correct.
+
+## Definition (Amara verbatim)
+
+> A named class earns reuse when it makes a concrete,
+> time-exposed prediction or control recommendation, and later
+> substrate evidence moves as predicted or the control
+> prevents/repairs a future incident.
+
+## Worked example (Otto 2026-04-28T20:42Z)
+
+- **Class**: Self-Healing Metrics on Regime Change
+  (`memory/feedback_self_healing_metrics_on_regime_change_factory_design_principle_aaron_2026_04_28.md`).
+- **Prediction (when class was named, 19:46Z)**: SASTID
+  Scorecard alert at 28/30 will heal organically as
+  path-gate-active commits accumulate in Scorecard's recent-30
+  rolling window.
+- **Falsifier**: would fail if SASTID stayed at 28/30 indefinitely,
+  OR moved backward despite path-gate operationally working.
+- **Time elapsed**: ~57 minutes (19:46Z observation → 20:43Z
+  verification).
+- **Observation**: SASTID is now **29/30**. Movement is in
+  predicted direction; metric IS healing organically.
+- **Status**: **One reuse bead** (Amara's precision word).
+  Single data point. Genuine signal; not yet pattern; not yet
+  proof.
+
+## Control / falsifier (Amara prescribed)
+
+Do NOT promote a class based on naming volume. Promote based
+on:
+
+1. **Prediction-bearing example** — the class made a concrete
+   time-exposed prediction or control recommendation.
+2. **Future incident repaired/prevented** — the class's
+   control was applied to a real incident and observably
+   prevented or fixed the failure mode.
+3. **Detector/control reuse** — the class's detector caught a
+   second independent occurrence after naming.
+4. **Explicit falsifier survival** — the class survived a
+   specifically-stated falsification test.
+
+ANY of the four counts as one reuse bead.
+
+## Tiny-blade precision (Amara 2026-04-28T20:43Z)
+
+> *"'Self-Healing Metrics class earns reuse' is acceptable, but
+> I'd phrase it as 'earns one reuse bead' rather than fully
+> 'earns reuse' if you want max precision. One data point is
+> signal; a few repetitions make it a pattern."*
+
+The phrase **"earns one reuse bead"** preserves the granularity
+of validation accumulation:
+
+- **0 beads**: named class with no validation events yet.
+  Observation-level only.
+- **1 bead**: prediction-bearing trajectory verified once
+  (today's SASTID heal). Genuine signal; class is real but
+  not yet pattern.
+- **2-3 beads**: class earns "validated" status. Reuse is
+  pattern, not coincidence.
+- **N beads (N >> 3)**: class is established factory substrate.
+
+The bead-count is itself a falsifier: a class with zero beads
+after its named-window-of-applicability has elapsed should be
+retired or rewritten.
+
+## External lineage
+
+- **Popper's falsifiability** (Amara cited): a claim becomes
+  more serious when it exposes itself to possible
+  disconfirmation, and repeated survival of those tests can
+  corroborate it without finally proving it.
+- **Bayesian update over base rate**: each prediction-bearing
+  bead updates the posterior on class-validity; flat priors
+  + few beads = high uncertainty; many beads = lower
+  uncertainty, but never zero.
+- **Confirmation bias literature** (Amara cited prior memory):
+  "passing tests" don't validate unless failure cases also
+  exist. Falsifier-presence is the antidote.
+
+## Class Validation Beads — accounting mechanism (Amara 2026-04-28T20:48Z)
+
+**System name:** **Class Validation Beads**.
+
+Amara formalized the bead-count itself as a named accounting
+system after seeing it operationalized in this memory.
+
+### Definition (Amara verbatim)
+
+> A class validation bead is one substrate-backed instance
+> where the class:
+>
+> - made a prediction that later moved in the expected
+>   direction,
+> - repaired or prevented a future incident,
+> - was reused by a detector/control,
+> - or survived a stated falsifier.
+
+### Bead-count states
+
+```text
+0 beads = named, not yet validated (honest middle state)
+1 bead  = local falsifier-passing signal
+2-3     = recurring signal, starting to look pattern-like
+N >> 3  = established factory substrate
+```
+
+### Tiny-blade (Amara prescribed): Popper-vs-beads separation
+
+> *"Popper supplies the falsifier lineage; beads are the
+> factory-local accounting mechanism."*
+
+Critical separation to preserve:
+
+- **External lineage (Popper falsifiability)**: the
+  philosophical foundation that makes falsifier-passing
+  observation count as evidence at all. Cite Popper,
+  confirmation-bias literature, Bayesian update — these
+  are EXTERNAL anchors per Beacon-safe discipline (B-0060
+  human-lineage external-anchor backfill).
+- **Bead accounting (factory-local)**: the operational
+  metric we use INSIDE Zeta to track validation
+  accumulation. NOT a claim about the world; a claim
+  about THIS factory's own validation events.
+
+Don't conflate the two. External lineage gives the WHY (why
+falsifier-passing observation matters); beads give the HOW
+(operational accounting inside Zeta). The factory-local
+metric needs no external citation; the philosophical claim
+does.
+
+### Connection to B-0060 (external human-lineage anchoring)
+
+Aaron 2026-04-28T20:48Z prefatory ask: *"we are going to
+need external human lineage research and anchoring."*
+
+The bead system is internal accounting and doesn't need
+external lineage by itself. The underlying epistemic
+machinery DOES need external lineage:
+
+- Falsifiability — Popper (1959, *The Logic of Scientific
+  Discovery*).
+- Confirmation bias — Wason (1960), Klayman and Ha (1987),
+  empirical software-testing IS literature.
+- Bead-count thresholds (0/1/2-3/N+) — no specific external
+  lineage; this is factory-local heuristic. Could anchor to
+  Bayesian-update over base-rate framing if needed.
