With a Baseline in Hand, Why Still No Peace of Mind
After Gu Ning saved the frequency of the eight comparable historical requests, Tang Ke asked: do we now know that R17 will very likely be confirmed? Gu Ning said she still does not know whether the client has checked the task list, nor whether the proposed receiving period is suitable. Tang Ke pointed at the four missing entries on the source table and asked again: are those the same kind of not-knowing?
Historical material not yet obtained, current conditions not yet determined, future confirmation not yet occurred—all three can be written "unknown" on the table. Yet they demand different actions. Recovering an old message cannot check the time period with the current client; clarifying the period cannot automatically complete the four historical records.
This chapter still uses the fictional Chengwan. We do not paste the unknown with a fixed set of labels able to explain all real systems; rather, within this one judgment, we identify what is missing, how the gap might be reduced, and which limitations will not disappear merely because one more message is found.
What Gu Ning needs is not permission to judge only after clearing all unknowns to zero. She needs to give the unknown an object and a destination: which parts can be looked up, which are temporarily unobtainable, which are conditions the representation has not handled, and which are cases where, under current conditions, no unique evidential basis for the outcome has yet been obtained.
To say only that "things always involve uncertainty" says nothing about the next step. To say only that "with more information we would know everything" risks describing the future as a definite answer merely waiting for data to be filled in. Both formulations settle, too early, the same conclusion for different questions.
What Happened in the Past, but the Material Has Not Arrived
R13 through R16 are registered; their original communication material is temporarily missing. The unknown here concerns the past first: which version the task used, what replies were received when, how the window ended. The relevant process has already run its course in the thought experiment, yet Gu Ning lacks material sufficient to adjudicate under the current definition.
Recovering the original messages may reduce this kind of gap. If what is found is only a later summary, the summary's object, time, and original use must also be identified. Obtaining a file does not guarantee all questions are resolved, but it may turn some previously uncheckable facts into usable material.
Gu Ning should save separately "original material not obtained" and "original material obtained but still ambiguous." The former needs searching; the latter needs interpretive correspondence or supplementary observation. If both are written only as "insufficient data," the sorter may repeatedly search files already in hand while failing to address the questions the files themselves cannot answer.
Past material may also prove ultimately unrecoverable. Then one must not back-fill outcomes for the sake of a complete record; a limited range can be kept, stating that this part of history does not support the current count. Admitting the unknown neither guarantees it will be traced later, nor directly pronounces the missing objects failures.
The key to this class of unknowns is separating factual material from its current retrieval status. Not obtained does not mean not occurred; occurred does not mean Gu Ning knows it at this moment. The event and time relations built in the previous two chapters point the data search here toward concrete objects.
Current Conditions Not Yet Determined
R17 is a different position. The repair station has drawn up the list, but whether the client has checked it is still unknown; the proposed receiving period is written into version one, but whether the client can accept it is also unknown. These are not blanks in four historical records but conditions in the current arrangement whose corresponding material has not yet been obtained.
Tang Ke may confirm specific questions with the client, and Gu Ning may check whether existing replies have already addressed a given condition. But asking may cost the client's time and may even participate in the communication process; obtaining answers cannot be treated as a free step requiring no action or support.
Questions should be as specific as possible. Asking "will you confirm" may earn only "still considering"; asking whether the list was received addresses material arrival; asking whether the scope was checked addresses object understanding; only asking whether the receiving period is suitable addresses the other dependency. The use of an answer changes with the question.
Even obtaining an affirmative reply closes only the corresponding gap. Receiving the list does not equal endorsing the whole scope; endorsing the scope does not equal agreeing to the payment terms. Gu Ning should not, because one question finally has an answer, convert all the other unasked matters into known ones.
Current unknowns can lead to limited verification, but verification does not guarantee a positive result. If the other party explicitly says the period is unsuitable, information has increased, yet version one may become harder to confirm. More knowledge and a more optimistic judgment are not the same direction.
Which Blank Did One Reply Close
At 11:20 on Thursday, Tang Ke saw a new reply on the agreed channel and passed it to Gu Ning at once. According to the action record she had saved, before this reply arrived she had not yet sent the inquiry mentioned in chapter two; the reply was sent by the client unprompted. It reads: "List received; I still need to check the receiving time." This is fictional material newly added to this chapter's main line—not a real business observation, and not a full confirmation of version one.
Gu Ning saved the original wording and the time of receipt, without back-filling it into the 10:00 initial information list. She now knows the client states the list has been received and that the receiving time still must be checked. The message's content is limited; from it one cannot reconstruct an item-by-item acceptance of the repair scope.
