LLenny's Podcast
← All episodes
Will Larson (Carta, Stripe, Uber, Calm, Digg)07 January 2024

The engineering mindset

6Frameworks
14Insights

Frameworks in this episode

Communication6 steps

Energy-Aligned Writing

Write only what energises you, only what overlaps your job, and publish nearly all of it.

Read →
Productivity5 steps

Imperfect Metrics as Education

Report an imperfect engineering metric on purpose, then use every question about it to educate upward.

Read →
Strategy5 steps

Stocks-and-Flows Pipeline Diagnosis

Model any org process as stocks and flows, then find where reality disagrees with the model.

Read →
Strategy5 steps

The Boring Engineering Strategy

Write the strategy down, get the diagnosis honest, then constrain options so energy goes to what matters.

Read →
Leadership4 steps

The Honest-Applicable-Reversible Values Test

Three tests that separate values which drive decisions from values that just describe your self-image.

Read →
Leadership5 steps

The Shared-Fate EM/PM Pair

Give the eng manager and PM the same performance rating, and make them understand before they argue.

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster16:30

You Already Have a Strategy, It's Just Not Written Down

Will pushes back on the common complaint that a company or team has 'no strategy.' Product, engineering, and business always have a strategy, it's just often unwritten and applied inconsistently. The first rule of strategy is that writing it down is what lets you debug and improve it.

  • The absence of a filled-in template doesn't mean the absence of a strategy.
  • There's always a strategy, it's just sometimes bad and often unwritten.
  • Written strategy lets you diagnose whether a PM is bad, misunderstanding it, or correctly applying a strategy that's wrong for the problem.
  • Debugging a mediocre written strategy beats arguing over an invisible one.

“there's always a product strategy sometimes it's bad but there's always one”

Will Larson · 18:00

“the the first rule of strategy is that if you write it down then you can like improve it right”

Will Larson · 18:00
#strategy#product#engineering-strategy#leadership

Hot Take· 4

Hot Take06:30

Stop Coddling Engineers, Treat Them as Adults

Will argues the industry treats engineers like children, sheltering them from the important, hard problems, which is actually bad for their growth. In the previous era, obsessing over retention led managers to coddle engineers, which was bad for the engineers, the teams, and the org. He sees a healthier shift toward holding engineers accountable, which is exactly what lets you put them into genuinely senior roles.

  • Sheltering engineers from what's important stunts their growth as adults.
  • Retention-obsessed management led to coddling, which is bad for efficiency and for the engineers themselves.
  • You can only put engineers in senior roles if you're comfortable holding them accountable.
  • Accountability is how senior engineers finally get the responsibility they've wanted.

“I think that we often treat Engineers a little bit like children instead of giving them like the responsibilities and ability to actually Thrive as…”

Will Larson · 06:30

“get to like give Engineers real hard problems and we get to we get to actually hold them accountable and that means we can put…”

Will Larson · 08:00
#engineering-culture#career#accountability#leadership
Hot Take19:00

Good Strategy Is So Boring It's Hard to Talk About

Will's contrarian point is that the best engineering strategies are boring and constraining, like 'we only use the tools we have today.' Engineers hate these because they feel like a loss of control, but the power of a boring strategy is that it focuses everyone's energy on the problems the company actually values.

  • A great strategy for almost all companies is to just use the standard kit you already have.
  • Boring strategies frustrate engineers who want new languages, databases, or cloud providers.
  • The value is focus: energy goes to what the company cares about, not tooling sprawl.
  • Alignment can be painful up front but pays off over time.

“a really good strategy for almost all companies is like we just use the standard kit we already have today”

Will Larson · 19:30
#strategy#engineering-strategy#constraints#focus
Hot Take32:30

The Biggest Risk to Writers Is Quitting, Not Starting Late

Will rejects the 'it's too late to start' fear about Substack, podcasts, or Medium. The real risk isn't growing too slowly or missing the wave, it's quitting early from burnout. If you keep writing good stuff, you build an audience over time and can carry it across platforms, treating content as an infinite game rather than a competition.

  • Everyone converges on the same hot topic (crypto, then AI), making it hard to say anything novel.
  • The biggest risk is burning out and quitting, not starting too late or growing slowly.
  • Keep writing good work and the audience compounds and travels with you across platforms.
  • It's an infinite game: other creators aren't competition, there's no cap on good thinkers.

“the biggest risk to content creation of any sort is quitting soon because you get burned out”

Will Larson · 32:30

“if you just like keep writing good stuff you'll build an audience over time and you can take that audience from platform to platform”

Will Larson · 33:00
#writing#content-creation#audience#longevity
Hot Take44:00

Give the EM and PM the Same Performance Rating

To fix misaligned incentives between engineering managers and product managers, Will forces the idea that an EM/PM pair is a pair and generally gets the same performance rating, calibrated together in the same meeting. Carta extends this to 'trifectas' that include business leadership, so everyone is graded on solving the entire set of constraints, not just their own function.

