Learning Ledger — Priya S. · Mastery by Standard

1 · Mastery by standard (11 AIHC anchors touched, A1–D2)

† n/r = numeric not recorded — schema drift, flagged, not papered over. Lessons 01–06 write a ledger_append block with numeric mastery_before → mastery_after_simulated pairs; Lessons 07–12 switched to a ledger_write block whose score fields carry only the literal ` placeholder (and a drifted lesson-id convention:L3-07-B3vs.L3-cons-priya-01-diagnostic`). This compiler records what the lessons wrote and invents nothing — consistent with the playbook §12's own scope honesty that fully runnable lesson/ledger schemas are round-3 work. The gap is itself instrument output.

2 · Viz 1 — Mastery by anchor, start → end

Static unicode bars; scale: each ▓ = 0.05 mastery (0–1.0, 20 cells).

AIHC.1.A1 Specification    start 0.60 ▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░░
                           end   0.78 ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░  Δ +0.18

AIHC.1.A2 Delegation       start 0.35 ▓▓▓▓▓▓▓░░░░░░░░░░░░░
                           end   0.60 ▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░░  Δ +0.25

AIHC.1.A3 Elicitation      start 0.20 ▓▓▓▓░░░░░░░░░░░░░░░░
                           end   0.50 ▓▓▓▓▓▓▓▓▓▓░░░░░░░░░░  Δ +0.30

AIHC.1.B1 Verification     start 0.45 ▓▓▓▓▓▓▓▓▓░░░░░░░░░░░
                           end   0.68 ▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░  Δ +0.23

AIHC.1.B2 Failure lit.     start 0.30 ▓▓▓▓▓▓░░░░░░░░░░░░░░
                           end   0.65 ▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░  Δ +0.35

AIHC.1.C1 Provenance       start 0.25 ▓▓▓▓▓░░░░░░░░░░░░░░░
                           end    n/r ????????????????????  L08 taught it; no numeric written †

B3, C2, C3, D1, D2: no start or end numeric exists on file (†) — charting them would require inventing data.

3 · Viz 2 — Coverage: 11 anchors × completion

Anchor        A1   A2   A3   B1   B2   B3   C1   C2   C3   D1   D2
Lesson        L02  L03  L04  L05  L06  L07  L08  L09  L10  L11  L12
Diagnosed L01  █    █    █    █    █    ·    ·    ·    ·    ·    ·     5/11
Taught         █    █    █    █    █    █    █    █    █    █    █    11/11
Numeric SIM    █    █    █    █    █    ░    ░    ░    ░    ░    ░     5/11 †

█ = present  ░ = placeholder only (schema drift †)  · = out of that lesson's scope

Arc complete: every one of the 11 anchors has exactly one dedicated lesson (plus the five-anchor diagnostic); the numeric-record row shows where the instrument still leaks.

4 · Per-lesson record

Calibration-gap column: this is the construct the design measures — Playbook §8 names the auto-score vs. self-score gap a first-class metacognitive signal; the next lesson reads it (under-confidence → encourage, over-confidence → re-verify, per the Cell-29 worked trace). All values SIMULATED.

‡ L01–06's ledger_append schema has no calibration_gap field; L07–12's ledger_write schema added it (placeholder-valued). Unifying the two is the round-3 schema work.

Journal slots — ALL EMPTY by design

The learner's words are never generated for her (Playbook §8: auto-collection never displaces the learner's own journal entry and self-score). Prompts on file, entries blank:

• J01 — Of the five practices you just measured yourself on, which result surprised you most — and why is that the one you didn't already know about yourself?
• J02 — What's one ask on your pitch tracker or notes you know you'll never send vague again — and what fixed element did you almost forget to name?
• J03 — Which task did you catch yourself about to hand off blind — and which column (Claude-side, Priya-side, check-method) was hardest?
• J04 — What did asking "what would you cut" catch that re-reading the section yourself would have missed?
• J05 — Which claim almost shipped before you checked it — source mismatch, or a cutoff-risk you hadn't thought to name?
• J06 — What's the rule you're proudest of writing this week — and which of the five diagnostic practices needs a lesson like this next?
• J07 — Pick one claim you trusted as-is without checking a cutoff or citation. Would you still — and what changed your mind (or didn't)?
• J08 — How did you track what was yours vs. Claude's vs. a source's before this lesson? Name one place that habit would have failed you.
• J09 — Was your source-protection practice a habit or nothing formal? What does having it written down change for your next investigation?
• J10 — Was "this habit works" a feeling or something written down? What's the first thing your tracker told you that you didn't already know?
• J11 — How many of your chats sat outside the right project — and what would another month of that have cost you?
• J12 — Which habit from Lessons 07–11 would silently break the next time a model ships, if never re-tested? How will you catch it?

5 · What the tutor reads next (personalization compounds)

The next generated lesson reads THIS file, not a reset baseline. What it sees:

1. Repair C1's record first. The ledger's lowest number on file is still C1 = 0.25 (profile, pre-arc) — L08 taught provenance but wrote no numeric, so the record and the teaching have diverged. Next C1 touch: an Analyze-level at-bat, medium scaffold (she has L08's exposure; the record just can't prove it), writing a real before → after pair.
2. A3 is the lowest recorded post-arc score (0.50). Next A3 touch climbs Apply → Analyze with light-medium scaffold — the elicitation habit exists (L04, reused in L05/L07); what's untested is her judging which elicitation answer to act on.
3. Fade fastest on A1 (0.78) — her strongest anchor; next A1 contact should be an unscaffolded at-bat only, else expertise-reversal risk.
4. Backfill before new content for B3–D2: the exit tickets exist in L07–12; the ledger blocks don't carry their numerics. One scoring pass (or re-run) populates five anchors — otherwise personalization for the Judgment/Stewardship/Evolution strands resets instead of compounding, the exact failure §8 exists to prevent.
5. Standing cadence, not a lesson: L12's regeneration checklist hands off to the freshness beat (§10) — the ledger should re-surface it to her whenever the model/cutoff facts it depends on change.

Seed-chaining already demonstrated across the arc: L01's scaffold plan ("heaviest on A3, fade fastest on A1") became L02–04's actual scaffold levels; L05's cutoff-miss became L06's worked failure case; L09's confidentiality policy became the instruction layer L11 inherits; L07's moving-line habit became L12's trigger. That chain — each lesson consuming the previous ledger write — is the personalization-compounds claim, running.