Some Costs Must Be Checked Before They Are Averaged
At 13:00 on Saturday, Gu Ning opened another page beside the scenario-weighting sheet. The previous chapter's demonstration showed that an action with a one-tenth chance of losing nine units must bear all nine if failure occurs, even when its expectation remains positive. Her question now was not how likely D17 was to succeed, but which outcomes would remove the station's ability to keep fulfilling and correcting.
This work cannot wait for complete probabilities. If an action floor moves with the preferred plan, a calculator can always send that plan through with optimistic inputs. If the floor is recorded first, then when probabilities and prices arrive, everyone at least knows which consequences cannot be offset by average gains.
By 13:00 the client still has not sent new material and Version 2 has not formed. The classifications below use only the established ten labor slots, six existing commitments, twelve working-capital units, and eight-unit protection. They invent no price or success rate for a nonexistent order.
“Cannot Be Tried” Does Not Protect Dignity
An attempt may disappoint or expose mistaken judgment. Disappointment alone does not prohibit trial and error. If the station spends a small, recoverable amount of time clarifying a question and receives a negative answer, the result is adverse but completes the information task without breaking survival conditions.
Gu Ning named the protected objects concretely: the six accepted half-day jobs in the next working week can still be completed as promised; the eight units reserved for payments on existing jobs, basic operations, and unexpected rework remain available when needed; and if the new attempt stops, at least one recovery path can actually be taken within the present decision period.
Embarrassment, no reply, a calculation error, or surrendering an imagined gain cannot be promoted to a survival crisis without explanation. Calling every discomfort decisive failure strips the floor of its screening function and makes every plan look impossible.
Survival Is a Set of Conditions, Not a Slogan
Survival here does not mean the station operates forever or can recover from every shock. It applies only to D17's present decision period: after acting, the station can still fulfill explicitly assumed work, retain protected resources for that work, and return to an executable arrangement when the new path fails to form.
Each of these three conditions has adjudicative material. Existing fulfillment asks whether the six jobs still have personnel, materials, and time. Resource protection asks whether the eight units are occupied, frozen, or prepaid. The recovery path asks whether, after the new action stops, remaining work can be rescheduled within the agreed period rather than leaving “we will find a way” on the page.
If an outcome breaks any condition, high scores on other goals cannot make it disappear. Conversely, preserving all three says only that a plan stays above this chapter's floor. It does not prove the plan most valuable or guarantee success.
Eight Units Are a Protected Use, Not a Mysterious Constant
At least eight of the twelve working-capital units must remain for existing payments, basic operations, and unexpected rework. This is Chengwan's internal arrangement under facts known on Saturday, neither an industry standard nor an optimal reserve ratio derived from theory.
“At least eight” first protects a use. Temporarily transferring the units to D17 and replenishing them later is possible only if a verifiable inflow will occur before the corresponding obligation. Expected future income is not an accomplished replenishment. The critical point is availability when needed, not merely the closing balance.
The line can change with facts. Completion of an existing payment or definite release from rework responsibility may change what can be released. But revision needs a corresponding record, date, and responsible person. The station cannot lower protection from eight to six because a favored D17 plan happens to require six deployable units and then call it compliant.
Six Labor Slots Protect Things Already Undertaken
The six slots are not physical blocks that can never move. They represent concrete internal or external arrangements already made. Their state can be updated if the relevant object agrees to a new date, a substitute person is actually available, or the task finishes early.
Without such material, a D17 reservation must state whose capacity it displaces. “Everyone will work faster” does not create capacity. Treating nights or rest as free compensation invents a resource outside the consequence table.
A new preparation requiring a fifth available slot cannot be justified by saying that the week contains ten. Six are committed; only four are presently schedulable. A plan exceeding four must first obtain release conditions, reduce its action, or wait, rather than execute early and hope existing work does not fluctuate.
Ordinarily Adverse and Unacceptable Are Different Levels
Gu Ning first divided results into ordinarily adverse, hard to recover, and unacceptable. An ordinarily adverse result consumes resources or misses a gain, but after stopping the station can still fulfill work and retain protection. No reply, limited communication time, or an unused draft usually falls here when kept within its ceiling.
A hard-to-recover result does not immediately breach the three survival conditions but substantially contracts later choice. Examples include occupying most available capacity for a prolonged period, buying narrow-purpose materials that might still be repurposed, or making a scheduling commitment that requires several rounds of negotiation to withdraw. These can still be adopted, but require explicit recovery steps, deadlines, and bearers.
An unacceptable result breaks at least one survival condition under the current plan: the six existing jobs cannot be fulfilled, the eight protected units are unavailable when needed, or there is no actual recovery path after exit. The classification follows consequences and the subject's conditions, not whether the actor likes the result.
