Neither Window Was Completed
V2 paused at its conditions gate. V3 stopped after the technical work because the handoff version was wrong. Neither window achieved item-by-item acceptance, and neither final payment was triggered. The primary commercial goal failed twice in succession.
This manuscript does not let the corrected archive package overwrite V3, nor does one successful internal drill interrupt the sequence of external failures. Once an outcome event has been adjudicated within its original window, it remains with its original object.
Stating failure clearly is not pessimism. It separates a problem of conditions from a problem of execution, gives resource losses a ledger, and prevents a next version from acquiring an attractive history by deleting the past.
A new commission may still appear. An open future does not change an outcome that has occurred, just as a failure does not prove that the future contains only failure.
What Relations Did Action Add to Judgment?
The probability record in Part 1 required the object, material, adoption, and result to remain retrievable. Part 2 adds goals, consequences, survival lines, budgets, stages, shared exposure, waiting, thresholds, and paths.
These relations prevent action from becoming simply “believe the probability and act.” A value of 0.5 did not decide whether to send a four-question message. Passing G1 did not authorize all of G2. A cash total did not make every unit freely usable.
Action also creates information. Thursday's investigation did not observe the system from outside; it consumed one capacity slot. Waiting protected E4 but reduced the handling time remaining before the original window. Knowing and intervening met in the same process.
Action judgment must therefore reconcile two questions continually: what did we learn, and what did we change?
Margin Does Not Mean Spending Whatever Remains
The eight protected cash units and recovery labor slot ran through D17. They did not produce success, but they kept both failures from penetrating existing commitments.
Preserving resources is not inherently correct. What a margin protects, when it can be released, and how it is recalculated across periods must be explicit. An unused slot does not accumulate into the next period, and the eight protected units cannot be used to preserve an attractive ledger by refusing actual obligations.
Sustainability does not mean never consuming resources. It means knowing before consumption whether the adverse path still lets the subject act again. Choice after failure comes from conditions preserved in advance, not from an optimistic slogan.
Thresholds Make Rules Visible
h may move gradually and d may change continuously, yet 2.0 separates passing from failing. A window deadline likewise maps minutes into on time and late.
A discrete consequence does not prove that a natural process breaks suddenly at the same point. Every threshold has a proposer, a protected object, an applicable version, and an action after crossing. Changing a threshold changes the rule and cannot secretly repair history.
Likewise, one slot and eight units are boundaries of action in this case, not natural constants. They function because they connect to concrete commitments and consequences of recovery.
Neither Good nor Bad Results Own the Explanation
After V3 failed, the entry judgment could still have been supported under the rules then in force, and the two-person check did prevent the wrong package from being sent externally. The plan had a gap in error prevention, execution actually deviated, and observation discovered it in time.
The first T22 drill succeeded, but the recent-files shortcut did not identify the object. Changing only the list order produced failure. A good result did not make a bad rule good.
Result, judgment, plan, execution, and observation need separate assessment. Separating the levels does not sever their relations; it locates improvement at the interface that actually broke.
If judgment receives credit for every benefit while luck receives blame for every harm, the actor can never learn. If one bad endpoint condemns every level, effective controls are deleted as well.
Recovery Does Not Restore a Past Without Failure
When D17 ended, cash stood at eleven, time had been consumed, and two failures to deliver remained in the relationship history. The team could not return to Saturday's twelve units and four open slots.
They could settle the old object, protect the floor, repair the single interface, inventory the next period again, and leave the unresolved mechanism of conditions in its proper place. Recovery means regaining eligibility to assess from a new state.
Eligibility does not require continuation. When no new object exists, remaining closed is the present action. When a new proposal arrives, scope, deadline, resources, and exit must be established anew.
Repeated failure does not automatically generate a failure probability or demand a third attempt at reversal. The count triggers a check; material must still answer the question of mechanism.
How Success at Different Levels Can Coexist
RC praxis holds that when room for error is secured and decisive failure excluded, the remaining possibilities lead toward success at different levels or degrees. D17 unfolds this proposition at the scale of action: the delivery goals of V2 and V3 failed and cannot be renamed. Yet the eight-unit protection, existing commitments, and recovery entrance remained, so failure did not remove the subject's eligibility to keep observing, revising, and creating a separate object.
“Levels of success” does not replace a commercial result with the consolation of learning. It first specifies the scale of observation. At the scale of a single window, the result is non-delivery. At the scale of subject stability, preventing spillover, preserving feedback, and restoring choice realize sustainable decision-making. These are different objects of judgment and can therefore both hold.
If decisive failure penetrates the protection and removes the subject's capacity to act again, local experience is insufficient to constitute sustainable success. D17 turned two local failures into conditions for continued evolution precisely because it preserved redundancy, low-loss paths, heterogeneous candidates, and channels of correction in advance.
How a Handoff Becomes a Result
Lin Jian did not participate in D17. On first reading the template, he interpreted the three unallocated units as immediately usable and almost treated the correct archive package as a file waiting to be sent.
These misunderstandings show that what appears complete to an original author may still depend on memory. After the template was revised, he could locate the absence of an active object, cash authority, the model's final state, failure types, unresolved questions, and the entrance for a new proposal.
Passing the handoff does not mean every future reader will make the same value choice. It means disagreements can begin from the same facts and boundaries without guessing the missing state.
The action record thereby becomes one result of the process: it allows judgment, loss, and unresolved matters to continue beyond the original participants.
Order Is a Retrievable Relation
Finding order in chaos does not mean predicting every change or calculating an optimal answer for every action. It means knowing which object contains the present question, which material was usable at what time, and which commitment grounds each line.
It also means knowing when to stop. Stopping is not a failure button outside the process but part of action design. Recovery is not a story of success but a set of steps with completion material.
D17 does not leave a perfect ending, but it has a definite final state: old objects are closed, the floor remains intact, losses are recorded, local capability has formed within limits, and the next action has a trigger.
That is enough for action to keep receiving reality. The next choice need not begin with forgetting, and past commitments need not answer it in advance.