What Is Missing from “Only One Last Step”
At 16:18 Thursday, the other party replies to the explanation of the pause: if Chengwan can provide a one-page revised arrangement before 10:00 Friday, the parties can discuss receiving the work again from 14:00 to 16:00 next Tuesday. The original total fee of six units remains unchanged, the final payment is still triggered after item-by-item receipt, and no additional advance will be paid.
This does not mean a new window has been established. The message says “can discuss.” It confirms neither object conditions, environmental entry criteria, nor whether the existing stage result may be carried forward, and it does not settle the remaining unit of the initial payment. It only forms a new proposal with a deadline.
Looking at the G2 record, Ye Cheng says, “Of the two segments, we have finished one, and we spent another slot investigating. There is only one last step.” The statement captures historical progress but does not list what remains between the present and a new delivery.
This chapter does not decide whether every project should continue. It examines only D17’s incremental choice at this moment. Labor, funds, and readings remain fictional units within the case.
Historical Inputs Must Be Recorded
G1, G2’s first segment, and Thursday’s investigation used one slot each, for three slots in total. G1 and G2’s first segment spent one unit each, for two units in total. The four-question message used another twenty-five minutes, while Thursday’s threshold diagram and path ledger used management time.
These inputs explain how cash and capacity reached their current state. They also support evaluation of the original plan, budget estimate, and stage design. Deleting them from review would leave the team unable to tell what one attempt cost.
Yet neither canceling nor continuing the next step will turn the three elapsed slots back into uncommitted capacity or return the two spent units to cash. They matter to historical accounting, but their size does not automatically authorize a future choice.
“Cannot decide continuation” does not mean historical cost may not be mentioned. The columns answer different questions: history asks what has happened; increments ask which path to choose now.
Sunk Means the Current Action Cannot Change It
At D17’s present decision point, a cost is sunk in this comparison if neither continuing nor stopping can recover it through the choice now being made. The consumed labor and the two process units already paid meet that condition.
“Sunk” describes alterability. It does not say the original decision was foolish or that the input left no result. The G1 reading, G2 stage state, and Q1 evidence all arose from these inputs.
The same item can be classified differently at different decision points. On Monday morning, the one-unit G1 expenditure was still an avoidable future cost. By Thursday it had become history. The classification must carry a time.
If a refund right, resalable material, or cancelable reservation exists, that part may not be sunk. The words “already paid” do not justify reducing the entire amount to zero.
Existing Results Are Not Negative Historical Costs
Someone may subtract “learning value” from the three slots in an attempt to derive a net sunk amount. The problem is that there is no reliable unit for converting S-3 records, a relational explanation, or the reusability of partial work into labor slots or units of working capital.
A more robust method uses separate columns: resources have been consumed; results still exist; and the future action in which a result can be used requires separate verification. This neither declares learning worthless nor invents a price large enough to cancel the loss.
For example, G1’s 1.6 and the Q1 pairs can help design a new entry condition, but they cannot fund next week’s operations. If G2’s first segment can be carried forward after review, it may reduce future work, but one full slot cannot yet be deducted from the next estimate.
Reuse value appears only within a specific plan. The same result may serve merely as an explanation in a closing settlement, alter the process in a revised delivery, and have no direct use in an unrelated task.
The Conditional Balance Is Neither Sunk nor Free Income
Of the current thirteen units, eight are protected, four are the original unallocated capacity, and one is the conditional balance remaining from the initial payment. That last unit is still in the account and therefore is not a spent, sunk cost.
Its disposition is unclear. Stopping may require a refund or settlement; continuing may permit it to cover revised work; the parties may agree to another use. Every case requires confirmation of conditions.
Treating it as unrestricted earned income makes continuation look too cheap. Treating it as already lost artificially raises the cost of stopping. Gu Ning places it in the “resources pending disposition” column.
Whether funds exist, whether they may be used, and what their use counts toward are three separate questions. A net balance cannot replace them.
The New Proposal First Produces Three Candidate Actions
Action S ends V2 and settles it. It requires organizing existing records, confirming disposition of the one-unit balance, and stating that no later production will be scheduled. Estimated management time is forty minutes. The actual refund remains subject to agreement, so the cash consequence is blank.
Action P submits only a one-page revised arrangement, attempting to turn the proposal into a version that can be judged. Drafting is estimated at forty-five minutes, uses no half-day production slot, and spends no working-capital unit. The document must state that sending a draft does not commit Chengwan to execution next week.
Action R directly commits to resume and deliver. Current material cannot produce a complete estimate: next week’s capacity has not been inventoried, and the stabilization step, reusability of G2’s first segment, and receipt rules remain unconfirmed.
The actions carry different depths of commitment. S and P can be defined from current information. Even R’s boundaries are incomplete; prior completion of one segment cannot make it executable by assumption.