Tang Ke said that the other party's willingness to reply is at least a positive signal. This view may enter subsequent evidential judgment, but the signal's discriminating power has not yet been checked against condition material. Gu Ning does not at once assign it a fixed probability, nor mark the event as held because it is useful.
A single message may reduce one unknown while revealing another. Previously Gu Ning did not know whether the list had arrived; now there is a corresponding statement. Previously the receiving period merely awaited confirmation; now the other party has explicitly said it still must be checked. The table does not necessarily just fill up with fewer blanks; it may also become more specific.
Version one in the main line still has not received full acceptance. The new reply supplies condition material for judgment; it gives no guarantee for the future outcome. Updating the list of unknowns as the material changes—that is the action this chapter needs to demonstrate.
Knowing the Conditions Does Not Mean Possessing a Unique Outcome
Set a comparison first: the client explains further that the list and the period are both understood, that no changes to the payment terms are proposed, but that final confirmation still must be made within the window. Gu Ning has obtained more corresponding information, yet she still cannot record this as acceptance merely because there is no objection.
In this comparison the listed conditions can be clearer while the outcome may still vary. The client may confirm, may keep considering, or may reply only after the window. We provisionally call this variation—where no unique evidential basis for the outcome has been obtained under the current conditions of representation—outcome fluctuation.
This is only a description within the practice of judgment. It does not claim to have proven some ontological randomness, nor assert that all futures are in principle unpredictable. Gu Ning knows only that the existing conditions do not suffice to yield a guaranteed result; she cannot reason from her own not-knowing to how the whole world must be.
Nor should all remaining unknowns be classified as fluctuation. Some important condition may still be missing from the table—for example, communications unfinished within the client's own organization. If material is obtained later, the explanation can be revised; the name "fluctuation" must not become a reason to stop searching for actual causes.
Knowing more conditions can therefore make judgment better grounded without automatically converting probability into an outcome guarantee. This chapter retains that limit; the next chapter examines further why the model selects certain conditions and leaves others outside.
Not Every Difference Has Been Proven Random
Of the eight comparable historical requests, five were confirmed and three did not satisfy the condition. This difference shows that different outcomes occurred under the same set of stated corresponding conditions; it does not by itself prove that the differences were all produced by an inexplicable random mechanism.
There may be conditions not yet recorded; some window or scope may still be misread; or the current representation may simply support only outcome tendency. Each possibility must be examined along the material; one cannot assign a strength to every explanation directly from "same group, different outcomes."
If Gu Ning says the confirmation process always involves luck, she still needs to state what that sentence does here. It may remind against guaranteeing single outcomes, but it cannot replace checking the reasons for non-confirmation, data quality, and the actual conditions.
Conversely, if Gu Ning believes all differences could be explained by some hidden condition, she still cannot write a condition not yet found as one she already possesses. That an explanation may hold and that an explanation has material are different positions. Expecting to find the cause someday does not supply today's definite grounds.
This book does not use the repair station's example to adjudicate determinism or natural randomness in general. What we handle is judgment within limited observation: which conditions have been obtained, which explanations are unproven, and how strong a conclusion the existing representation can support.
Not Knowing Is Not Fifty-Fifty
Tang Ke said: since confirmation and non-confirmation are both possible, is it now half and half? That neither outcome has been excluded means only that both remain within the range under discussion; it does not mean they carry equal judgment strength. The number of options cannot directly generate a probability distribution.
Gu Ning already has the eight-record historical frequency and the new reply, each with a limited role. If everything is switched to one-half merely because unknowns remain, the material already obtained would be deleted as well. Honestly stating that one does not know does not require every unconfirmed object to adopt the same neutral number.
Likewise, having a baseline does not justify ignoring the unknowns. Five eighths is the counting result of that historical set; adopting it for R17 still requires a domain bridge. How the message changes current conditions, whether the missing items affect the comparison range—no old ratio can settle these on their behalf.
This chapter does not compute a final probability for R17. We only exclude two unsupported substitutions: inferring equality from two options, and canceling all historical material because unknowns remain. If a number is to be adopted, it should state the conditions and assumptions it comes from.
If material is truly insufficient, a qualitative judgment or a range restriction can be retained; one need not fill in a deceptively neutral number just to complete a cell. Formal completeness differs from a supported judgment; the unknown should exist in the statement, not be covered up by "one half."
Uncertainty About the Estimate, Versus the Outcome Possibly Failing
Gu Ning may also lack confidence in the estimate she adopts. A small historical group, incomplete sources, unchecked condition differences—these make her unsure whether the estimate fits the current case. This differs, at a different layer, from an already-adopted probability whose outcome may still fail to hold.