“Irreversible” Must Name a Time and Object
Many real-world actions are not irreversible in a cosmic sense. Advance payments may be recovered, commitments renegotiated, and specialized materials altered. Writing only “irreversible” can let a forceful word replace examination of recovery paths.
The case limits the term: for the protected object, within D17's current decision and fulfillment window, there is no already-existing recovery path that can be executed in time at a bearable cost. A payment might theoretically be recovered after a long dispute, but if the eight protected units are needed next week, it can still be judged irreversible within this working-capital window.
Difficulty of recovery does not equal irreversibility. If a reservation can be canceled by noon Monday at a cost of one recorded hour and without delaying existing work, it is more accurately recoverable or hard to recover. Classification needs deadline and evidence; rhetorical force cannot decide it.
Why Hard-to-Recover Consequences Need Their Own Category
A binary reversible/irreversible division pushes many real costs into the reversible side. Even an action that can ultimately be withdrawn may occupy money, consume a relationship, require rework, and close other windows while recovery proceeds.
Gu Ning required a hard-to-recover item to answer four questions: who initiates recovery, by what latest time, what additional labor or working capital it requires, and which capacity is unavailable during recovery. Without answers, “we can sell it later” or “we should be able to reschedule” is only a wish.
Hard to recover does not mean automatic rejection. If the goal is sufficiently important, the bad scenario still preserves the floor, and recovery responsibility is clear, the plan can enter another comparison. The separate category puts withdrawal cost inside the plan so that an exit button cannot impersonate costless exit.
Unknown Does Not Automatically Mean the Worst
D17 contains many unknowns: whether the client really wants a new version, its scope, delivery and acceptance, price, and payment. Unknown means corresponding consequences cannot yet be quantified; it does not prove every value will be worst-case.
Treat every unknown as infinite loss and no action is possible. Treat each as zero and deep preparation appears free. A more reliable response limits the depth of action, preventing unknowns from crossing the survival line before material arrives.
A time-bounded clarification does not require all future costs to be known. Advance purchasing turns an unknown specification into actual occupation and therefore needs more concrete conditions. Both actions face the same unknowns through different exposures. Unknowns are managed mainly by changing action structure, not assigning an arbitrary extreme probability.
Low Probability Does Not Shrink the Full Consequence
If a consequence receives a one-percent estimate, expectation weights it at one percent. The survival check still asks whether, if realized, it makes the eight protected units unavailable or prevents completion of existing work. Multiplication by one percent does not reduce the realized consequence to one percent of its magnitude.
This does not require rejecting every one-percent event. Commitment can be reduced until full failure is bearable; an effective transfer can be built; or, where a goal is important and no alternative exists, a person with authority can assume it explicitly. The requirement is that an average not secretly delete a severe consequence.
Probability must remain faithful to material. Severity does not permit changing one percent to ten, and attractive gains do not permit reducing it to one in ten thousand. Recording estimates and floor adjudications separately reveals which field new material should change.
Low Probability and No Evidence Are Not Interchangeable
“We have never seen this failure” is sometimes rewritten as “failure probability is low.” Without an observation count, object range, and detection capacity, the rewrite has no basis. Zero observations can coexist with low probability, insufficient sample, missing records, or newly changed conditions.
D17 has no Version 2 and no valid batch of similar actions. Gu Ning neither marks the probability of incompatible advance materials as “small” nor calls it certain. She marks the path condition unknown and blocks the purchasing action at the action level.
When specifications, return conditions, or evidence of general applicability arrive, the consequence range may narrow. What changes then is material and exposure, not a claim that the prior unknown secretly equaled a low number.
Protection Counts Only If It Works in the Bad Scenario
Someone might say failed preparation can be reused elsewhere, a cash shortage covered by payment from the new order, or a capacity conflict solved by temporary help. Each could become a recovery path, but it must remain available when needed.
Repurposing requires a corresponding task and time; future payment requires established payment terms; temporary help requires authority, capability, and attendance. An inflow existing only in the favorable scenario cannot protect the failure scenario. A backup triggered by the same condition as the original failure may fail with it.
The chapter does not require a backup for every catastrophe. It refuses to treat a name as a function. Each protection on D17's survival card states trigger, resource, time, and evidence. Protection not yet obtained remains a proposal and cannot be deducted from losses.
The Bearer Cannot Hide Inside “We”
The station's accounting loss, Tang Ke's communication time, Ye Cheng's existing jobs, and Gu Ning's authorization responsibility are not identical. If a plan attributes benefits to the station while assuming individuals will bear overtime and default pressure, a positive aggregate hides transfer among subjects.
