A Small Action Still Needs a Clear Question
At 08:50 on Monday, Tang Ke opened D17-A0, adopted on Saturday. The card did not authorize her to accept the job, quote, or schedule production. It required only that at 09:00 she send one four-question message group through the existing text channel and keep checking, sending, and archiving within one hour.
“Take the minimum step first” sounds prudent, but small does not automatically mean useful. A casual “Do you still need it?” may take one minute yet fail to supply scope, timing, and transaction terms. Half a day spent preparing a polished plan may involve no purchasing yet still exceed the information task.
This chapter tests one question only: can the station use A0 to obtain the minimum material needed to decide whether to create a V2 conditions proposal? Action size, direction of inquiry, adjudication of results, and entrance to the next step must all hold for it to count as a minimum useful probe.
The story remains a fictional derivation. Part 1 did not disclose the specific items or amounts on the Version 1 checklist, and this chapter does not invent them as new details supposedly known at the time. References to the old checklist mean only that both parties can return to the same stored file.
Minimum Does Not Mean Writing the Cost as Zero
Tang Ke must check the old record, inspect the message, send it, and archive it. A0's one-hour ceiling acknowledges that this work has a cost. If communication were called free, ten successive instances of “just one question” could disappear from the ledger.
Minimum is relative to the present question and available actions. Compared with reserving a half-day bay, purchasing, or making a sample, a clarification within one hour is a shallower commitment. Compared with no contact, it still adds attention, communication, and relationship consequences.
This chapter therefore does not call A0 zero-input. The zero is a stage number before resource commitment. It occupies no half-day production slot and uses no working-capital unit, but the actual minutes of communication remain on the result page.
First Check Whether Saturday's Assumptions Have Failed
At 08:52, Tang Ke confirmed that the original text channel remained accessible and displayed conversation history and arrival times. There was no platform failure or need to switch channels. She then asked Ye Cheng whether the six existing tasks had developed new problems over the weekend.
Ye Cheng checked the schedule: the six commitments remained in place and four slots were still schedulable. No expenditure had occurred from the twelve working-capital units, and the protected use of eight units had neither been released nor expanded. This check did not reserve the four slots; it showed only that Saturday's resource state still supported A0 execution.
At 08:58, Tang Ke confirmed that the four questions matched the approved Saturday draft and contained no quotation, delivery guarantee, or statement of readiness. All three premises held, so A0 did not need reopening before sending.
The Execution Record Begins with a Receipt
At exactly 09:00 Monday, Tang Ke sent the approved message group through the original channel. The system showed successful delivery. She saved the complete text, sending time, and channel receipt. A0 then moved from “adopted” to “executed, awaiting result.”
The text first stated that Version 1's original receipt period had ended and that this inquiry concerned a possible new version. It then asked, in order, whether the client wished to continue, what object and scope it proposed, the desired receipt or delivery window, and the proposed fee, payment, and acceptance terms. It ended by saying the station would arrange subsequent discussion at noon based on the material received.
It did not say “we have reserved Wednesday,” attach a new quotation, or promise an immediate start upon reply. Execution matched the adopted version. Successful sending alone did not complete the information task.
Preserving the Actual Cost
From opening the card at 08:50 to completing archival of the text and receipt at 09:05 took fifteen minutes. Tang Ke would also need to preserve any reply, so A0's final time had not yet been settled. The record showed only fifteen minutes used so far.
Fifteen minutes does not establish a universal duration for small actions. It is an actual stipulation in this fictional execution. Another channel, a more complex check, or additional responsible people could require a different amount and cannot use this as a communication benchmark.
Labor-slot occupation remained zero, meaning none of the four half-day production slots had been removed. Working-capital expenditure also remained zero. Separating time cost from scheduling units prevents both exaggerating it into half a day and erasing it because it is less than one.
Waiting Must Not Secretly Expand the Action
At 09:20 the channel showed that the message had been viewed. Tang Ke did not classify “viewed” as a reply or infer that the client was preparing complete conditions. It was only an interface record supplied by the platform and answered none of the four questions.
She also sent no reminder. A0 authorized one message and passive receipt of replies in the original channel until noon. Sending a second message out of anxiety would turn one clarification into continued pursuit, changing time and relationship consequences and requiring another version.
Waiting did not mean staring at the screen. Tang Ke returned to existing work and checked the channel only at agreed points. The one-hour ceiling constrained actual handling; the three-hour observation window was not recorded as three hours of labor.
