I spent five years at a startup that looked right on paper.
The mission was real. The product was working. The people were talented — genuinely talented, some of the sharpest engineers and designers I'd worked with. The market was there. The funding was there.
And yet. Something was off. Not catastrophically — nothing you could point to and say "there, that's the problem." Just a persistent, low-level unease. A feeling of walking on unstable ground even on good days. A sense that the rules of the game were unclear, or that they changed without notice.
I couldn't name it then. I just knew that I was tired in a way that sleep didn't fix.
What I Noticed But Couldn't Articulate
Looking back, there were patterns I observed without having language for them.
Information moved strangely. Important decisions happened in small rooms and arrived to the rest of the team as announcements. When things went wrong, the conversation focused immediately on who was responsible rather than what had happened and why. Raising a concern felt like a risk — not because anyone explicitly punished it, but because the conversation that followed always seemed to end up somewhere uncomfortable for the person who had raised it.
I compensated. I learned to read the room before speaking. I learned which topics were safe and which ones weren't. I learned to phrase problems as questions rather than observations. I got good at navigating the fog.
I didn't realise that adapting to fog was itself a signal.
The Research
Years later, working with other companies and building a consulting practice, I was reading through the Google Cloud State of DevOps report — the research that tracks what distinguishes high-performing engineering organisations from low-performing ones.
Buried in the references was the work of Ron Westrum, an organisational sociologist who had spent decades studying how information flows in complex organisations — originally in aviation and healthcare, later applied more broadly.
His framework was simple. Three types of organisational culture, defined by how they handle information:
Pathological organisations are power-oriented. Information is withheld because it is a source of power. Messengers are shot. Responsibilities are siloed. Failures are hidden. Bridging between functions is discouraged.
Bureaucratic organisations are rule-oriented. Information flows through established channels. Responsibilities are defined by role. Failures are judged against procedures. Bridging is allowed but only through proper channels.
Generative organisations are performance-oriented. Information flows freely to where it's needed. Messengers are trained. Responsibilities are shared. Failures are treated as opportunities to learn. Bridging is actively encouraged.
I read it three times.
The Chill of Recognition
The reason I read it three times was not because it was complicated. It was because it was describing something I had lived.
The startup was pathological. Not unusually so — not in the way that makes headlines. Just in the ordinary, structural way that many early-stage companies are, where power concentrates at the top, information becomes a resource to be managed rather than shared, and the culture around failure makes honesty feel dangerous.
The chill I felt was recognition. Five years of unease, explained in three paragraphs.
But it was also relief. Because Westrum's framework was not a character assessment. It was a structural one. The culture I'd spent five years navigating was not the result of bad people. It was the result of a particular set of incentives and information flows that produced predictable, well-documented patterns.
Which meant: systems can be changed. Cultures can shift.
What Westrum's Research Actually Shows
The reason the State of DevOps report references Westrum is that organisational culture turns out to be one of the strongest predictors of software delivery performance. More predictive than technology choices, methodology choices, or team size.
High-performing engineering teams — the ones that deploy frequently, recover from incidents quickly, maintain low change failure rates — are almost universally found in generative cultures. The correlation is not subtle.
This matters because it reframes the question most engineering leaders are asking. The question is usually some version of "how do we go faster?" or "how do we improve quality?" or "how do we reduce incidents?"
Westrum suggests the prior question is: what kind of organisation are you? Because the answer to that question determines whether any of the tactical interventions will work.
You can implement Shape Up in a pathological culture. It won't do what you hope. You can run blameless post-mortems in a culture where people don't feel safe being honest. They won't be blameless. You can hire talented engineers into an environment where information is hoarded and failure is punished. They'll leave.
Structure is not enough. Culture is the substrate that everything else runs on.
The Questions Worth Asking
When Westrum wrote to me — I had emailed him directly after reading the research, not expecting a reply, and he responded — he recommended I read David Abrashoff's It's Your Ship. A naval captain who inherited the worst-performing ship in the US Pacific Fleet and turned it into the best, not through authority but through a systematic practice of listening, distributing responsibility, and treating failure as information.
The Westrum/Abrashoff combination gave me a frame that has shaped how I work ever since.
The questions that matter when entering a new organisation are not primarily technical. They are cultural:
- When something goes wrong, what happens next? Is the first conversation about learning or about blame?
- Do people share bad news upward? Freely, or with careful preparation?
- Are there things everyone knows but nobody says?
- Does information flow to where it's needed, or does it pool at the top?
- When someone in a different function has a relevant perspective, do they share it? Are they invited to?
These questions don't have clean answers. But the texture of how people respond to them tells you more about an organisation than any process document or org chart.
The Silver Lining
The five years of unease were not wasted. They gave me a lived understanding of what pathological culture feels like from the inside — not as a concept, but as a daily experience. That understanding shapes how I read organisations now, what I listen for in early conversations, what I treat as a signal versus noise.
And Westrum's research gave me the relief of knowing: it wasn't me. It was the system. And systems can be changed.
That's not idealistic. It's what the research shows. Cultures shift when the information flows change, when failure is treated differently, when people feel safe enough to say what they actually see.
It's not fast. It's not easy. But it's possible.
And it's where the real work begins.
Thomas Riboulet is a Fractional VP of Engineering working with European tech companies. He writes about engineering leadership, team structure, and sustainable delivery at insights.wa-systems.eu.