Pathological stability
Why organisations fail
At the end of 2024, I wrote a piece on why organisations fail—A failure of facts.
It was written from inside the problem. Less as a theory of failure, and more as an attempt to understand why familiar explanations—bad leaders, the wrong professions, bureaucracy—so often leave people with little they can actually do.
As I move into a later phase of my work, I wanted to take stock from a little distance. What emerged was unexpectedly academic. Perhaps I was satisfying an unmet calling to philosophy.
Even so, there are a few ideas in that work that seem worth extracting.
To spare the reader the tedium of the full paper, what follows is a brief account of the definitions and claims that appear most relevant to daily organisational life.
I suspect it will be useful to someone who feels quietly stuck inside a large organisation—not because they lack effort or care, but because the system itself seems strangely resistant to change.
The paper also gestures toward a way forward—not a fix, but some questions we may ask.
Pathological stability
Pathological stability describes a condition where an organisation is busy, competent, and stable—yet increasingly fragmented and drifting from intent.
Locally, every team is fulfilling its remit.
Globally, intent has slipped, year by year.
Everyone knows the organisation has problems—many can point to what they are—but no one seems able to do anything about them.
Why does this happen?
To answer that question, we need to define the agency cycle and the five conditions for organisational learning.
The agency cycle
Agency is the capacity of an entity to act toward an intent, observe the effects of its action, and revise its orientation accordingly.
The agency cycle consists of three states and one movement:
Expectation: a judged orientation toward the world
Effort: action taken under that orientation
Effect: what actually occurs in reality
Feedback: the interpretive act that links effect back to expectation
Formally:
Expectation → Effort → Effect → (feedback) → Expectation
Feedback is not a fourth node. It is an act of judgement. It requires attention, interpretation, and willingness to revise expectation.
In human systems, this act is always carried by someone capable of holding the loop in view rather than defending a local position.
The five conditions for organisational learning
For an organisation to advance in its agency cycle—to revise expectations in service of intent—it needs to learn.
There are five conditions for organisational learning: Pain, Freedom, Responsibility, Authority, and Attention.
Humanly, they correspond to:
The person who is hurt
(experiences friction, overload, failure, or moral discomfort)The person who is free to act
(can change assumptions, redirect effort, or alter structures)The person who must act to remain faithful to intent
(recognises that inaction would violate purpose)The person who controls the routing of consequence
(buffers, escalates, delays, or absorbs pain)The person who can hold the whole picture in view
(sustains attention to misalignment rather than reacting to salience)
Learning occurs only when these roles are structurally connected.
Many organisational analyses bundle Responsibility and Authority together. Separating them surfaces a painful reality: the person who must act (Responsibility) often has no power to change the system (Freedom), while the person with power (Authority) is shielded from consequence (Pain).
When this configuration persists, learning becomes structurally impossible—and pathological stability follows.
Why do organisations end up in pathological stability?
Pathological stability does not require incompetence, bad faith, or malice.
It emerges through sensible adaptations to scale, risk, and uncertainty:
In complex environments, coherence becomes an easy proxy for truth.
As learning becomes costly, systems minimise pain rather than integrate it.
To preserve stability, organisations eliminate instability risk, especially single points of failure.
Each step is locally rational. Each improves short-term stability.
But overall, the five conditions for organisational learning—Pain, Freedom, Responsibility, Authority, Attention—become increasingly fragmented. It becomes harder and harder to pay attention to the feedback loop itself.
Eventually, the organisation reaches a state where the agency cycle is broken and meaningful revision becomes structurally unavailable. Not forbidden. Not denied. Simply absent.
As a result, complex organisations do not merely fall into pathological stability—they converge on it.
Coherence trap & single-signal collapse
Two symptoms are typical of an organisation in pathological stability.
Coherence trap
As organisations grow in size and complexity, learning directly from reality becomes difficult.
Consequences are delayed.
Causal chains span domains.
Interpretation fragments across roles.
No single actor can reliably hold the full feedback loop in view.
Under these conditions, organisations increasingly privilege coherence: internal consistency, narrative alignment, and explanatory simplicity.
