Skip to content
finalstate.aiLatent Exposure: The Loss You've Already Agreed To
105Atlas

105THE LOSS ALREADY AGREED TO

Latent Exposure: The Loss You've Already Agreed To

A latent exposure is an unrecognised dependency, contract term, or knowledge concentration that can turn today's operating setup into a future loss. Accepting the dependence does not mean choosing the loss.

One supplier feeds an operation through a single connection.The crate, connection and machine form an illustrative sole-supplier dependency. Accent identifies the connection to examine. Nothing breaks, and the drawing predicts neither a date nor a loss.SupplierOperation

THE NEGATIVE MIRROR

Recognition exposes two different decisions.

An upside candidate requires secured access; a supplier dependency already connects to the operation.The outlined candidate is separated from access by a gate. Securing investment opens that gate. Below, a crate already feeds a machine through a single connection; recognition highlights the dependency without adding another source. The paths are an editorial comparison, not symmetric payoff curves.CandidateSecureDependence
This upside move needs secured access. The supplier comparison is an editorial application with different economics.

A shadow option is a candidate move in existing resources. In Bowman and Hurry's option sequence, recognition and securing investment are distinct. A latent exposure is downside already embedded in a dependency. Recognising either state leaves a further decision to make.

WHERE INDICATORS EXIST

A scan can follow an indicator. Some cases offer none.

Two observable paths and a third case with no usable indicator.A candidate opportunity has a separate investment gate; the supplier path carries a dependence. Their indicator paths can reach a scanning aperture. A separate outlined case has no connecting indicator and remains outside it. Positions and counts are schematic, not coverage estimates.OpportunitySecureExposureNo indicator

The frequency-impact matrix helps place consequences once there is a case to examine. A scan can flag an opportunity before expiry or a dependency before its trigger, where indicators exist. It cannot establish complete coverage, and noticing an upside candidate does not secure it.

Some responses need time that a failure takes away.

Qualifying a second supplier takes time. A handover may need the departing expert present; a hedge may need arranging before a contract resets. There is no universal cost curve. Identify the responses available now and the conditions under which they would disappear.

TIME AND UNDERSTANDING

Time to respond and understanding can fail separately.

Response slack narrows separately from the appearance of unexpected connections.The upper connection loses response slack. The lower process gains cross-connections that obscure a disturbance's route. These independent schematic changes illustrate Perrow's coupling and interaction dimensions; they do not assert that all tight systems lack warnings or assign a measured response time.SlackInteractions
Perrow, Normal Accidents, chapter 3, p. 78 and table 3.2, p. 96. Schematic tendencies, not measurements.

Perrow's account separates interaction complexity from coupling. Complex interactions can obscure a disturbance's path; tight coupling leaves little slack for recovery. Their combination can let a cascade outrun diagnosis. These are limits on weak-signal recognition, not a rule that every failure has zero warning.

THE MOVING BASELINE

Local success can make a deviation look normal.

Current practice and a working reference move away from a fixed original comparison.A solid practice path and a dashed working reference step away from the original horizontal comparison. That original stays fixed and is highlighted at the end. There is no measured danger threshold, guaranteed detection or claim that the original practice was safe.OriginalPracticeReference
Editorial schematic applying Vaughan and Dekker. The fixed comparison is a proposed check, not either author's measured figure.

Vaughan's Challenger study traced repeated acceptance of anomalous O-ring performance. Dekker's account of drift follows ordinary decisions under pressure. A fixed comparison and review outside the local decision chain are proposed checks for the recognition engine; neither guarantees detection.

JUDGMENT IN ONE HEAD

A manual can remain when practiced judgment leaves.

A manual remains in a workshop while the expert and their integration of cues move toward the exit.The book stays on the left. The person, cue marks and their connecting lines share a moving group toward the doorway; no complete knowledge diagram remains behind. The distinction concerns practiced judgment, not a claim that expertise can never be discussed or taught.ManualJudgment

An operator may hear trouble before a machine's instruments show it. Procedures can survive their departure while tacit knowledge needed to interpret an unfamiliar case leaves. The exposure depends on which tasks someone else can perform, including exceptions they have practised handling.

EVIDENCE OF CAPABILITY

A test can establish a capability. Avoided loss needs a comparison.

A stand-in attempts a task, with a result record distinct from an unresolved estimate of avoided loss.A person approaches a task with a retained manual. A trial can produce a record of observed performance, represented by neutral writing lines without a pass mark. A separate dashed field for avoided loss remains empty. The drawing does not show a successful transfer or an amount of loss prevented.TrialResultAvoided loss
Illustrative capability test. A result can record success, failure or limits; avoided loss remains a separate inference.

A supplier failure can belong in the rare-and-severe corner of the frequency-impact matrix. A contract term can also create frequent, smaller losses. The prevention problem requires a distinction between testing a response and estimating what would have happened without it. An archive that answers back must face that test too.

BEFORE ANYONE LOOKS

Keep the exposure open until the response has evidence.