The First Reply Answers Only Part
At 09:42, the client replied: “We would like to continue. Change the time to Wednesday afternoon; everything else can be discussed using the previous checklist.” Tang Ke saved the original and its arrival time without immediately asking what hour.
The message answered the intention to continue, offered a broad time expression, and referred to the earlier checklist. It did not define the boundaries of “Wednesday afternoon.” Nor did “using the checklist” show whether fee, payment, and acceptance were a proposal, an acceptance, or a starting point for discussion.
Under A0's result classes, this was a partial reply. The existence of a reply did not make all four items complete, and a positive tone did not form V2. Tang Ke listed the missing exact window and the current status of transaction terms without completing the answer for the client.
Receiving More Material Does Not Require Another Question
At 09:50, without waiting for a follow-up, the client added: “The scope remains the stored Version 1 checklist, with no new items. We would like receipt between 14:00 and 16:00 Wednesday. Use the total from that checklist; half will be paid after the new version is confirmed, and the other half after item-by-item inspection and receipt against the checklist. If you cannot arrange this, please do not prepare materials yet.”
Tang Ke preserved this text and arrival time as well. Together, the two unsolicited replies covered intention, a retrievable scope, a specific receipt window, and proposed fee, payment, and acceptance terms. The last sentence also prohibited the station from preparing materials before it knew it could schedule the work.
This did not violate the “send once” rule. Tang Ke asked no additional question; she merely received and archived material the external party supplied in the same channel. The action boundary constrained the station's behavior, not the number of messages another subject could send.
What Must Be Checked When Returning to an Old Checklist
“Use the stored Version 1 checklist” is useful only if both parties point to the same file. Tang Ke compared the file identifier, sending time, and version name in the original conversation and confirmed that the reply uniquely referred to the Version 1 checklist stored for R17.
She did not reinterpret its entries or invent undisclosed items for this chapter. Retrievability is enough for internal scope review. Readers need to know that the reference exists, not guess what particular object was to be repaired.
Had the client said only “the old items” while several lists existed, scope would remain ambiguous. The case explicitly stipulates that the identifier matches uniquely, so the second minimum condition is obtained. That is a fact of this case, not a rule that validates every shorthand reference.
The Time Expression Has Become an Adjudicable Window
“Wednesday afternoon” supplies an everyday range but not a precise receipt boundary. The supplementary message gives 14:00 to 16:00 Wednesday, with both ends explicit. The station can later connect capacity and handoff conditions to the same time object.
This window is a condition proposed by the client, not a delivery commitment accepted by the station. Ye Cheng has not reserved labor and Gu Ning has not said the schedule is feasible. A clear proposal and joint adoption are separate states.
If the parties later change the period, they must preserve a new version. They cannot erase 14:00–16:00 on the V2 card and make it appear that the final arrangement was known on Monday from the start.
Transaction Terms Are a Proposal, Not Cash Received
The client proposes the total in the retrievable checklist, half upon mutual confirmation of V2, and half after item-by-item inspection and receipt. This creates a discussable payment structure and acceptance interface; no payment has occurred.
“After confirmation of the new version” still needs a record of how both parties confirm. The station cannot treat the client's unilateral terms as completed confirmation or count the possible first payment among its current twelve units.
Item-by-item checking is more retrievable than “pay when satisfied,” but whether the checklist suffices for adjudication still needs later review. A0 obtains a proposal and does not complete every detail of a formal agreement.
The Client's Restriction Also Enters the Action Material
“If you cannot arrange this, please do not prepare materials yet” is not noise outside the four questions. It directly limits the next step. The station must not buy materials to appear proactive and hope to solve scheduling later.
The restriction aligns with Saturday's survival card but has a different source. The Saturday rule came from the station's internal protection of resources; Monday's sentence is new primary material from the client. Both support no advance preparation but cannot be merged into a single piece of evidence.
If Gu Ning extracted only the Wednesday window and omitted this line, the later stage table would overstate current authority. Preserving the full original lets unexpected information constrain action instead of retaining only the portion favorable to continuation.
A0 Expires Early When the Reply Becomes Complete
At 09:55, Tang Ke gave Gu Ning the two replies, file-reference check, and omissions table. All four minimum conditions were covered, triggering A0's early reopening condition. There was no reason to wait until noon.
A0's result was recorded as “complete set of conditions obtained within the window,” composed of two original messages. This did not mean “cooperation succeeded” or “the new version was accepted.” It meant the information task obtained enough material to define a conditions proposal.