Coherence is attractive because it is:
faster than sustained interpretation
easier to communicate across scale
less destabilising than competing signals
sufficient to maintain coordination
Over time, coherence becomes a cheap proxy for truth. Explanations are judged less by whether they revise expectations correctly, and more by whether they preserve internal consistency with existing structures and stories.
The question shifts from “What is reality saying?” to “Are we following the methodology, framework, or process?”
The problem is not coherence per se, but coherence becoming a non-revisable substitute for engagement with reality. Once fully in this state, the organisation has entered a coherence trap: a condition in which the success or failure of the proxy becomes a substitute for reality itself.
Under these conditions, single-signal collapse is no longer accidental, but attractive.
Single-signal collapse
The second symptom is single-signal collapse: the collapse of attention onto a single signal, causing feedback to fail.
Sustained attention is difficult and costly. Optimising around a single signal is the path of least resistance.
This occurs when an organisation:
treats one survey, metric, or incident as an explanation rather than a trigger
mistakes salience for causality
routes all interpretation and action through that signal
In this state:
other signals are discounted as anecdotal
existing capability is ignored or overwritten
optimisation replaces judgement
intent collapses into proxy
The error is not the signal itself, but allowing it to eclipse all others.
When a single signal becomes non-revisable, learning becomes impossible.
Traditional management practices accelerate convergence
Many orthodox management practices are designed to improve clarity, distribute accountability, and increase predictability.
Locally they succeed, globally, the weaken the organisation’s ability to learn.
They reward coherence of internal narratives over contact with reality.
They minimise pain that signals the need for change.
They fragment the conditions for learning.
Most importantly, they redirect attention away from organisational intent and toward artefacts—plans, metrics, reports, and narratives.
The system becomes more organised, more legible to itself, and less responsive to reality.
Why reform programs are not enough
Fixing many things at once does not repair the feedback loop itself.
Reform programs typically focus on incremental change: replacing leaders, redefining strategy, restructuring teams, launching transformations, or fixing visible problems. None of these are inherently wrong.
They fail when they operate locally—addressing symptoms, processes, or structures—without restoring the feedback loop that allows learning to occur.
Learning depends on whether pain, freedom, responsibility, authority, and attention are structurally aligned—on whether reality is able to revise expectation. As long as these remain fragmented, learning cannot occur.
When reform concentrates on improvement rather than alignment, activity increases but learning does not. The organisation may change many things while remaining fundamentally unable to revise expectation.
Total pathology
In its terminal form, pathological stability becomes total:
Everyone is responsible, but no one is free.
Those who carry pain have no authority; those who carry authority have no pain.
No one is paying attention to the loop.
This occurs when roles divide into functions, functions cement, resources scale with hierarchy, and everyone optimises locally within the loop.
This is the stable configuration that emerges when governance diverges from reality.
At this point, failure is not dramatic. It is procedural.
The way forward
The paper does not offer a solution.
What it offers instead is a reorientation: restore attention to the loop.
In practice, this means asking questions such as:
Have we fallen into a coherence trap and lost touch with reality?
Has attention collapsed onto a single dominant signal—a metric, survey result, framework, or governance artefact?
Is this decision optimising for safety rather than alignment with intent?
Is responsibility meaningfully coupled with freedom, or increasingly separated?
Is activity locally optimising rather than revising global expectation?
Has intent fallen out of the active feedback loop?
Are exits functioning as signals of agency preservation rather than ordinary turnover?
Not solutions—but diagnostic questions.
Conclusion
I would summarise the paper in a single line:
When consequences no longer threaten revision, misalignment settles into persistence.
The paper examines three case studies — the Boeing 737 MAX crisis, the UK Post Office Horizon scandal, and the Grenfell Tower fire. I have since applied the same framework to other contemporary problems our nation seems stuck in, and I have found it diagnostically useful when applied to other contemporary problems our nation seems stuck in.
How this account relates to A failure of facts—whether it contradicts, deepens, reframes, or simply stands alongside it—I will leave to the reader.
And no—I hope I don’t develop a habit of writing “Why organisations fail” once every year.
Full paper including deep dives into technical leadership, big organisations and why good people leave: Attention Over the Loop (SSRN)