Future Increments Start at the Current Point
For action P, the new direct consumption after 16:18 Thursday is the time required to draft, check, and send the arrangement. The past three slots and two units are not added to P’s incremental cost.
P may produce an explicit new version or an explicit refusal. Its possible losses are management time, misunderstanding, and the other party’s belief that Chengwan has already committed. Explicit status can reduce the last risk, but it cannot be assumed to be zero.
For action S, new cost consists of settlement communication and a possible refund. Benefits include releasing future attention, closing the conditional balance, and ending new exposure. The refund amount is unknown, so S still requires clarification, but this does not make the action undefinable.
Action R lacks inputs and cannot currently support an incremental table of equal quality. Unknown does not mean zero cost, nor does it justify assigning infinite cost immediately.
A Project May Lose Overall While Its Next Step Is Still Worth Taking
Consider a demonstration separate from D17. A project has already spent nine units. Stopping creates no further expenditure. Continuing costs one more unit and is certain to recover three new units. The whole project still ends at negative seven, but from the present point continuation adds two units relative to stopping.
The demonstration shows that overall historical profit and the next increment can point in different directions. It does not show that D17 will gain three units, nor does it make money the only objective.
If an action crosses the survival line, violates a commitment, or creates an unacceptable consequence, a positive financial increment still cannot authorize it. Incremental comparison remains constrained by Chapter 5’s floor.
“The project has lost overall, so stop now” and “only a little remains, so continue” both compress different questions into one sentence.
A Project May Be Profitable Overall While Its Next Step Is Not Worth Taking
Now consider another independent demonstration. Earlier activity has produced twelve units of net recovery. Continuing costs five new units and can add at most two units, with no contribution to any other objective. The project remains profitable if it stops, but continuation worsens the present position by three units.
Historical success does not give later actions a permanent pass. A stage having passed once does not establish that a new environment, window, and capacity still satisfy their conditions.
D17’s G1 pass is a past-stage result. Wednesday’s failure and the closed window have changed the conditions of the new action. The 1.6 cannot overwrite 2.2 and 2.3.
The unit of judgment must match the current action. Project reputation, past profit, and completion percentage may enter the background, but they do not supply values for future increments.
Completion Percentage Often Lacks a Stable Denominator
“Half completed” assumes that the original two G2 segments still constitute the current total workload. The new proposal may require condition stabilization, review of the existing segment, renewed version confirmation, and scheduling a new window. The denominator may already have changed.
Even if fifty percent of the physical work is complete, the remainder may hold more uncertainty or shared dependencies. Equal duration does not imply equal consequences.
Conversely, learning already acquired might shorten the final step. Both directions require current process evidence; neither follows from a completion percentage.
Gu Ning removes the progress bar from the continuation page. She replaces it with four entries: completed and verifiable, reusability pending review, newly required, and conditions unsettled.
Stopping Does Not Make Historical Inputs Wasteful
Historical inputs have already occurred. Whether they were “wasted” must be reviewed through the results they left and the original objective, not defined by whether the next step continues. Stopping does not spend the two units a second time.
If the team adds action merely to prove that the input meant something, each new round can become the reason for the next. Even if every round yields a little learning, the project may continue shrinking its recovery margin.
After stopping, the S-3 process, condition-recording method, and failure material can still be archived, although transfer remains a separate question. After continuing, they do not automatically become universally valid.
Making the past meaningful is an interpretive task, not a resource authorization. Separating the two lets the team evaluate the original plan honestly.
Stopping May Still Abandon Recoverable Value
Rejecting sunk-cost distortion does not mean stopping has no opportunity cost. If G2’s first segment can be reused after a small review, stopping may forgo its future contribution. If the other party will form a suitable new window, stopping may also forgo that opportunity.
Both belong in the future difference between stopping and continuing rather than being discarded with historical cost. The key is to attach conditions and required actions to every recoverable value.
At present, “reusable” is unverified and the “new window” is unformed, so both remain conditional. Gu Ning does not count possible value as certain return in advance.
The attention released by stopping is likewise a real consequence, even without a ready monetary scale. The incremental page allows monetary and nonmonetary entries to stand together.
Split Continuation at the Next Available Stop
Action R spans too much. The team splits it into P0, drafting the revised arrangement; P1, jointly confirming object and window; P2, inventorying next week’s capacity and the condition entry; P3, reviewing whether the existing stage can be carried forward; and only then P4, deciding whether to resume production.
Stopping is allowed after every node. The present decision therefore compares spending forty-five minutes to form P0, rather than committing now to delivery next week.
Decomposition must not deliberately hide high later costs. The P0 page still states that entering production would use next week’s capacity, may require new working-capital expenditure, and remains constrained by the eight-unit safeguard and recovery margin.
