How Serious Systems LearnPart IV — When Disciplines Fail
Chapter 13 — The Seduction of Coherence
Why clean explanations are often the enemy of learning.
Coherence feels like competence.
A coherent account explains what happened, why it happened, and what to do next. It aligns teams, calms stakeholders, and restores momentum. After uncertainty and conflict, coherence feels like relief.
That relief is precisely why coherence is dangerous.
In complex systems, tidy explanations often close inquiry before reality has finished speaking. They convert partial interpretation into stable certainty, then reorganize attention around preserving that certainty.
What begins as orientation becomes enclosure.1
Chapter 13 examines coherence as a failure mechanism: not because narrative is bad, but because narrative can become a substitute for continued correction.
Coherence Is a Coordination Technology
Coherence is not merely cognitive. It is organizational.
Institutions need shared stories to coordinate action across distance. Without narrative convergence, large systems fragment into local interpretations and stall.
So coherence is rewarded:
- in leadership communication,
- in board updates,
- in media framing,
- in post-incident reports,
- and in strategy documents that must travel quickly.
This reward structure creates asymmetry.
Stories that are simple, directional, and identity-consistent spread faster than stories that are conditional, multi-causal, and temporally open.
The faster a system must move, the stronger this asymmetry becomes.2
From Sensemaking to Sense-Closing
Sensemaking is necessary under uncertainty. Sense-closing is premature certainty under uncertainty.
The transition is subtle and usually unintentional.
It happens when provisional accounts acquire institutional commitments: resource allocations, reputational claims, policy locks, and authority stakes. Once that happens, updating the story requires undoing more than an interpretation. It requires undoing identity and investment.
At that point, disconfirming evidence is not evaluated neutrally. It is filtered through what continuation requires.
The story that helped the system orient now constrains what the system is allowed to see.3
The Narrative Economy
Every organization runs a narrative economy.
In that economy, coherence produces legitimacy. Legitimacy produces authority. Authority stabilizes coherence.
This loop is efficient in stable contexts. In volatile contexts, it can trap learning.
Why? Because admitting narrative instability can be interpreted as loss of control. Leaders then face a structural incentive: preserve credibility through consistency even when evidence suggests revision.
The result is not always lying. More often, it is selective attention:
- anomalies are reframed as edge cases,
- timeline inconsistencies are called transition effects,
- contradictory data is assigned lower confidence until it can be absorbed into the dominant story.
Coherence remains intact. Contact with reality thins.4
The Comfort of Monocausal Explanations
Complex failures are usually multi-causal. Monocausal stories are easier to communicate.
So systems simplify: "this failed because of X," "this worked because of Y," "the issue is leadership," "the issue is culture," "the issue is process."
These statements are often directionally true and operationally insufficient.
They obscure interaction effects: how incentives shaped reporting, how metrics suppressed weak signals, how authority delayed correction, how time pressure narrowed interpretation.
Monocausal coherence reduces cognitive load. It also removes the levers required for serious adaptation.
When the map has one cause, the solution has one knob. Real systems rarely cooperate with that hope.5
Coherence and Moral Certainty
Narrative closure is not only operational; it is moral.
Coherent stories assign roles: responsible and negligent, disciplined and reckless, visionary and resistant. These assignments stabilize identity and accountability politics.
In some cases, that clarity is necessary. In many cases, it is premature.
Moral certainty that arrives too early can disable diagnosis. Once actors are typed, evidence is interpreted through role expectation. Behavior that fits the story is amplified; behavior that complicates the story is discounted.
Correction then becomes reputationally asymmetrical: some actors can be revised; others can only be defended or condemned.
Serious systems resist this asymmetry by keeping moral judgment tied to evolving evidence, not fixed narrative position.6
How Coherence Suppresses Weak Signals
Weak signals are fragile by definition. They are partial, ambiguous, and easy to dismiss.
Coherence amplifies dismissal pressure:
if a weak signal does not fit the dominant story, it appears as noise costing attention the system believes it cannot spare.
Over time, this produces a critical pattern:
- strong signals are those that confirm narrative,
- weak signals are those that challenge it.
But in many failures, early warnings appear exactly as weak, narrative-inconvenient signals.
When coherence governs relevance, the system filters out the information most likely to preserve corrigibility.7
Preserving Narrative Without Surrendering Learning
The goal is not to eliminate coherence. The goal is to discipline it.
These systems use “provisional coherence”: narratives strong enough to coordinate, weak enough to revise.
This requires explicit practices:
- labeling major interpretations with confidence levels and expiry conditions,
- separating operational directives from causal certainty claims,
- requiring periodic narrative challenge reviews,
- maintaining minority interpretive tracks in high-stakes domains,
- and making revision events visible as competence rather than failure.
These practices keep coherence instrumental instead of sovereign.
A story should guide action. It should not govern what evidence may count.8
The Test of Seriousness
A useful test is simple:
Can this institution tell a new story before crisis forces it?
If the answer is no, coherence has become captivity.
If the answer is yes, coherence remains a tool.
Serious systems are not storyless. They are revisable.
They understand that in uncertain environments, epistemic dignity is not the ability to maintain one coherent account indefinitely. It is the ability to update account, authority, and action together as reality changes.
Footnotes
-
Weick, Karl E. Sensemaking in Organizations. Thousand Oaks, CA: SAGE Publications, 1995. ↩
-
Scott, James C. Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven, CT: Yale University Press, 1998. ↩
-
Berger, Peter L., and Thomas Luckmann. The Social Construction of Reality: A Treatise in the Sociology of Knowledge. New York: Anchor Books, 1966. ↩
-
Selznick, Philip. Leadership in Administration: A Sociological Interpretation. New York: Harper & Row, 1957. ↩
-
Perrow, Charles. Normal Accidents: Living with High-Risk Technologies. New York: Basic Books, 1984. ↩
-
Arendt, Hannah. Responsibility and Judgment. Edited by Jerome Kohn. New York: Schocken Books, 2003. ↩
-
Weick, Karl E., and Kathleen M. Sutcliffe. Managing the Unexpected: Sustained Performance in a Complex World. 3rd ed. Hoboken, NJ: Jossey-Bass, 2015. ↩
-
Senge, Peter M. The Fifth Discipline: The Art and Practice of the Learning Organization. New York: Doubleday/Currency, 1990. ↩