Gu Ning stipulated that changes touching the six slots require confirmation by the corresponding responsible person and resource authorization from her. Tang Ke may draft and send a clarification within authorized scope, but cannot reserve a bay or purchase. Ye Cheng may report conflicts, but cannot manufacture capacity from unconfirmed overtime.
The subject field is not intended to turn every small action into a complex approval. It brings the people who actually bear floor consequences into judgment. Willingness expressed by someone without authority does not dissolve commitments the station has already made.
Hard Constraints and Risk Preferences Remain Separate
The eight-unit protection and six-slot fulfillment are hard constraints in the current record. They state the minimum conditions for a plan to become a candidate. Gu Ning's greater caution about volatility and Tang Ke's willingness to try once more are risk attitudes and action preferences, expressed when plans that pass the constraints are compared.
Put preference into hard constraints and every disagreement becomes “you do not value survival.” Reduce hard constraints to preference and passing excitement or anxiety can overturn them. Separation lets the group debate the basis of protection and whether a compliant plan is worth taking.
A hard constraint is not eternal truth. It can be revised when existing work finishes, protected uses are released, or the deadline changes, but the old version and reason must remain. Risk preference can also change without being backfilled into the past decision.
The Floor Cannot Answer for Every Goal
A plan with no contact and no input probably preserves the resource line but may miss a clarification opportunity. One vague question—“Are we still doing it?”—may be cheap but fail to answer scope or timing. Survival checking does not automatically select these actions.
It first excludes plans that break the capacity to continue, or requires them to return after conditions change. Remaining plans must still compare primary goals, information results, relationship expression, opportunity occupation, and deadlines.
“Safest” cannot become “best” without definition. Complete inaction has opportunity consequences, and permanent refusal of reasonable probes may weaken operations. Chapter 5 specifies which failures may enter trial and error; Chapter 6 combines the floor and goals into a provisionally adopted action.
Past Spending Must Not Lower the Floor
If the station has already spent a day preparing D17 and then discovers that continuation requires protected resources, the earlier input does not lower the eight-unit line. Occurred costs enter the record; the next step still compares future incremental consequences.
The actors may find exit “too wasteful” and rename an unacceptable occupation as a special case. That does not improve the chance of success; it lets past spending determine future survival conditions. Unless new facts change the protected object itself, the floor remains.
The case has no such input yet. Stating the point now prepares staged action: one failed probe may end without authorizing the next segment through its prior cost. Chapter 17 returns to sunk inputs and continuation; here we fix the check that does not move with them.
A Bad-Scenario Stress Check
Gu Ning used scenarios without probabilities to check three candidate actions. First, send one time-bounded clarification. If the worst result is no reply, communication time is capped, and neither a bay nor purchasing is committed, it touches neither the eight units nor the six slots. Its adverse result is currently bearable.
Second, reserve all four available slots. If no version forms, existing six-slot commitments remain intact, but the station loses all capacity for new arrangements while waiting. Recovery depends on timely cancellation. This is at least hard to recover and needs a concrete cancellation point, not merely “release if needed.”
Third, assume a version will arrive and spend five units on materials. Only four are deployable, so the fifth enters the protected eight. However attractive success might be, the action crosses the present floor and does not enter adoption comparison.
These judgments use no success probability. They compare the complete bad scenario with present bearing conditions. New material may permit redesign of the second and third actions, but cannot make Saturday's gap already solved.
Stress Checking Does Not Predict That the Worst Must Occur
Listing no reply, difficult cancellation, or incompatible materials does not mean Gu Ning predicts them as certain. A stress check asks conditionally whether the protected object survives if the path occurs.
It does not remove desired success from the table. After a plan passes the floor, successful consequences and probabilities still enter value comparison. Looking only at bad scenarios loses sight of reasonable gains; looking only at averages omits whether the subject remains eligible to continue.
Unknown paths need bounded stress. Gu Ning does not imagine an infinite chain of losses. She follows the action's actual commitments through direct consequences, identifiable subsequent obligations, and the present deadline. Fear beyond the causal path does not enter the table; occupation created by the action cannot be removed merely because it is uncomfortable to see.
A Recovery Path Must Be Startable
An effective recovery path contains at least a trigger, initiator, usable resources, completion deadline, and restored state. For example: “If no complete reply exists by 12:00 Monday, cancel the reservation; Gu Ning confirms; release four slots; create no obligation beyond communication time already used.” This is more checkable than “exit if things go badly.”