If later review showed that the file identifier was not unique, the result could be corrected from the original evidence to partial reply. The case stipulates that it is unique, so Gu Ning can proceed to the V2 definition check.
Settling Total Cost at Reopening
Tang Ke spent another ten minutes preserving replies, mapping them to the four fields, and producing a handoff page. Together with the prior fifteen minutes, A0 took twenty-five minutes, below its one-hour ceiling.
No call, second outgoing message, quotation, bay reservation, or purchase occurred. Four slots remained schedulable and twelve working-capital units remained on the books; six slots and eight units remained protected.
Finishing below the ceiling did not authorize using the remaining thirty-five minutes. A budget is maximum exposure, not an amount that must be consumed. The question was answered; more polishing or questioning would be new input without a task.
Defining Success for a Minimum Action
A0 succeeds when one authorized message is executed within its cost ceiling, reaches an adjudicable result, and lets that result enter the established reopening interface. It does not require favorable conditions from the client.
This time the complete set arrived, so the information goal was achieved. Even an explicit refusal could have completed the identification task while failing the cooperation goal. Multiple goals cannot be compressed into one “win” after the result.
With no reply, A0 would still close at noon with less information gained. Whether adoption remained reasonable would return to Saturday's grounds. This favorable reply cannot prove that every similar clarification will work.
What Happens When an Action Is Small but Its Question Is Unclear?
If Tang Ke had asked only “Are we still talking next week?” and received “Yes,” the shorter message would still leave scope, window, and transaction terms unknown. The action would be small but would not create an interface usable by the next stage.
At the other extreme, listing twenty production details or attaching a complete preparation plan might obtain more information while turning A0 into advance design. More information is not automatically more effective at the present step; excess questions add costs and imply commitment.
A minimum useful step contains enough questions for the next adjudication without doing later-stage work in advance. The four questions are not a universal optimal template; they correspond exactly to D17's current gaps.
What One Step Can and Cannot Test
Execution tested that the original channel worked, the message could be completed within an hour, and the four questions could obtain one complete proposal. It did not test whether advance materials increase the chance of agreement or whether reserving a bay is worthwhile.
One A0 cannot estimate a general response probability for structured clarification. A message sent at nine and completed by 09:50 does not establish a fifty-minute response law for every object.
Evidence from action covers what the action actually changed. An unexecuted deep-preparation path still has no performance data, and passive waiting was not implemented simultaneously as a control. A good outcome cannot supply results for paths not taken.
Establishing the V2 Conditions Proposal
At 10:10, Gu Ning created the V2 conditions-proposal card. It links to but does not overwrite R17 Version 1. Scope references the same stored checklist with no added items; proposed receipt is 14:00–16:00 Wednesday; fee is the checklist total; proposed payment is half after mutual confirmation of V2 and half after item-by-item checking and receipt.
The card cites the two messages received at 09:42 and 09:50 and notes that the file reference was checked. The client wishes to continue and asks the station not to prepare materials if it cannot schedule the job.
State: “client conditions proposal formed; station has not accepted.” No production schedule, first payment, materials, sample, or formal delivery exists. V2 is now an object that can be judged; “formed” does not smuggle in a contract.
A New Object Does Not Inherit the Old Probability
V2 relates to Version 1 but has a new proposal time and receipt window. Version 1's 0.5 and Y = 0 remain on its old page. They are neither copied into V2 nor revised because the client replied positively this time.
No numerical V2 event has even been defined. The station first decides whether to confirm and commit in stages. If a prediction is later needed, it must define a new event, information time, and deadline.
“This reply was fast” is a process observation, not proof that V2 will be completed. Action material has increased, closing some unknowns and opening new questions about capacity, stages, and exit.
Accepting the Job Still Requires Another Judgment
Ye Cheng must return V2's scope to the checklist and estimate labor and working-capital needs by stage. Gu Ning must check the Wednesday window against the six existing slots. Tang Ke must clarify how both parties will adopt the version.
None of this belongs to A0. If the team accepted immediately because the reply was complete, the information step would upgrade itself into a resource commitment and nullify the authority boundary established earlier.
Gu Ning replies only that the message was received and that the station will confirm after checking its arrangements. This ordinary receipt does not answer whether the work can be accepted or change V2's state. Its time belongs to reopening work, not to A0's count of outgoing questions.
Information Has Value When It Changes the Next Step
A0 closes four direct gaps: intention, scope reference, receipt window, and proposed transaction terms. It also changes available actions. Permanent closure is no longer the only low-input path; preparing against unknown specifications is no longer the question; the team can design stages for a formed V2.
