Education OS — Depth Repair Loop Specification

Version: v1.0 (Public Protocol)
Use When: D-FAIL dominant (Depth is lowest or explanation/rebuild fails)

Purpose

Depth repair rebuilds the skill’s internal architecture so the learner can construct it independently, explain it, and reproduce it without copying.

Depth repair is not “more practice.”
It is rebuilding the structure that practice depends on.

Closed-loop target:
After repair, the learner should pass D-01 (Explain + Rebuild) and show stable accuracy untimed.

Depth Failure Signature (D-FAIL)

Common signs:

  • “I memorised but don’t understand.”
  • Can do it when shown, cannot start alone.
  • Explains with vague words, not steps/logic.
  • Forgets quickly.
  • Makes basic conceptual errors.

Depth Repair Loop (Core Sequence)

Clarify → Structure → Micro-Retrieval → Feedback → Rebuild → Retest

Step 1: Clarify (5–10 minutes total per session)

Goal: remove confusion and define the core idea.

Actions:

  • Define the concept in one sentence (“This skill is about…”)
  • Identify the rule/method in plain words
  • Identify the “why” (what problem the method solves)

Pass indicator:

  • Learner can state the concept and purpose clearly.

Step 2: Structure (build the method map)

Goal: make the steps visible and ordered.

Actions:

  • Break the method into 3–6 steps
  • For each step, add: “what” + “why”
  • Highlight the most common misconception

Pass indicator:

  • Learner can list the steps in order and explain why each exists.

Step 3: Micro-Retrieval (no notes)

Goal: convert understanding into recall.

Actions:

  • Ask learner to recall the steps without looking
  • Do 2–3 mini examples
  • After each, ask: “Which step are you on? Why?”

Pass indicator:

  • Learner reproduces steps from memory with minimal prompting.

Step 4: Feedback (error classification, not just marking)

Goal: turn mistakes into repair instructions.

Actions:

  • For every error, classify it as one of:
  • Concept error (definition wrong)
  • Step error (sequence wrong)
  • Condition error (wrong rule applied)
  • Slip error (careless execution)

Pass indicator:

  • Learner can name what went wrong and how to correct it.

Step 5: Rebuild (independent production)

Goal: learner does the whole thing alone.

Actions:

  • One full question/task with no hints
  • Learner narrates steps aloud
  • Learner checks answer using their own checklist

Pass indicator:

  • Learner can complete independently and self-correct.

Step 6: Retest (closed loop)

Run D-01 again:

  • “Teach me how to do it from scratch.”

Pass threshold:

  • If D < 4, repeat Depth loop for another cycle.

7-Day Depth Repair Cycle (Daily 15–25 minutes)

Day 1: Clarify + Structure (build steps + why)
Day 2: Micro-retrieval (steps from memory) + 2 examples
Day 3: Micro-retrieval + error classification drill
Day 4: Independent rebuild task + checklist
Day 5: Mixed 3-question set untimed + explain each
Day 6: “Teach-back” (learner teaches you) + patch misconceptions
Day 7: Retest D-01 + add one mild variation (start Transfer-lite)

Expected movement:

  • D rises first, then L and T become trainable.

Upgrade Rule (after D stabilizes)

When D reaches 4:

  • begin adding Load scaffolding (short timed sets)
  • begin Transfer variations (new formats)

Depth is the foundation. Do not skip it.

Education OS — Load Repair Loop Specification

Version: v1.0 (Public Protocol)
Use When: L-FAIL dominant (collapse under time/pressure, “careless mistakes”, inconsistent)

Purpose

Load repair stabilizes performance under constraints.

Load failure is not “carelessness.”
It is insufficient automation + poor pressure stability.

Closed-loop target:
After repair, learner should pass L-01 with stable accuracy under time.

Load Failure Signature (L-FAIL)

Common signs:

  • “Good at home, bad in exam.”
  • Panic, blanking, rushing.
  • Big accuracy drop when timed.
  • Careless errors spike under speed.
  • Inconsistent performance across days.

Load Repair Loop (Core Sequence)

Stabilize → Automate → Time-Scaffold → Stress-Inoculate → Retest

Step 1: Stabilize (remove chaos)

Goal: remove unnecessary difficulty so stability can form.

Actions:

  • Use easier question set (same skill, lower complexity)
  • Set a calm rule: accuracy before speed
  • Establish a fixed routine (same time daily)

Pass indicator:

  • Accuracy returns when untimed.

Step 2: Automate (fluency building)

Goal: reduce cognitive load per step.

Actions:

  • Repetition of the core move (the smallest unit)
  • Short bursts: 2–5 minutes per burst
  • Focus on smooth, correct execution

Pass indicator:

  • Learner performs the core move quickly without thinking hard.

Step 3: Time-Scaffold (controlled timing)

Goal: introduce time pressure gradually.

Actions:

  • Start with generous time limits (success first)
  • Reduce time slowly across days
  • Keep accuracy target fixed

Example scaffold:

  • 5 questions / 10 minutes (day 1–2)
  • 5 questions / 8 minutes (day 3–4)
  • 5 questions / 6 minutes (day 5–6)
  • 5 questions / 5 minutes (day 7)

Pass indicator:

  • Accuracy stays within an acceptable band as time reduces.

Step 4: Stress-Inoculate (exam realism)

Goal: train stability in realistic conditions.

