Home › Culture & Change › Pathological, Bureaucratic and Generative Cultures
Culture & Change · Westrum Organizational Typology

Pathological, Bureaucratic and Generative Cultures

Whether bad news reaches the top often depends on what happens to those who carry it.

◷ 4 min read▤ 4 true storiesWorks with AI, with changesSee its neighbors on the map →

A Nokia middle manager remembers how a chart could change on its way up. The supervisor would tell you to move the data points to the right, to give a better impression. Then the supervisor would take the chart to the higher-level executives and present it.

Through those years, Nokia sold more phones than any company on earth. It was about to lose the smartphone race, and fast.

The manager later described the routine to two researchers, Timo Vuori of Aalto University and Quy Huy of INSEAD. “The information did not flow upwards,” the manager said. “Top management was directly lied to.”

Shot, neglected or trained

Ron Westrum, a sociologist at Eastern Michigan University, created a typology in 1988 to compare how organisations handle information, especially information that suggests something is wrong. In 2004 he set it out for medicine in Quality and Safety in Health Care.

He starts with the leader, because staff learn what is rewarded: “rewards, punishments, and resources follow the leader’s preferences.” A preoccupation with personal power, needs and glory produces what he called a pathological culture. Rules, positions and departmental turf produce a bureaucratic one. A focus on the mission produces a generative one.

His table, in part:

Pathological Bureaucratic Generative
Orientation Power Rules Performance
Messengers Shot Neglected Trained
Failure leads to Scapegoating Justice Inquiry
Novelty Crushed Leads to problems Implemented

Whether bad news reaches the top often depends on what happens to those who carry it.

In pathological circles, Westrum wrote, information is “a personal resource, to be used in political power struggles.” In generative ones the guiding question is “who needs the information now?” The bureaucratic middle doesn’t punish the messenger. It sends the message through standard channels, which “are often insufficient in a crisis.”

Why tell them?

The Nokia manager explained the calculation. Sometimes everybody at that level knew things were going wrong, “but we were thinking, ‘Why tell top managers about this? It won’t make things any better.’” Fearing top managers’ reactions, Vuori and Huy concluded, middle managers “remained silent or provided optimistic, filtered information.” Optimistic reports went up, so the top saw no reason not to push harder.

A smoke alarm that gets shouted at every time it beeps soon loses its battery. The kitchen goes quiet. The fire risk stays.

Silence doesn’t need shouting, either. In the Dutch childcare benefits scandal, an internal memo listing the benefits office’s own mistakes never reached the state secretary, who told parliament in 2017 that nothing had shown parents suffering unnecessary harm. No one has to shoot a messenger whose memo stops short of the decision.

How strong is the proof?

Westrum is frank about the limits. The idea that these climates shape communication, cooperation and problem solving “is a hypothesis,” he wrote, “based on considerable observation and study,” and the link to performance is “statistical, not deterministic.” Case studies are plentiful, systematic tests rarer, and of the studies tying the three cultures to medical outcomes, some show a relationship while “others fail to show it.”

The largest data set comes from software. Google Cloud’s DORA research program measures culture with survey items based on Westrum’s table, one of which reads: “Messengers are not punished when they deliver news of failures or other bad news.” Its 2023 report, drawing on nearly 3,000 professionals, found that teams with generative cultures had 30% higher organizational performance. Culture and performance both come from respondents’ own answers, so this is an association across many organisations, not an experiment that changed a culture and watched results move.

The amber report

Imagine a project lead whose data migration is slipping. The last colleague who marked a project red spent the next steering meeting defending himself. So she marks hers amber, writes “risks being managed,” and hopes to catch up. If her director’s usual first answer to red were “what do you need?”, the same slip would surface weeks earlier, while there was still time to move people or cut scope.

When things go wrong, Westrum wrote, pathological climates look for a scapegoat, bureaucratic ones seek justice, and generative ones try to find the basic problems with the system. That third answer is what blameless postmortems and psychological safety try to make routine.

When the first messenger is a machine

AI systems add messengers that fear no one, such as logs and sensors. The relay is still human. After a driverless Cruise car dragged a pedestrian in San Francisco in October 2023, an engineer told a Slack channel of 77 employees by 3:45 the next morning that the car had moved after the impact. Investigators hired by the boards saw too much of an “us versus them” attitude toward regulators; some staff worried that regulators had missed the dragging, and no one acted on it.