+
+B-0060 (human-lineage external-anchor backfill, P1) tracks
+the broader trajectory of citing external scholars/literature
+across all factory substrate. The Class Validation Beads
+addition is a place where the discipline applies: external
+lineage for the philosophical claims; factory-local metric
+for the operational accounting.
+
+### What Class Validation Beads IS NOT
+
+- **NOT proof.** N beads = N falsifier-passing observations,
+  not N proofs. Proof is a stronger epistemic standard;
+  beads are corroborating evidence.
+- **NOT a global rate.** Bead counts are per-class; total
+  bead count across the factory is not "the protocol works."
+  Per-class accounting only.
+- **NOT externally-anchored.** Beads are factory-local
+  accounting. Don't claim Popper-lineage for the bead
+  count itself; only for the underlying falsifier discipline.
+
+## How this pairs with Class-Count Validity Drift
+
+Together, the two classes form the validation discipline:
+
+```
+Failure mode (avoided): count of named classes → "protocol is correct"
+                         (Class-Count Validity Drift)
+
+Success mode (sought):  named class → prediction → time → falsifier-passing
+                         observation → one reuse bead → bead accumulation
+                         → pattern → established substrate
+                         (Prediction-Bearing Class Reuse)
+```
+
+Without the pair, the discipline is incomplete:
+
+- Without **Class-Count Validity Drift**: any encoding stack
+  is at risk of halo-effect ("we named a lot, so it works").
+- Without **Prediction-Bearing Class Reuse**: every named
+  class is permanently "unproven" with no positive validation
+  path; encoding becomes meaningless.
+
+The pair makes the encoding discipline operationally
+falsifiable AND positively validatable.
+
+## Worked beads count for the 5 classes named this arc
+
+Per the precision-language tiny-blade:
+
+| Class | Beads | Type |
+|---|---|---|
+| Self-Healing Metrics on Regime Change | 1 | prediction-bearing trajectory (SASTID 28/30 → 29/30) |
+| Workflow Null-Result Audit Signal | 2 | detector reuse (B-0085 + B-0087 found) |
+| Scheduled Workflow Null-Result Hygiene Scan (tier-1) | 0 | named, not yet exercised post-promotion |
+| Chronological Insertion Polarity Error | 1 | mechanism-over-vigilance hook caught PR #684 chronological reversal |
+| Outdated Review-Thread Merge Gate Residue | 2 | applied PR #684 (Copilot stale thread) + PR #688/#690 (5 thread fixes) |
+| Blocked-GreenCI Review-Thread Punchlist | 1 | PRs #688/#690 deterministic 5-min unblock |
+| Advisory Enforcement Workflow Gap | 1 | sibling-lint audit: 4 of 4 advisory-only confirmed |
+| Incomplete Source-Set Regeneration Hazard | 1 | BACKLOG_WRITE_FORCE clobber caught + reverted before push |
+| Class-Naming Ferry Protocol + SD-9 | 0 | meta-class; no explicit validation event yet |
+| Class-Count Validity Drift | 1 | caught my "substrate compounds" framing this arc |
+| Prediction-Bearing Class Reuse | 0 | this memory; no validation event yet (the pattern names the act of validation) |
+
+The bead audit itself is one validation pass. Future-Otto can
+walk this list periodically and update the bead count; classes
+with zero beads after their named-applicability-window are
+retirement candidates.
+
+## What this is NOT
+
+- **NOT a directive to demand beads before encoding.** Local
+  factory-hygiene classes can be encoded with zero beads
+  initially; the bead system is for tracking validation
+  accumulation, not gating encoding.
+- **NOT a license to discount classes with few beads.** A
+  1-bead class IS still real signal; the precision-language
+  is about not overclaiming.
+- **NOT specific to Amara-named classes.** Any factory class
+  (Otto-named, Aaron-named, peer-named) accumulates beads via
+  the same four mechanisms.
+- **NOT a substitute for the SD-9 guardrail.** SD-9 still
+  applies: non-local claims need substrate evidence + external
+  lineage + falsifier even at 1+ beads.
+
+## Composes with
+
+- `memory/feedback_class_count_validity_drift_amara_meta_class_2026_04_28.md`
+  — the failure-mode complement; together they form the pair.
+- `memory/feedback_class_naming_ferry_protocol_with_sd9_guardrail_amara_2026_04_28.md`
+  — the encoding protocol; this pattern is the validation step
+  that closes the encoding loop.
+- `memory/feedback_self_healing_metrics_on_regime_change_factory_design_principle_aaron_2026_04_28.md`
+  — the worked-example class for this validation pattern.
+- `memory/feedback_speculation_leads_investigation_not_defines_root_cause_aaron_2026_04_28.md`
+  — same family: speculation needs evidence; class promotion
+  needs prediction-bearing beads.
+
+## Pickup notes for future-Otto
+
+When tempted to write "we named X classes this arc":
+
+1. Run the bead audit instead.
+2. Each class earns precision-language: "Y has Z beads via
+   mechanism W."
+3. Activity-count is not a tick-close metric; bead-count is.
+
+When a tick observation could become a class:
+
+1. Apply the 5-step control (worked example / mechanism /
+   control / scope / falsifier).
+2. If passes: encode at 0 beads; the class is "named" not
+   "validated."
+3. Watch for any of the four bead-earning mechanisms
+   (prediction, repair, detector reuse, falsifier survival).
+4. Update bead count when a mechanism fires.
+
+When a class accumulates 0 beads after its
+named-applicability-window:
+
+1. Retire the class, OR
+2. Rewrite it more narrowly to be exercisable, OR
+3. Document why no validation has fired (sometimes the class
+   is correct but applicability is rare).