"Reality is always right. Your model is always wrong if it conflicts with reality. That gap is where you learn."
Stocks accumulate (candidates, fish in a lake, incidents)
Flows are rates between stocks — find the real bottleneck
Build model first, then compare to reality — iterate fast
Danger: getting so lost in measurement you forget to act
Main Framework
Engineering Strategy: Rumelt's Three Parts + Will's Boring Constraint Rule
3
books published
~1K
blog posts written
16+
years writing online
Every company has a strategy — written or not. Unwritten = unimprovable.
Good strategy is boring: constrains choices to focus energy on what matters most
Uber: no cloud → launched in China in 3 months (crane-lifted racks through the roof)
Stripe: Ruby monolith → engineers built features for users, not internal tooling
Bad strategy = wrong diagnosis, usually from wilfully ignoring real constraints
Inert strategy = no one actually acts on it — the biggest failure mode
Will's key insight
The goal of strategy is not to make everyone happy. It's to dictate how limited capabilities get invested into the problems the company actually cares about.
Deep Dive
EM–PM Alignment, Values, and Measuring Engineering Velocity
The EM/PM Alignment Fix
Most conflict = incentive misalignment or not understanding the other's hidden constraints
Will's fix at Carta: EM + PM get the same performance rating — shared fate, shared incentives
Trifecta model: EM + PM + business lead, all graded on holistic team execution
Before solving conflict: understand their actual needs vs. what they're saying
Values framework (Will's 3 tests)Honest? Do you actually operate this way? Applicable? Can you use it to make real daily decisions? Reversible? Could an opposite value also be valid? If not — it's identity theater, not strategy.
Measuring Engineering Velocity
DORA metrics (Accelerate — Forsgren, Kim): lead time, deploy frequency, MTTR, change failure rate — useful for diagnosis, not judgment
Sprint points = "totally fake thing to report to a board" — but start somewhere imperfect
Best signal: just talk to engineers. They know if the team is effective and why it isn't.
Best board answer: show a roadmap of meaty, impactful things shipped in the last 6 months
Benchmark against peer companies for defensibility — not insight, but board comfort
The measurement paradox
Experts know every metric is imperfect so they measure nothing — and look clueless. Use imperfect metrics as an education tool: show them, then explain the nuances. That's how stakeholders get sophisticated.
Tactics
Write More, Stay in the Game Longer
Write only what gives you energy — stop writing what doesn't. That's why Will has ~1,000 posts.
Align writing with work: topics that sharpen your thinking AND help you perform in your role
Just want career credibility? Write 2–3 great things. Don't start a newsletter.
Want to write consistently? Publish everything. Hundreds of drafts = zero impact.
Biggest risk to any creator: quitting early from burnout. The long game always wins.
You're not competing with other writers — it's an infinite game. You help each other grow.
"Will anyone remember what we decided in six months? If no — just do something reasonable and move on to more important things."
The Digg V4 lesson
Complete rewrites never work. Will learned it the hard way — the site was down for a month. But he became an eng manager 2.5 years into his career because everyone else quit. Grim moments = fastest growth.
Contrarian
Engineering Leadership Myths Will Larson Wants You to Unlearn
✗Coddling engineers retains your best talentINSTEAD →✓ Sheltering engineers from hard problems is bad for them AND you. Accountability enables senior roles. The ZIRP era of coddling stunted engineering careers for a decade.
✗No written strategy = no strategyINSTEAD →✓ Every company always has a strategy — it's just unwritten. Unwritten strategies can't be improved or debugged. Write it down even if it's bad. That's step one.
✗Great engineering strategy is bold and ambitiousINSTEAD →✓ The best strategies are boring constraints: "use only the standard kit." Boring focus beats exciting sprawl. Engineers hate constraints — and constraints are exactly the point.
✗More systems thinking analysis = better decisionsINSTEAD →✓ Systems thinking without action is a trap. At Stripe, Will's incident team measured so deeply they forgot to fix things. Measure twice, cut once — not measure forever and never cut.