Summary of "An Elegant Puzzle"

Core Idea

Management is a learnable system, not innate talent — rooted in organizational design, clear structures, and intentional practices
Your job: create conditions where teams thrive through right-sizing, clear direction, fair processes, and continuous learning
Build for durability, not heroics — sustainable organizations scale; individual brilliance doesn't

Size teams at 6–8 engineers per manager; managers-of-managers oversee 4–6 managers
Never go below 4 people per team — small teams leak individual complexity instead of abstracting it
Move teams through four states: Falling Behind → Treading Water → Repaying Debt → Innovating; apply the right system fix for each, not more people
Preserve high-performing teams — regelling costs outweigh efficiency gains
Shift scope or rotate individuals for growth, don't shuffle teams to optimize globally

Write one clear vision (1–2 pages): aspirational, describes future capabilities, links to business value
Write strategies for hard problems using: diagnosis → policies → actions framework
Define goals with four components: baseline, target, trend, timeframe (e.g., reduce p95 latency from 600ms to 300ms by Q3)
Create a visible dashboard with directional metrics: current value, goal, trend
Use metrics to guide, not control — attribution, benchmarking, and nudges beat mandates

Three phases: De-risk (design docs, embedded teams) → Enable (programmatic tools automate 90%) → Finish (deprecate legacy system at 100% adoption)
Stop the bleeding first — enforce new approach via linters and documentation for all new code
Celebrate completion, not startup — unfinished migrations compound debt
Track, push status to teams, escalate only when needed

Design loops from first principles: identify critical skills, write rubrics, test real-world scenarios (pair programming, debugging, presentations)
Run humane processes — be kind, be prepared, test actual skills
Use cold sourcing (LinkedIn outreach) to scale beyond referrals
Extend the hiring funnel to include onboarding, impact, promotion, retention — hiring is a lifecycle

Write career ladders first — describe behaviors and expectations at each level
Run consistent performance cycles (annual or biannual); calibrate across teams for fairness
Address poor performance immediately, not during reviews
Watch for designation momentum — set explicit goals for high performers stuck at same level

Opportunity: visible career paths, structured project selection, clear rubrics, tracked retention and diversity
Belonging: recurring events, ERGs, offsites, random coffee chats, psychological safety
Balance positive freedom (autonomy to act) with negative freedom (protection from constraints) — adjust based on team stability

Management is ethical work — your decisions shape compensation, growth, and dignity
Relationships trump processes — most problems trace to broken relationships, not bad systems
Do hard things now — don't defer poor relationships, performance issues, or misalignment
Decide trade-offs in order: Company > Team > Self

Map your organization: Size teams, identify weak structures, spot where you micromanage or abdicate
Write one strategy for your biggest challenge using the diagnosis → policies → actions framework
Define two metrics: one baseline (stability), one investment (growth); make visible
Document your hiring process: role requirements, 3–4 interview slots with clear rubrics, add cold sourcing
Schedule career conversations: Ask each report what they want to learn in 6 months; write it down together