Information is useful because it changes an uncertainty, action, condition, or adjudication—not because the number of words increases. A courteous greeting that reduces no critical unknown is new information but may have no present action value.
Value is not determined only by whether the station ultimately accepts. Even if capacity review leads to refusal, the information makes refusal answer to explicit conditions rather than an imagined object. Feedback has already entered the choice structure.
Minimally Sufficient Does Not Mean Globally Optimal
The four questions' ability to form a V2 proposal shows they were sufficient for this interface. It does not compare them with every possible message or prove twenty-five minutes a theoretical minimum. “Minimum” is a working judgment: removing a class of question leaves an identified gap, while adding production design exceeds the task.
A global optimum requires a complete candidate set, evaluation scale, and all relevant consequences. D17 has neither such a set nor a unified utility value. Gu Ning records A0's grounds and result relative to the present candidates without presenting it as a mathematical optimum.
This restraint does not weaken action. A plan can receive clear authority after limited comparison while remaining open to a better path later. When scope, grounds, and validity are preserved, later revision does not prove that the earlier choice was groundless.
More Replies Do Not Mean More Valuable Information
Ten paragraphs of background without scope or window would still prevent capacity review. One sentence naming file A and receipt at 14:00–16:00 Wednesday might close two critical gaps. Information value follows the uncertainty and action relation it changes, not its character count.
In these replies, “we wish to continue” changes intention; the checklist identifier changes scope status; the exact window permits schedule checking; and payment and acceptance proposals allow a new consequence table. Each points to a subsequent page. Courtesies and repeated positions need not count as extra pieces of evidence.
Repeating the same content three times does not create three independent supports. Tang Ke retains the number of original messages while consolidating their functions in the conditions table. Communication facts remain intact without inflating information gain.
An Adjudicable Result Also Needs Adverse Exits
A probe that states only the desired answer, without saying how partial replies, refusal, or silence end, extends indefinitely on adverse paths. A0 defined five result classes in advance so every path could reach a place in the record.
An adverse exit does not redefine every failure as success. Explicit refusal fails the cooperation goal; no reply leaves conditions incomplete. But neither leaves the action suspended. Information-task completion, cooperation-goal realization, and actual losses are adjudicated separately.
This execution followed the favorable path, so the adverse exits were not used. They remain part of the action structure. Deleting them afterward would hide why Saturday could authorize action under an unknown reply.
The Minimum Step Implements the Low-Loss Principle
RC's continuity of stability proposes that the parts most likely to vary should remain low-cost and low-loss. A minimum step realizes this condition in the domain: the smallest commitment sufficient to distinguish a next action intervenes in reality and obtains feedback without first locking cash, capacity, and the negotiation path.
Inquiry is not passive sampling from outside the ground of possibility. Wording, timing, and channel participate in the other party's response and reshape later visible paths. Minimum therefore does not mean the least possible action, but the least loss that completes an observational intervention with a clear question.
Record of the Minimum Useful Probe
At 10:20, Gu Ning completed the record. Question: can one bounded clarification obtain the minimum material for defining a V2 conditions proposal? Action: at 09:00 Monday, Tang Ke sends the approved four-question message group through the original channel, with no active follow-up.
Actual input: twenty-five minutes of handling, zero labor slots, zero working-capital units. Result: a partial reply at 09:42 and an unsolicited supplement at 09:50; together they cover all four fields, and the file reference is unique, so A0 expires early.
Product: a V2 conditions-proposal card, state “proposed by client, not accepted by station.” Not occurred: no reuse of the old 0.5, reservation, purchasing, sampling, payment, production, delivery, or generalization of response speed into probability.
Handing Off to Stage Design at 10:20 Monday
The standard for a minimum step is now reviewable. It needs a clear question, an action sufficient to answer without unnecessary expansion, a cost ceiling and stopping point, and a path by which every result enters the next round.
This successful A0 does not prove smaller is always better. If the four questions had been insufficient, the next object could not form; if the scope required an on-site survey, a message might not be the minimally sufficient action. Minimum is always relative to the relation awaiting adjudication.
At 10:20 Monday, D17-A0 closes and the V2 conditions proposal enters a new action judgment. The four schedulable slots and four deployable working-capital units remain uncommitted, and the client's proposed first payment has not arrived.
The next chapter separates confirmation, sampling, and formal scheduling into stages. Staging is not slicing one total commitment into attractive small pieces. Each segment must answer a new question and give feedback from the prior segment real authority to stop, modify, or enter the next.