The role of a minimum step is to obtain information needed for the next decision at limited cost. It is not to fragment a total commitment and induce acceptance one piece at a time.
The Revised Arrangement Must State What It Does Not Commit
The one-page draft is titled “Proposal for Conditions to Reopen V2” and marked station draft. It states that the original Wednesday window has closed and does not rewrite B_D. Next Tuesday’s window exists only after joint confirmation.
The draft asks four questions: whether the original scope still applies, how G2’s first segment will be recognized, how the operating-condition entry and review will be written, and how the remaining one unit of initial payment will be handled. It also requires an inventory of next week’s capacity before any production commitment.
These are not the same questions as Monday’s four-question message. Monday established the initial version. The present questions address a deviation, a stage result, and a funding state that now exist; copying the old questions would only imitate progress.
Stating that the draft “does not reserve capacity, prepare materials, or commit delivery” limits P’s relational risk and gives the other party room to accept, revise, or refuse.
Four Regions of the Continuation Page
The first region is history: three slots, roughly an hour or more of management time, two spent units, and nondelivery within the original window. These support total-project review and do not enter P’s incremental resource total.
The second is currently available or pending disposition: four originally unallocated units, one recovery slot this week, one conditional-balance unit, and existing stage results. The eight safeguard units appear separately as unavailable for discretionary use in this round.
The third is future increments: P requires forty-five minutes. Later production labor, condition handling, review, and cash remain blank until a new version and capacity inventory exist. The three-unit final payment remains a conditional inflow.
The fourth is action consequences: P may produce an explicit version or refusal, S closes and settles, and R cannot currently be authorized. Totals may not cross between regions.
Compare Stopping with Asking, Not Emotion with Numbers
Stopping may make the team feel that its earlier effort never paid off. Asking may provide the relief of “another chance.” These feelings can be recorded because they affect execution.
But emotion cannot appear only on the stopping side while numbers appear only on the continuation side. Asking also prolongs the demand on attention; stopping also preserves evidence and releases capacity.
Gu Ning requires both actions to use the same fields: new time, cash, capacity, commitment, information result, worst acceptable consequence, and next stop.
A symmetric format cannot eliminate value judgment, but it exposes omissions. The team still must decide whether forty-five minutes is worthwhile; the table cannot supply its objective.
A Past Forecast Need Not Be Vindicated by Continuation
R17’s first version assigned 0.5 in Part I, and the result was Y=0. There is no reason to carry 0.5 into next week’s new proposal, much less pursue a successful delivery to prove that the old judgment was “actually right.”
The new object’s conditions, window, and action structure are different. If a probability input is needed, it must be modeled again after the object has formed. Until then it stays blank.
Actors sometimes treat continuation as a comeback, using a future good result to compensate for past discomfort. That mixes model scoring, project outcome, and self-evaluation.
The continuation page asks only whether the future action fits current objectives and constraints. It does not restore the reputation of a past judgment.
Ownership Changes What Can Be Recovered
Chengwan completed G2’s first segment, but its right to use it, delivery conditions, and inclusion within the paid initial amount must be traced to the parties’ version. The team cannot assume unrestricted reuse because it performed the work, nor assume the result is worthless because the other party did not receive it.
Disposition of the conditional balance likewise depends on confirmation. If the parties agree to carry it into a new version, it may reduce new cash needs; if it must be refunded, the cash paths for stopping and continuing both change.
P’s draft can turn both unknowns into explicit questions. They do not justify immediate full resumption.
Recoverability includes physical existence, technical usability, and authority over disposition. One alone is insufficient.
Opportunity Cost Must Come from the Current Schedule
Only one recovery slot remains this week. P does not use a half-day slot and therefore does not directly release it. If the team uses that slot to review G2 before the draft, it consumes protected capacity before a new version exists.
Next week’s slots have not been inventoried. The four originally available this week cannot be projected into four next week, and the final slot cannot be copied across weeks.
Only after P1 forms will the team create a capacity table for the next cycle. If a new window conflicts with an existing commitment, opportunity cost must identify the displaced item.
The completed choice to wait for E4 supplies no automatic answer to next week’s priority.
Risk Is Not Diluted by the Fraction Already Invested
Someone may argue that half the work is complete and therefore only half the risk remains. Risk does not automatically shrink with the completion percentage. The unfinished steps may concentrate conditional sensitivity, receipt, and settlement.
G1’s pass reduced some unknowns but did not eliminate d crossing the line under high-range conditions. The investigation provided explanatory material but did not establish stable production conditions.
Future risk must be redrawn from the remaining actions, shared dependencies, window, and recovery capacity. Smooth historical steps are inputs; they cannot be deducted proportionally from total risk.