Actions:

  • One short “exam sprint” daily
  • No pausing, no hints
  • After sprint: analyze errors by type

Error types under load:

  • Rush error
  • Misread
  • Sequence skip
  • Arithmetic slip
  • Attention drift

Pass indicator:

  • Errors decrease week to week; confidence improves.

Step 5: Retest (closed loop)

Run L-01 timed micro-set.

Pass threshold:

  • L < 4 → repeat another Load cycle.

7-Day Load Repair Cycle (Daily 15–25 minutes)

Day 1: Stabilize + automate 1 core move
Day 2: Automate + gentle timed scaffold
Day 3: Timed scaffold + error type tracking
Day 4: Timed scaffold + one exam sprint
Day 5: Faster scaffold + exam sprint + review
Day 6: Mixed mini-set under time + review
Day 7: Retest L-01 (same format) + compare drop rate

Expected movement:

  • “Careless mistakes” reduce when stability returns.

Upgrade Rule (after L stabilizes)

When L reaches 4:

  • increase complexity gradually
  • add Transfer variants under time
  • build endurance (longer sets) slowly

Load stability is what makes exam performance reliable.

Education OS — Transfer Repair Loop Specification

Version: v1.0 (Public Protocol)
Use When: T-FAIL dominant (can do familiar forms, breaks on new formats/context)

Purpose

Transfer repair makes learning portable.

Transfer failure is not “low intelligence.”
It is overfitting to one pattern instead of learning the underlying structure.

Closed-loop target:
After repair, learner should pass T-01 and handle new formats with stability.

Transfer Failure Signature (T-FAIL)

Common signs:

  • “Worksheet okay, exam fails.”
  • “I don’t know because it looks different.”
  • Depends on memorised templates.
  • Cannot identify the underlying concept.
  • Breaks when context changes.

Transfer Repair Loop (Core Sequence)

Extract the invariant → Train variations → Mix formats → Explain mapping → Retest

Step 1: Extract the Invariant (what stays the same)

Goal: identify the underlying concept beneath changing appearances.

Actions:

  • Ask: “What concept is this really testing?”
  • Write the invariant in one line
  • Create a mini checklist: “If you see X, do Y”

Pass indicator:

  • Learner can name the underlying concept even when format changes.

Step 2: Variation Ladder (near → far transfer)

Goal: train adaptability gradually.

Actions:

  • Near transfer: small changes (numbers, wording)
  • Mid transfer: new format but same concept
  • Far transfer: multi-step, unfamiliar context

Pass indicator:

  • Learner survives variations without freezing.

Step 3: Mixed Practice (stop pattern dependence)

Goal: prevent guessing by appearance.

Actions:

  • Mix question types
  • Randomize order
  • Force concept identification before solving

Rule:

  • Learner must say: “This is a _ concept” before doing it.

Pass indicator:

  • Learner selects method correctly without relying on visual similarity.

Step 4: Explain Mapping (teach the adaptation)

Goal: make transfer conscious and repeatable.

Actions:

  • Learner explains:
  • what changed
  • what stayed the same
  • why the same method still applies (or how it adjusts)

Pass indicator:

  • Learner can verbalize adaptation logic.

Step 5: Retest (closed loop)

Run T-01 with a new-format variant.

Pass threshold:

  • T < 4 → repeat Transfer cycle.

7-Day Transfer Repair Cycle (Daily 15–25 minutes)

Day 1: Extract invariant + near transfer (3 variations)
Day 2: Near → mid transfer ladder
Day 3: Mixed set (identify concept first)
Day 4: Mid transfer + explain mapping
Day 5: Far transfer attempt + guided review
Day 6: Mixed set under gentle time
Day 7: Retest T-01 with a fresh variant + compare adaptation confidence

Expected movement:

  • “Looks different” stops being a blocker.

Upgrade Rule (after T stabilizes)

When T reaches 4:

  • add Load pressure (timed transfer)
  • increase variation complexity
  • mix across chapters/topics

Transfer is what makes learning future-proof.

Education OS — 7-Day Repair Cycle Template (Universal Wrapper)

Version: v1.0
Use When: you have any D/L/T dominant failure and want a standard weekly cycle.

Purpose

This wrapper makes all repairs closed-loop and comparable week to week.

Day 0 (Before starting)

Run D-01, L-01, T-01 and record scores.

Days 1–6

Run the dominant repair loop daily:

  • 15–25 minutes
  • small, consistent
  • accuracy first, then speed, then variation

Day 7 (Retest Day)

Run D-01, L-01, T-01 again and compare.

Interpretation:

  • If dominant axis improved to 4: upgrade difficulty or move to next lowest axis
  • If it improved but remains 3: run another 7-day cycle
  • If it did not improve: your probe might be wrong target skill, or feedback is missing — reset and narrow the target

Education OS — Quick Routing Map (One Screen Summary)

  1. Run probes: D-01, L-01, T-01
  2. Find lowest axis (dominant failure)
  3. Run its repair loop for 7 days
  4. Retest
  5. Upgrade and repeat

Depth builds the machine.
Load stabilizes the machine.
Transfer makes the machine portable.

If you want, next I can write a “DLT Probe Set for Primary / Secondary / Adults” (same plain-text spec style), so your system has ready-made probe examples by level and context.