An alternate supplier awaits a trial while the time for a response narrows and evidence remains open.The original supplier connection remains the only operating route. A dashed second-source route ends at an uncompleted trial boundary. A schematic bracket narrows the response window without assigning a countdown. The result field stays empty at the complete still; proposing a response never closes the exposure.TrialWindowEvidence open
Illustrative second source, awaiting qualification. The response window has no measured duration.

For each dependency, name the trigger, an observable indicator if one exists, and the response that remains available. Then specify evidence of reduced exposure: a qualified second source, a workable contract arrangement, or a stand-in who can handle the task. Use that record to act before the event, and continue to the archive that answers back and the #risk it was built to hold.

Read the transcript

01 · THE LOSS ALREADY AGREED TO

A business can depend on one supplier without recording what would happen if deliveries stopped. Its operation works today, while a future interruption has no qualified alternative. Other exposures sit in contract terms that can reset against the firm, or in judgment concentrated in an employee who may leave. Nobody needs to choose a loss outright. The dependence can form through ordinary operating decisions. A latent exposure is downside embedded in the operating setup, not yet recognised as risk. The loss remains conditional; the dependence already exists.

02 · THE NEGATIVE MIRROR

In Bowman and Hurry's account, a shadow option is a candidate move in existing resources awaiting recognition. The sequence illustrated here separates recognising that move from securing a right to pursue it. A securing investment can establish preferential access and create a real option: a right, without obligation, to commit later. The downside exposure is already embedded in a position the firm depends on. Recognising the sole supplier reveals a dependence; it does not qualify an alternative. The two cases share a recognition problem, but the decisions that follow have different economics.

03 · WHERE INDICATORS EXIST

A scan can search for both kinds of case. On the upside, an observable indicator may reveal a candidate opportunity before it expires. For this example, securing the candidate remains a separate investment decision. On the downside, an indicator may reveal that a supplier dependence is becoming more dangerous. There may also be exposures for which the scan receives no usable indicator. Recognition can be a bottleneck in either direction, without the upside and downside having matching distributions or equal visibility.

04 · BEFORE THE TRIGGER

Timing changes the available choices. Before a supplier fails, a firm may be able to qualify a second source. Before an employee leaves, someone else can practise the work with them. A contract hedge may need arranging before a reset. At the event, some responses may be slower, more expensive, or gone. There is no universal cost curve. Identify which response exists now and what would make it unavailable.

05 · TIME AND UNDERSTANDING

Charles Perrow's Normal Accidents separates interaction complexity from coupling. Tight coupling leaves little slack or time to intervene. Interactive complexity makes a disturbance's path difficult to understand: effects can travel through unfamiliar or unexpected connections. Where the two combine, a cascade can outrun diagnosis. These are tendencies, rather than two absolute categories. A tightly coupled system may still offer a warning, but detecting trouble does not ensure enough understanding or time to respond.

06 · THE MOVING BASELINE

Diane Vaughan's study of the Challenger launch decision traced repeated acceptance of anomalous O-ring performance within NASA's working culture. Sidney Dekker describes drift through small decisions made under pressures of scarcity and competition. Each local adjustment can seem reasonable while their cumulative direction goes unrecognised. A comparison loses force if the working reference moves with current practice. The proposal here is to retain a fixed comparison point and invite review from outside the local chain of decisions. Either can reveal a change that merits investigation; neither guarantees detection.

07 · Advertisement · Bubble AI App Builder

Some ideas do not need another document before they become testable. With Bubble AI, you describe the app you want, and Bubble creates a working starting point: interface, data, and logic you can inspect. From there, you refine visually, connect AI models and services, and turn the first version into something real enough to use.

08 · JUDGMENT IN ONE HEAD

In a small firm, the dependency may be a particular operator. They hear trouble in a machine before it shows, know which supplier can bend, or recognise which complaint matters. A manual may capture procedure while missing judgment learned through practice. When the person leaves, the manual stays. Their practiced integration of cues leaves with them. The continuity concern is specific: which tasks can someone else perform, and which unfamiliar cases still require that person? No fixed share of the firm's value follows from the example.

09 · EVIDENCE OF CAPABILITY

The supplier example can sit in the rare-and-severe corner of the matrix. The definition is broader: a contract term can create frequent, smaller losses. Surveillance can watch an indicator; a knowledge archive can support a handover. Neither activity establishes that exposure has fallen. Backtests and planted canaries test detection. A stand-in drill tests whether another person can carry out the work under stated conditions. These are observations of capability. Estimating a loss avoided also requires a credible comparison with what would have happened without the intervention.

10 · BEFORE ANYONE LOOKS

Return to the supplier. The firm can name its dependence, the trigger that would interrupt deliveries, and a second source it might qualify. That proposal still needs a trial. As the response window shrinks, naming an alternative does not make it ready. Keep the evidence field open until there is a result to record. The same discipline applies to a contract arrangement or a handover: specify what response is available, who can carry it out, and what test would show the dependency has been reduced. A quiet period is time to investigate, not evidence that the exposure has gone.

01 / 10 · THE LOSS ALREADY AGREED TO0:00 / 6:09