California has tried to write the messenger row into law. SB 53, the Transparency in Frontier Artificial Intelligence Act, signed on September 29, 2025, requires large frontier developers (those with more than $500 million in annual revenue, among other tests) to give staff who assess or manage the risk of critical safety incidents an internal way to report anonymously. It covers good-faith concerns about a specific and substantial danger from catastrophic risk, or about breaches of the act. The reporter gets monthly updates on the investigation, reports and responses go to officers and directors at least quarterly, and retaliation is barred.

Catastrophic risk is defined narrowly: a model materially contributing to the death or serious injury of more than 50 people, or over $1 billion in property damage, in a single incident where the model gives expert help toward chemical, biological, radiological or nuclear weapons, carries out a cyberattack or acts that would be crimes such as murder or theft with no meaningful human oversight, or evades its developer’s or user’s control. Only certain staff are covered. And a law can carry news to the board without making the board want it. Nokia had a reporting line too. The chart went up it every time.

Stories behind it

Play it

Sources

  1. Ron Westrum, A typology of organisational cultures, Quality and Safety in Health Care 13 (Suppl II), 2004
  2. Timo O. Vuori and Quy N. Huy, Distributed Attention and Shared Emotions in the Innovation Process, How Nokia Lost the Smartphone Battle, Administrative Science Quarterly 61(1), 2016
  3. Quy Huy and Timo Vuori, Who Killed Nokia? Nokia Did, INSEAD Knowledge, September 22, 2015
  4. DORA (Google Cloud), Accelerate State of DevOps Report 2023
  5. DORA, Capabilities, Generative organizational culture
  6. Tweede Kamer, Parlementaire ondervragingscommissie Kinderopvangtoeslag, Verslag Ongekend onrecht, December 17, 2020
  7. Quinn Emanuel, Report to the Boards of Directors of Cruise LLC, GM Cruise Holdings LLC and General Motors Holdings LLC regarding the October 2, 2023 accident in San Francisco, January 24, 2024
  8. California Senate Bill 53 (Chapter 138, Statutes of 2025), Transparency in Frontier Artificial Intelligence Act, chaptered September 29, 2025
  9. Engadget, Nokia CEO Stephen Elop rallies troops in brutally honest 'burning platform' memo (full text), February 8, 2011
  10. NBC News, Nokia standing on a 'burning platform,' CEO tells employees, February 9, 2011
  11. Microsoft, Microsoft to acquire Nokia's devices & services business, license Nokia's patents and mapping services, September 3, 2013
  12. The Guardian, Microsoft buys Nokia handset business for €5.4bn, September 3, 2013
  13. NHTSA, NHTSA Announces Consent Order with Cruise After Company Failed to Fully Report Crash Involving Pedestrian, September 30, 2024
  14. US Attorney's Office, Northern District of California, Cruise admits submitting a false report to influence a federal investigation, November 14, 2024
  15. CBS News Bay Area, Cruise admits to false report in 2023 dragging of San Francisco pedestrian, November 14, 2024
  16. NBC Bay Area, CPUC orders Cruise to pay $112,500 in penalties, June 20, 2024
  17. General Motors, GM to refocus autonomous driving development on personal vehicles, December 10, 2024
  18. CNBC, GM exits robotaxi market, will bring Cruise operations in house, December 10, 2024
  19. Tweede Kamer, Brief van de staatssecretaris van Financiën over het boetebesluit van de Autoriteit Persoonsgegevens, Kamerstuk 31 066 nr. 936, December 7, 2021
  20. Amnesty International, Xenophobic Machines, Discrimination through unregulated use of algorithms in the Dutch childcare benefits scandal, EUR 35/4686/2021, October 2021
  21. The Guardian, Dutch government resigns over child benefits scandal, January 15, 2021
  22. Amy C. Edmondson, Learning from Mistakes Is Easier Said Than Done, Journal of Applied Behavioral Science 32(1), 1996 (figures and unit details cited via the Psychological Safety Field Guide summary)
  23. Hidden Brain, Making the Most of Your Mistakes (interview with Amy Edmondson)