If a new action may consume the eight safeguarded units or eliminate the only recovery slot, Chapter 5’s floor still blocks it directly. Prior investment does not relax that boundary.
The Stopping Baseline Must Be Written as an Actual Action
Incremental comparisons often describe stopping as “do nothing,” leaving continuation with benefits and stopping blank. If D17 stops, the team must still handle the conditional balance, preserve stage records, notify the other party, and release later attention.
These tasks require about forty minutes of management time, and the refund remains unknown. Once completed, however, they put V2 into an explicit settled state. Stopping is a path with actions, costs, and results.
Only with this baseline can P’s forty-three to forty-five minutes be compared with a real alternative. Otherwise the team compares a concrete plan with an option artificially emptied of content.
Because S has not been selected, the forty minutes remains an estimate and does not enter actual consumption. Separating estimates from facts also prevents mutually exclusive paths from both entering the ledger.
The more concrete the baseline, the easier it is to verify continuation’s actual difference.
The Final Payment Cannot Be Counted Again with the Historical Initial Payment
If the revised version retains the total fee of six units, three units of initial payment have already arrived. The maximum future book inflow triggered by item-by-item receipt is the unpaid balance of three, not the full six counted again as new income.
The remaining unit of the initial payment may cover future cost, be refunded, or be settled separately. It is a current resource pending disposition. Putting it in current cash while listing all six units as future income counts the same amount twice.
Likewise, the two units spent in the past cannot reappear as future cost. If the existing stage cannot be reused and must be repeated, the resources required for repetition are the new item. Historical expenditure and future repetition may have the same number but arise from different paths.
Gu Ning adds “already recognized,” “conditional and unrecognized,” and “future addition” columns to J17-1. The final payment remains a conditional inflow until the new object, window, and acceptance trigger all exist.
What Would Make P Not Worth Sending
If the other party had clearly refused reopening, P would answer no remaining unknown. If the draft itself would be treated as an irrevocable delivery commitment beyond the team’s authority, it should not be sent. Neither condition appears in the current message.
If drafting required the only recovery slot or immediate expenditure, P’s cost structure would also change. The present estimate is forty-five minutes of management time, requiring checks by Tang Ke and Gu Ning but no production slot.
P also depends on being sent before 10:00 Friday. After that deadline, the original proposal ceases to be the current entrance; a late draft cannot be treated as an on-time response.
Explicit invalidating conditions prevent “small actions are always worthwhile” from becoming a new dogma.
Prospective Direction Cannot Be Replaced by Past Lock-In
Prior investment belongs to a reality locked in by earlier action. It changes current resources and reusable material, but it cannot answer the next prospective question. RC’s direction of action always faces paths now being generated. Continuation judgment must therefore restart from current margin, future increments, and exit conditions.
This does not deny history; it places history in the right relation. Sunk parts form present boundaries, recoverable results change candidate paths, and uncommitted resources preserve choice. If past investment commands future continuation, the feedback loop decays into reinforcement of the existing path.
D17’s Continuation-Judgment Page
At 17:00 Thursday, the three complete J17-1. The history region retains three slots, two units, and the result of the original window. The recoverability region marks G2’s first segment as pending review, record-keeping capability as usable within limits, and the conditional balance as pending disposition. The future region contains no uninventoried figures for next week.
Action S can be defined, though settlement of the balance is unknown. Action P adds forty-five minutes and directly asks four questions that determine later plans, with an explicit boundary against committing production. Action R fails current authorization because the version, capacity, condition entry, and reusability remain unsettled.
Given D17’s supporting objectives of honest negotiation and answering explicit unknowns, and because P touches neither the eight units nor the only recovery slot, the team adopts P0: prepare a one-page proposal for reopening conditions. Adoption does not mean the other party accepts or that R will later be chosen.
The choice follows from P’s future consequences and current constraints, not from “we have spent three slots, so we must keep going.”
Sending at 17:45 Thursday
Tang Ke completes the check and sends the draft at 17:45, using forty-three minutes in practice. It states that the original window has closed, lists four matters awaiting confirmation, withholds a production commitment until next week’s capacity is inventoried, and asks the other party to confirm by 10:00 Friday whether a new version can be formed on that basis.
Sending uses no half-day production slot, spends no working-capital unit, and does not alter the conditional balance. P0 moves from adopted to executed; its result awaits a reply.
If no reply arrives Friday, no new window forms by default. If the reply confirms only the time but not the object, entry, and balance, production still cannot be released directly. Chapter 18 will establish a reopening protocol for judging complete and partial replies.
The historical investment remains in P17-1, while the new forty-three minutes enters J17-1. The next chapter will decide which changes merely update an existing model, which require a new object, and how pause, reopening, and abandonment connect through explicit triggers.