Files

T

Antigravity Agent f8b21af4be Wiki cleanup: error-doc removal, dedup merge, link normalization

10_Wiki/Topics 대규모 정리:
- 오류 캡처/미완성 stub 문서 227개 제거
- 교차폴더 중복 43클러스터 병합 (63파일 → redirect)
- 링크명 정규화: 깨진 링크 수정·redirect 직결·개념 매핑 ~2,400건
- 카테고리 MOC 6개 신규 생성
- Graph 섹션 미해결 related-keyword 링크 10,058건 제거

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-20 23:52:15 +09:00

4.3 KiB

Raw Blame History

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

title

Pyramid Principle

매 한 줄

"매 conclusion first, supporting reasons next, evidence last". Barbara Minto 의 McKinsey (1967) 의 develop 의 top-down structure 의 consulting/exec communication 의 standard. 매 reader 의 cognitive load 의 minimize — main idea 의 immediately 의 deliver.

매 핵심

매 Three Rules

Ideas at any level must summarize the ideas grouped below.
Ideas in each grouping must be the same kind (MECE — Mutually Exclusive, Collectively Exhaustive).
Ideas in each grouping must be logically ordered (time / structure / degree).

매 SCQA Intro Pattern

Situation: 매 reader 의 known fact.
Complication: 매 disrupt 의 event.
Question: 매 implicit question.
Answer: 매 thesis (= pyramid top).

매 응용

Exec memo — 매 1-page 의 BLUF (Bottom Line Up Front).
Consulting deck — 매 action title slide 의 each.
PR description — 매 "what / why / how" 의 top-down.
RFC docs — 매 abstract 의 conclusion-first.

💻 패턴

Bad → Good (technical email)

<!-- BAD: Bottom-up -->
We profiled the API. CPU was 80%. Then we added caching.
After caching, p99 dropped from 800ms to 120ms.
Therefore we should ship the cache to prod.

<!-- GOOD: Pyramid -->
**Recommend shipping Redis cache to prod (p99 800ms → 120ms).**

- Profiling showed 80% CPU on repeated DB reads.
- Cache hit ratio 92% in staging.
- No correctness regressions in 48h soak test.

SCQA opener template

**S**: Our checkout API serves 10k req/s.
**C**: Last week, p99 spiked to 2s during flash sale.
**Q**: How do we prevent recurrence?
**A**: Add request-coalescing cache + circuit breaker. Details below.

Action-title slide structure

# [Slide title = the takeaway, not the topic]

# BAD: "Q3 Revenue Analysis"
# GOOD: "Q3 revenue grew 22% driven by enterprise tier"

  - Bullet 1 (evidence)
  - Bullet 2 (evidence)
  - Bullet 3 (evidence)

PR description pyramid

## TL;DR
Fix race in `OrderQueue` causing lost messages under high load.

## Why
- Reproduced 1/1000 message loss in load test.
- Root cause: missing mutex in `enqueue()`.

## How
- Added `sync.Mutex` (commit a1b2c3).
- New stress test (commit d4e5f6).

## Risk
Low — change is local; backward compatible.

MECE grouping check

# Pseudo: ensure subpoints partition the parent claim
parent = "User churn increased"
children = [
    "Onboarding friction",  # acquisition phase
    "Feature gaps",          # activation phase
    "Pricing",               # retention phase
]
# MECE: each is distinct phase, together cover funnel.

매 결정 기준

상황	Use pyramid?
Exec / time-poor reader	Yes — BLUF
Persuasive memo	Yes — SCQA
Narrative storytelling	No — suspense matters
Tutorial / step-by-step	No — sequence matters
Research paper abstract	Yes — conclusion-first

기본값: 매 business 의 technical writing 의 pyramid first.

🔗 Graph

변형: SCQA · MECE
응용: Technical-Writing
Adjacent: Mental_Models · Decision-Making

🤖 LLM 활용

언제: draft 의 restructure top-down, 매 generate exec summary. 언제 X: creative narrative — 매 pyramid 의 kill suspense.

❌ 안티패턴

Buried lede: 매 conclusion 의 last paragraph — exec 의 never reach.
Non-MECE buckets: 매 overlap 의 reader confuse.
Topic title slides: "Q3 Analysis" 의 X — "Q3 grew 22%" 의 O.
Pyramid 의 misuse 의 fiction: 매 mystery 의 spoil.

🧪 검증 / 중복

Verified (Minto, "The Pyramid Principle", 1987 / McKinsey internal training).
신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — Three rules + SCQA + before/after examples

4.3 KiB Raw Blame History