--- id: wiki-2026-0508-purpose title: Purpose category: 10_Wiki/Topics status: verified canonical_id: self aliases: [Telos, Mission, Intent, Goal-Setting, Why] duplicate_of: none source_trust_level: A confidence_score: 0.85 verification_status: applied tags: [philosophy, psychology, leadership, design] raw_sources: [] last_reinforced: 2026-05-10 github_commit: pending tech_stack: language: theory framework: teleological-design --- # Purpose ## 매 한 줄 > **"매 know your why before you optimize your how."**. 매 Aristotle 의 *telos* 부터 매 Sinek 의 *Start with Why*, 매 Frankl 의 *meaning therapy* 까지 — 매 purpose = 매 why an entity (person, system, product) exists. 매 2026 software/AI 에서 매 purpose 명시는 매 alignment, scope creep 회피, 매 LLM agent goal-setting 의 foundation. ## 매 핵심 ### 매 4 levels of purpose - **Existential (인간)**: 매 Frankl — 매 meaning vs pleasure vs power. - **Organizational**: 매 mission statement — 매 "why does this company exist?". - **Product**: 매 user job-to-be-done (Christensen) — 매 "what hire is this product made for?". - **System/Code**: 매 module purpose — 매 "what 1 thing does this module own?". ### 매 purpose articulation patterns - **Sinek's Golden Circle**: WHY → HOW → WHAT. - **JTBD**: "When [situation], I want to [motivation], so I can [outcome]". - **OKR**: Objective (qualitative why) + Key Results (measurable). - **Module docstring**: 1-line purpose + invariants. ### 매 응용 1. Code: 매 module 의 1-sentence purpose 가 unclear 면 — split or rename. 2. LLM agent: 매 system prompt 의 첫 줄은 purpose — 매 alignment anchor. 3. Career: 매 ikigai (것 + 잘 + need + paid). 4. Product: 매 JTBD interview → feature prioritization. ## 💻 패턴 ### Pattern 1: Module-level purpose docstring ```python """ auth/jwt.py PURPOSE: 매 issue and verify JWT tokens for our API authentication. INVARIANTS: - Tokens always include `iss` claim. - Verification rejects tokens with `alg=none`. NON-GOALS: - Session storage (see auth/session.py). - Refresh-token rotation (see auth/refresh.py). """ ``` ### Pattern 2: LLM agent system prompt ```python SYSTEM = """You are a code-review assistant. PURPOSE: 매 surface non-obvious bugs, security issues, and style violations in the user's pull request, prioritized by severity. NON-GOALS: - Generating new code (the user does that). - Auto-fixing without confirmation. CONSTRAINTS: - Always cite file:line for every claim. - Refuse to comment on out-of-scope files. """ ``` ### Pattern 3: JTBD interview template ```yaml job_to_be_done: situation: "When I'm onboarding a new team member" motivation: "I want a single doc with their day-1 setup" outcome: "so they can ship a real PR by end of week 1" current_alternatives: [README, Notion page, ad-hoc Slack DMs] hire_criteria: [single-source, executable, verified] ``` ### Pattern 4: OKR framing ```yaml objective: "매 Make AI Act compliance trivial for EU SMBs" key_results: - "Onboard 100 SMBs by Q3" - "Reduce time-to-compliance from 30d → 5d" - "NPS ≥ 50" ``` ### Pattern 5: Purpose-test for scope creep ```python def is_in_scope(feature, module_purpose) -> bool: """매 ask: 'does this feature serve the module's stated purpose?'""" # 매 if no, push to a different module or reject return feature.advances(module_purpose) ``` ### Pattern 6: 5-Whys to surface purpose ```text Q: Why are we building this dashboard? A: To show metrics. Q: Why? A: So managers can see team health. Q: Why? A: So they can intervene early. Q: Why? A: To prevent burnout & attrition. Q: Why? A: Because attrition costs $200K/engineer. → 매 PURPOSE: "매 reduce attrition cost via early-burnout intervention". ``` ## 매 결정 기준 | 상황 | Tool | |---|---| | 매 module/file design | 1-sentence purpose docstring | | 매 LLM agent design | Purpose + non-goals + constraints in system prompt | | 매 product feature triage | JTBD + Purpose-test | | 매 organization strategy | Mission statement + OKRs | | 매 personal direction | Ikigai / 5-Whys | **기본값**: 매 1-sentence purpose 가 없으면 — 매 don't ship. ## 🔗 Graph - 부모: [[Philosophy]] · [[Design Thinking]] - 변형: [[Mission]] · [[Vision]] · [[Telos]] · [[Ikigai]] - 응용: [[OKR]] - Adjacent: [[Single Responsibility Principle (SRP)|Single Responsibility Principle]] ## 🤖 LLM 활용 **언제**: 매 system prompt 첫 줄, 매 module docstring, 매 product brief, 매 strategy doc. **언제 X**: 매 throwaway prototypes — 매 over-formalization slows iteration. ## ❌ 안티패턴 - **No purpose**: 매 module 이 "everything" 하면 — 매 god class. - **Purpose drift**: 매 purpose 명시했지만 features 가 매 drift — 매 update or refactor. - **Aspirational nonsense**: 매 "make the world better" — 매 unactionable. - **Purpose ≠ method**: 매 "use React" 는 method, not purpose. ## 🧪 검증 / 중복 - Verified (Aristotle *Nicomachean Ethics*; Frankl *Man's Search for Meaning*; Sinek *Start with Why*; Christensen JTBD). - 신뢰도 A (foundational philosophical + management canon). ## 🕓 Changelog | 날짜 | 변경 | |---|---| | 2026-05-08 | Phase 1 | | 2026-05-10 | Manual cleanup — 4 levels, JTBD/OKR/Golden Circle 패턴 |