  • EM/PM pairs generally get the same rating, with exceptions only for clear individual non-performance.
  • The two are discussed and calibrated in the same meeting.
  • Carta's CEO wrote about 'trifectas' extending shared grading to business leadership.
  • Everyone is graded on solving for the entire set of constraints, not just their functional ones.

“empm pairs are Pairs and they generally have the same performance rating”

Will Larson · 44:00

“you all get graded the same score based on your ability to um evaluate and solve for the entire set of constraints not just your…”

Will Larson · 45:00
#management#incentives#product#performance-reviews

Explainer· 3

Explainer04:30

How the Market Shift Rewired What Great Eng Leaders Do

Will describes how the move from the zero-interest-rate era to today changed the engineering job. In the boom, a top engineering director could excel just by hiring well; now that hiring has collapsed to a couple of interviews a month, that same person can look like a low performer unless they lead the team, get into the details, and get right-sizing decisions correct. Teams now get cut, consolidated, and disappeared, which the industry avoided for a decade.

  • During the last decade, managers spent half their time or more hiring; now it can be two interviews a month or zero.
  • A director who was a top performer purely for hiring well can now be perceived as a low performer.
  • New critical competencies: leading the team, getting deeper into the details, and correct team sizing/allocation.
  • Teams disappearing, getting cut, and being consolidated is now a core part of the job.

“teams are actually disappearing teams are getting cut down teams are getting Consolidated”

Will Larson · 06:00
#engineering-leadership#layoffs#hiring#management
Explainer48:30

How to Actually Measure Engineering Productivity

Will explains that funding-based benchmarking gives a defensible answer that calms your board but doesn't help you run the org. The real signal comes from talking to engineers, who know if their teams are effective, and from showing a roadmap of meaningful things shipped in the last six months. DORA/Accelerate metrics are excellent for diagnosis but shouldn't be used to judge whether a team is good, and refusing to measure anything just cedes ground to non-experts.

  • Funding-based benchmarks are mechanical and mainly reduce board anger, not improve the org.
  • Engineers themselves will tell you if their teams are effective and often why not.
  • Showing a roadmap of meaty, high-impact work earns you space; an empty list rightly raises concern.
  • DORA metrics (from Accelerate) are diagnosis tools, not verdicts; a slow deploy tells you where to invest, not that you're a bad company.
  • Measuring something imperfect and educating people on its limits beats measuring nothing.

“if you're a good leader and you talk to the engineers they will tell you like the engineers know if their teams are effective or…”

Will Larson · 50:00

“metrics are about educating the people consuming the metrics about the reality of the rich data underneath”

Will Larson · 54:30
#engineering-productivity#metrics#dora#measurement
Explainer56:00

Good Company Values Are Honest, Applicable, and Reversible

Will lays out three tests for a useful value. It must be honest (you actually do what you claim), applicable (you can use it to make a real decision, like Stripe's 'optimize globally' vs Uber's implicit 'do what's good for your team'), and reversible (a sane company could hold the opposite). Values that fail reversibility, like 'we build good software' or 'Integrity,' are just identity values that feel good but can't guide any decision.

  • Copying another company's values (e.g. writing Facebook's on your own walls) doesn't work.
  • Honest: your decisions must match the value, or it undermines trust.
  • Applicable: Stripe's 'optimize globally' vs Uber's implicit 'optimize for your team' both guide real choices.
  • Reversible: if no sane company would hold the opposite (e.g. Integrity), it's a useless identity value.
  • Cut values down to about four that people can remember, and only keep ones that some group doesn't fit.

“these are the three that I found really useful for any value like is it honest is it a applicable and can you reverse it”

Will Larson · 1:00:00

“Integrity like everyone has that nobody wouldn't want Integrity”

Will Larson · 1:01:30
#company-values#culture#leadership#decision-making

Story· 3

Story09:00

The Systems-Thinking Trap: Measuring Forever, Never Fixing

Will tells the story of Stripe's incident management work, where the team got so deep into analyzing incidents that they lost track of whether anything was actually improving. He admits he was the one caught in the model, prioritizing measurement over improvement. His broader lesson: when your model conflicts with reality, reality is never the thing that's wrong.

  • At Stripe, incident analysis became so absorbing the team optimized for measurement, not improvement.
  • The mistake was his own, not the team's, showing how seductive a clean model is.
  • "Measure twice, cut once" doesn't mean measure infinitely and never cut.
  • When model and reality conflict, the model is wrong, and that gap is where the real learning is.

