Thoughtbox

export const FIVE_WHYS_CONTENT = `# Five Whys Iteratively ask "Why?" to drill down from symptoms to root cause. ## When to Use - A problem keeps recurring after fixes - You're treating symptoms not causes - Need to understand failure chains - Post-incident analysis ## Process ### Step 1: State the Problem Clearly Start with a specific, observable problem. - Bad: "The system is slow" - Good: "API response time increased from 200ms to 2s at 3pm yesterday" ### Step 2: Ask Why (First Level) Why did this happen? State the immediate cause. - "Why did response time increase?" → "The database queries started timing out" ### Step 3: Ask Why Again (Second Level) Why did THAT happen? - "Why did queries start timing out?" → "The connection pool was exhausted" ### Step 4: Continue (Third through Fifth Levels) Keep asking until you reach something actionable at a systemic level. - "Why was the pool exhausted?" → "Connections weren't being released" - "Why weren't they released?" → "Error handling didn't close connections on exception" - "Why not?" → "No code review checklist item for resource cleanup" ### Step 5: Verify the Chain Read the chain backward: does fixing the root cause prevent all the downstream effects? ## Key Principle Stop when you reach something you can: - Change systemically (process, architecture, policy) - Prevent from recurring - Actually control ## Signals You've Gone Too Far - "Because humans make mistakes" (too generic) - "Because physics" (can't change) - "Because requirements changed" (may be valid, but ask if process could handle change better) ## Signals You Haven't Gone Far Enough - Fix would only address this instance - Same type of failure could easily recur - You're fixing what broke, not why it was breakable ## Example Application **Problem:** "Customer received wrong order" 1. Why? → Warehouse shipped wrong item 2. Why? → Picker grabbed from wrong bin 3. Why? → Bins not clearly labeled 4. Why? → Labels fade and aren't replaced 5. Why? → No scheduled label maintenance **Root cause fix:** Implement quarterly bin label audit **Symptom fix (insufficient):** Re-train that one picker ## Common Mistakes - Stopping at the first satisfying answer - Following only one branch (problems often have multiple causes) - Asking "who" instead of "why" (blame vs understanding) - Not verifying the causal chain backward `;