D17 has not decided to reserve four slots on Saturday, so the sentence remains a structural demonstration. If the future action card contains no reservation, it should not invent cancellation steps. Recovery paths correspond to actual actions, not to a desire to fill a template.
Recovery need not return to the origin. Communication time has passed, a relationship impression may have formed, and other opportunities may have moved on. State the feasible state to which the system returns instead of claiming that every trace disappears.
How the Floor May Change
Gu Ning established three legitimate sources of update. First, a protected obligation is actually completed or changed with confirmation by the relevant party. Second, a new reliable resource has arrived or is usable when required. Third, the action is reduced, staged, or supplied with an effective recovery arrangement so that the bad scenario no longer touches the line.
“We feel more confident,” “there should be enough time,” and “we have already prepared a lot” are not update sources. They may change subjective judgment or emotion, but cannot alone change facts about labor and working capital.
Every update preserves its time. Material obtained Monday can change Monday's version, but cannot be backfilled as Saturday's protection. Outcome review can then assess whether the decision respected the line under material available at the time.
Decisive Failure Must Also Be Named by Scope
This book uses “decisive failure,” but each use must state what it decides. For D17, it means an action consequence that, within this round's deadline, prevents fulfillment of committed work or retention of necessary protection and lacks a bearable recovery path.
It does not mean the end of a person's life, an organization's history, or all future possibilities, nor prove that the failure can never be remedied. The more specific the scope, the more an actor can reduce commitment, delay a promise, or build recovery. The more grandiose the term, the more it becomes an untestable threat.
No reply is not a decisive failure. It may close the clarification step or temporarily close D17, but the station retains resources and can handle other work. Closure of an action object and loss of the subject's ability to continue are separate.
The D17 Survival-Condition Card
At 13:30, the three condensed the discussion into one card. Protected object one: six committed labor slots in the next working week, unless the relevant arrangement leaves a time-stamped change record. Protected object two: at least eight of the twelve working-capital units remain available when existing payments, basic operations, or unexpected rework need them. Protected object three: after stopping a new action, the station retains at least one bearable recovery path that can begin within this round's deadline.
Unacceptable triggers include an action that directly or through an identified obligation delays existing work; makes the eight protected units unavailable when needed; or relies for exit on payment, personnel, or a repurposing object that does not yet exist. Hard-to-recover triggers include cancellation requiring new negotiation, prolonged occupation of available slots, or a not-yet-confirmed destination for materials, provided the bad scenario still stays above all three lines.
Ordinarily adverse results include no response to bounded communication, limited time consumed within authority, and stopping on time without the expected information. The classes may change with new facts. Gu Ning owns updates, Ye Cheng records task conflicts, and Tang Ke preserves communication execution and original replies.
How Present Candidates Pass the First Screening
No further contact consumes no new resources and passes the survival check, but its fit with the goals of a suitable new commission and honest negotiation remains unjudged. Waiting for complete conditions also passes the resource line but needs an expiry; otherwise waiting becomes delay without adjudication.
One clarification with explicit questions and a time ceiling, reserving no resources, has a full failure consequence of limited communication and no reply. It may therefore enter value comparison. A bay reservation needs a cancellation time and release owner. Advance purchasing or deep preparation cannot describe recovery while specifications are unknown, and directly fails if it exceeds four deployable units.
“Enter value comparison” does not mean an action is authorized or a message sent. It means only that the candidate's identified bad scenario does not break present survival conditions. The next chapter must still state the questions, authority, deadline, and outcome branches.
Saving the Boundary at 13:40 on Saturday
At 13:40 on Saturday, Gu Ning saved the D17 survival-condition card. The client still has not sent new content. The station has formed no Version 2, reserved no bay, bought no material, and sent no clarification. The eight units and six slots remain in their original states.
The card prevents several kinds of failure from being conflated: ordinarily adverse results can be tried at limited cost; hard-to-recover results need concrete recovery arrangements; unacceptable results break current protected conditions; and irreversible within the deadline means no actual path exists when recovery is needed. Unknown remains separate, neither impersonating low probability nor expanding automatically into unlimited worst consequences.
When available margin in the ground of possibility enters Chengwan's scale of observation, it must be realized through actual capacity, protection, authority, and recovery paths. Sustainable decision-making does not promise realization of every local goal. It first excludes decisive failure that would cancel the subject's eligibility to continue. While that floor and feedback capacity remain, stopping, reducing, learning, and reentering form paths of success at different levels without disguising one non-delivery as goal fulfillment.
The next chapter makes one clarification that passed the floor concrete. Its goal, questions, communication ceiling, owner, deadline, and reopening conditions enter the first action card. Only adoption of that card grants authority to execute, and adoption still does not mean Monday's message has already been sent.