“we weren't actually prioritizing improvements we were just prioritizing measurement”

Will Larson · 10:00

“reality is never wrong reality is always right your model is always wrong if it's in conflict with with reality”

Will Larson · 10:30
#systems-thinking#incident-management#stripe#decision-making
Story22:00

Uber's No-Cloud Rule and Stripe's Ruby Monolith

Will gives two concrete strategy examples. Uber's 2014-era 'no cloud, own data centers' policy was annoying and forced them to reinvent everything, but it let them spin up in China in three months and move in and out of geopolitical constraints that cloud-dependent companies couldn't. Stripe's 'we run a Ruby monolith' policy focused engineers on user features instead of multi-language tooling. In both, lots of engineers hated the constraint, but that's not the point.

  • Uber ran its own data centers and could stand up in China in ~3 months (even craning racks in through the roof).
  • Cloud-dependent competitors are constrained by where AWS/Google/Azure have built out.
  • Stripe's Ruby monolith kept engineers building features rather than supporting many languages.
  • The goal of good strategy is to direct limited capacity at the problems you care about, not to appease everyone.

“at Uber we only used our own data centers we didn't use the cloud”

Will Larson · 22:00

“the goal of good strategy is not to um appease everyone the goal of good strategy is to dictate how we invest The Limited capacities…”

Will Larson · 24:00
#strategy#uber#stripe#infrastructure
Story1:02:00

The Digg v4 Rewrite That Went to Zero

Will's failure story is Digg's v4 complete rewrite, a decision made to add social functionality that he says never works out for anyone. On launch day there was champagne and sushi, but the site kept crashing; it took a month to get fully functional, with servers needing a restart every 12 hours until he finally traced the core bug to how Python initializes default parameter variables. The team pulled off a heroic recovery, but Digg still went to zero and sold for parts.

  • A full rewrite to add social features was a decision made ~2.5 years before he joined; it 'never works out for anyone.'
  • Launch day had caterers, champagne, and sushi, but the site wouldn't stay up.
  • Servers had to be restarted every 12 hours even with a hastily written caching system.
  • The core bug was Python's handling of variables used as default parameters, written by someone new to Python.
  • Despite a heroic recovery, layoffs cut the team from ~100 to ~30 and Digg eventually sold for parts.

“a complete rewrite in order to get there this is a decision that never works out for anyone”

Will Larson · 1:03:30

“related to the way that python initiates variables used as default parameters”

Will Larson · 1:06:00
#failure#digg#rewrites#engineering-war-story

Tool· 1

Tool25:00

Will Larson's Reading List for Getting Better at Strategy

Asked how to actually get better at strategy, Will recommends a specific stack of books. Good Strategy Bad Strategy by Richard Rumelt is the starting point, Thinking in Systems by Donella Meadows for modeling reality, plus engineering-specific titles. He notes the canonical engineering-strategy book still feels missing, which is part of why he's writing one.

  • Start with Good Strategy Bad Strategy (Rumelt); read The Crux second.
  • Thinking in Systems (Meadows) helps you model reality and improve your diagnosis.
  • Engineering-specific: Technology Strategy Patterns, The Value Flywheel Effect, The Phoenix Project.
  • Will feels the canonical engineering-strategy book still doesn't exist.

“the Richard rumelt book I think good strategy bad strategy is probably the right starting point”

Will Larson · 25:30
#books#strategy#recommendations#engineering-strategy

Takeaway· 2

Takeaway27:30

Write What Energizes You, Then Just Publish It

Will's core writing philosophy is that you can write far more if you write what you actually want to write, not for money, deadlines, or an audience. He abandoned an infrastructure book that just wasn't coming together, because forcing unenergizing writing takes you negative. His publishing rule: he ships almost everything he writes, and if he can tell up front he won't publish something, he doesn't even start it.

  • Writing on someone else's topic and deadline is draining; owning the schedule is where the energy is.
  • He gave up an infrastructure engineering book when it wouldn't come together.
  • He publishes almost everything he writes, accepting that some of it isn't that good.
  • A quick internal check ('can I write and publish this?') stops him from starting doomed drafts.

“you can write a lot more if you write what you want to write”

Will Larson · 27:30

“I publish pretty much everything I write”

Will Larson · 39:00
#writing#content-creation#energy-management#career
Takeaway1:14:30

Why High-Standards People Get Labeled 'Hard to Work With'

Will's proudest recent piece, 'Hard to Work With,' names a trap he lived through early in his career: talented people who hold peers to a high standard get labeled combative or difficult, even though they think they're just being rigorous. Companies say they want high standards, but when you actually enforce them, people can turn on you. He hopes more people read it because many of the most talented workers never crack this idea.

  • Very talented people who hold peers to a high standard often get seen as combative or hard to work with.
  • It comes from his own early career, where trying to hold people accountable made colleagues say he sucked to work with.
  • Companies claim high standards but don't always actually want them enforced.
  • Many talented, caring people go their whole careers without resolving this and are despised precisely because they care so much.

“I see a lot of people who are incredibly talented but they try to hold their peers to a high standard and then they're viewed…”

Will Larson · 1:14:30
#career#high-standards#collaboration#writing