2nd/10_Wiki/Topic_Programming/Frontend/Levels of Understanding.md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-levels-of-understanding	Levels of Understanding	10_Wiki/Topics	verified	self	Knowledge Hierarchy Bloom's Taxonomy for Code	none	A	0.85	applied	learning mental-model pedagogy knowledge		2026-05-10	pending	language framework meta cognitive

Levels of Understanding

매 한 줄

"매 knowledge 의 매 layered model — 매 surface mimicry 매 deep generative comprehension 까지". Bloom's Taxonomy (1956) 와 Feynman Technique 의 매 영향, 매 software engineering 매 context 매 "can copy" → "can use" → "can debug" → "can teach" → "can extend" 매 progression.

매 핵심

매 5 levels (engineering 적용)

1. Recognition: 매 코드/concept 매 본 적 있음. 매 시 read-only.
2. Reproduction: 매 example 매 따라 작성 가능. 매 copy-modify.
3. Application: 매 새 context 매 동일 pattern 적용. 매 transfer.
4. Debugging: 매 broken case 매 root cause 까지 trace. 매 inverse reasoning.
5. Generation: 매 from-scratch design + teach others. 매 generative model 보유.

매 signals

• Recognition: "본 적 있어" — but cannot describe.
• Reproduction: tutorial 따라 동작하는 결과물 — but breaks on변형.
• Application: 매 다른 problem 매 동일 solution 매 적용 — but mechanism 모름.
• Debugging: error message 의 매 원인 가설 + 매 fix path 제안.
• Generation: 매 design decision 매 trade-off 매 articulate + 매 alternative 매 비교.

매 응용

1. Self-assessment: 매 새 tech 매 학습 매 어느 level 인지 정직히 평가.
2. Hiring: interview 매 level 별 question 설계 (recognition vs generation).
3. Onboarding: 매 신규 팀원 매 level 별 task 배정.
4. Documentation 작성: target level 명시 (tutorial=reproduction, ADR=generation).

💻 패턴

Self-test 매 Feynman Technique

1. 주제 선택 (e.g., "React Suspense")
2. 매 8세 아이에게 매 설명 매 작성 (no jargon)
3. 매 막힌 지점 = 매 understanding gap
4. Source 매 다시 학습
5. Repeat
→ 매 step 4 매 도달하면 매 generation level

Code review 매 level probe

Reviewer: "왜 useMemo 사용?"
- L1 답: "튜토리얼에서 봐서"
- L2 답: "성능 최적화"
- L3 답: "expensive computation 결과 캐시"
- L4 답: "render마다 reference 변하면 child re-render — useMemo 매 stable reference"
- L5 답: "이 case 는 child memo 안 했으니 useMemo 무용지물 — 제거 권장"

Bloom's mapping (engineering)

Remember     → Recognition  (L1)
Understand   → Reproduction (L2)
Apply        → Application  (L3)
Analyze      → Debugging    (L4)
Evaluate     → Debugging    (L4)
Create       → Generation   (L5)

Dreyfus 모델 비교

Novice         ≈ L1 Recognition
Advanced Beg   ≈ L2 Reproduction
Competent      ≈ L3 Application
Proficient     ≈ L4 Debugging
Expert         ≈ L5 Generation

매 progression 매 actionable steps

L1 → L2: Tutorial 매 따라 매 직접 type (copy-paste 금지)
L2 → L3: 매 다른 problem 매 적용 (변형 challenge)
L3 → L4: Bug 매 fix (intentional break + restore)
L4 → L5: Source code 읽기 + 매 design doc 작성

Learning log template

## <Topic>: <Date>
- Level before: L?
- 매 학습 source: <link>
- 매 핵심 insight: <one sentence>
- 매 confused 지점: <gap>
- Level after: L?
- Next: <action to advance>

매 결정 기준

상황	Required Level
Use library in side project	L2 Reproduction
Use in production	L3 Application
Own subsystem in production	L4 Debugging
Lead architecture / teach team	L5 Generation
Write spec / RFC	L5 Generation
Code review approver	L4+

기본값: 매 production code 매 책임지려면 매 L3 minimum, 매 L4 권장.

🔗 Graph

• 부모: Mental_Models
• 응용: Code-Review · Onboarding
• Adjacent: Deliberate-Practice

🤖 LLM 활용

언제: 매 self-assessment, 매 학습 priorities 결정, 매 team capability mapping, 매 documentation target audience 명시. 언제 X: 매 trivial task (level distinction overkill), 매 emotional/soft skills (different framework).

❌ 안티패턴

• L2 → "할 줄 안다" 자칭: 매 Dunning-Kruger — tutorial 통과 ≠ application.
• L1 결과물 매 production 배포: 매 Stack Overflow copy-paste — 매 silent failure 보장.
• Level 매 fixed trait 가정: 매 매 domain 별 level 다름 — React L5 + Rust L1 매 가능.
• Recognition 매 understanding 혼동: 매 "들어봤다" ≠ "이해한다".
• Generation 매 가정 매 검증 X: 매 자칭 L5 매 실제 L3 매 case 흔함 — Feynman test.

🧪 검증 / 중복

• Verified (Bloom 1956, Dreyfus 1980, Feynman lectures).
• 신뢰도 A (cognitive science consensus).

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — 5-level engineering taxonomy with progression patterns