../talks
enpt

Engineering culture

The references behind the talk. Where everything I said came from.

The bottleneck was never technical

1968

Melvin Conway

How Do Committees Invent?

The original Conway's Law paper. Mind the direction of the arrow: a system's design isn't influenced by the organization, it is a copy of its communication structure.

1975

Fred Brooks

The Mythical Man-Month

You can't buy a deadline with people: coordination cost grows faster than the capacity you add.

1987

Tom DeMarco e Tim Lister

Peopleware

The major problems of our work are not so much technological as sociological in nature. Almost twenty years after Conway, the idea finally becomes a management book.

What we say and what we do

1974

Chris Argyris e Donald Schön

Theory in Practice

They separate the theory you profess from the theory your behavior reveals. And the finding that keeps the conversation from turning into an accusation: people usually don't know the two diverge. It isn't hypocrisy, it's invisibility.

1985

Edgar Schein

As três camadas da cultura organizacional

Artifacts, espoused values and basic assumptions. When layer 2 and layer 3 disagree, layer 3 wins. That's why writing the doc changes nothing on its own: the artifact has no access to the layer that decides.

The term was born as a criticism

1992

Gideon Kunda

Engineering Culture

The book that coined the phrase, and the pun in the title is the thesis: not the culture of engineers, but the engineering of culture. Kunda argues the informal, flexible workplace rhetoric operates as normative control. The first canonical use of the term was already a suspicion, and it went on to become a job requirement.

Where the unwritten decision kills people

1996

Diane Vaughan

The Challenger Launch Decision

Where normalization of deviance comes from. None of those decisions was wrong on its own, each was defensible on the day, and the sum brought down the shuttle.

2004

Ron Westrum

A typology of organisational cultures

Classifies culture by information flow rather than by values. The question isn't what the company values, it's what happens to bad news. It's the most honest test here, because a well-written doc can't cheat it.

2012

John Allspaw, Etsy

Blameless PostMortems and a Just Culture

If you read only one, read this. The blameless postmortem is a cultural practice with a mechanism, not a value on a wall: it has steps, and it was designed to change a decision nobody can mandate, the decision to say what actually happened.

The manifestos, and the hangover

2009

Netflix

Culture Deck

The document that founded the genre, with over 15 million views. Worth noticing it's a talent deck, not an engineering one, which explains why these documents read like recruiting material: they are.

2015

Google

Project Aristotle

Around 180 teams measured. The single biggest predictor of effectiveness wasn't composition, collective IQ or technical skill, it was psychological safety. That is: whether the person decides to speak up.

2018

Nicole Forsgren, Jez Humble e Gene Kim

Accelerate e os relatórios DORA

They take Westrum's typology and validate it empirically. Look at what exactly got validated: information flow. Not values, not rituals, not the doc.

2020

Jeremiah Lee

Spotify's Failed #SquadGoals

The most copied cultural artifact of the 2010s described a company that didn't exist, not even at the company that published it.

2016

Zach Holman

How GitHub (no longer) Works

The same author of How GitHub Works coming back years later to say the company adopted hierarchy, managers and top-down decisions. The two pieces side by side are a complete case study in ten minutes.