Even if an estimate later proves quite useful, a single event may still fall on the less frequent side. One non-confirmation does not automatically prove the estimation method wrong; one confirmation does not automatically prove the comparison class and evidence were used correctly. Evaluation requires more corresponding records.
Conversely, adopting a high number does not mean Gu Ning is confident in the estimate itself. The number expresses the target event's tendency; material quality and applicability conditions state what support this representation has. The two should not be conflated into a single degree of self-assurance.
This chapter does not write this estimate uncertainty directly as a statistical confidence interval. How such an interval would be defined, and on what data and assumptions it rests, will be treated in the later delivery of judgment. For now only concrete reasons are saved; upper and lower bounds are not computed automatically from the count of gaps.
With the two layers separated, Gu Ning can say explicitly that a historical starting point leans positive but its transfer is limited; or that the existing conditions are fairly clearly handled, yet there is no guarantee for a single case. Expression need not equate caution with pessimism, nor optimism with reliability.
Does Finding More Messages Reduce the Same Unknown
Gu Ning can continue to inquire, search old archives, or collect more comparable requests. But every action must first return to its target. Recovering R13's original material addresses the historical range; confirming R17's receiving time addresses the current dependency; adding new request outcomes may help future tests of the representation.
If the key blank is the current receiving period, rereading the display cards may add nothing. As material multiplies one may feel better prepared, while the actual gap is unchanged. The quantity of informational labor does not substitute for object correspondence.
A new message may also be merely a retelling of old words. Tang Ke hearing once more that the other party is "willing to consider" does not amount to obtaining two independent bases of confirmation. Shared sources will be discussed in detail in Part Two; this chapter asks only whether the message actually addresses the original blank.
The cost of verification needs identifying too. To learn a private detail irrelevant to the confirmation conditions may add burden without changing the judgment. Gu Ning should state why a piece of data is needed, rather than claiming unlimited rights of demand on the grounds that all unknowns deserve elimination.
A useful action should be able to point to which gap it intends to reduce, what material it may obtain, and how the judgment updates afterward. If that cannot be said, narrow the search first, rather than covering "not knowing what to do next" with more gathering.
If the Explanation Is Unclear, What Alternative Descriptions Remain
For the unconfirmed receiving period, Gu Ning can save several possibilities: the client has not yet checked it; the period genuinely conflicts; or the reply is simply not yet complete. These are explanations to be tested, not established facts; no definite value may be assigned to each at will.
Different explanations point to different follow-up questions. If there is a real conflict, one needs to know whether it can be adjusted; if the item is merely unchecked, one needs to know whether an arrangement can be obtained; if a stage of the reply is missing, the communication status should be saved. A classification is useful insofar as it elicits corresponding material.
If explanations overlap, that too should be stated. For instance, "not yet checked" may coexist with "an existing time conflict"; reality should not be forcibly divided into three mutually exclusive cells with one cell excluding the others. Convenient classifications must still submit to the actual process.
Keeping several explanations does not mean declaring them equally reliable. Existing messages may support one; contrary material may shrink another; explanations lacking evidence remain hypotheses. Processual completeness allows continued calibration, rather than granting every imagination the same standing.
Nor does Gu Ning need to enumerate every possible cause before continuing. Keeping a few competing explanations around the current key blank, and stating which materials would distinguish them, is enough. Endless listing may be just another form of information overload, making actual verification hard to begin.
How Current Gaps and Model Omissions Connect
Some unknowns can be expressed with existing fields—for example, whether the receiving period has received explicit acceptance. Other questions never entered the current representation—for example, several requests sharing and occupying a stretch of communication time. The latter may require adding conditions, not merely filling one more blank.
Gu Ning has not yet shown that R17 is affected by such shared conditions. She may list it as a model limitation to be checked; she should not claim a cause discovered just because the theory permits omissions. A model gap must be formed through material related to the specific object.
If an important condition is later found, the current judgment can be updated, and the comparison class may need reopening. The old frequency still describes the old set; what the new representation changes is its interpretive use for the present. Adding conditions need not render all past observation invalid.
Likewise, model omission must not become an after-the-fact exemption for any outcome. Saying "some condition was surely omitted" when the prediction fails, and "the original model was valid" when it holds, would protect the representation no matter what happens. The original conditions and the visible consequences must be saved, so revision has a real object.
The next chapter discusses model selection, compression, and boundaries specifically. This chapter completes the interface: data unknowns can change the inputs; model omissions may change what inputs are needed; the two layers should not be merged into one problem that merely requires gathering more of the same material.
Uncertainty Does Not Mean No Limited Judgment Can Be Formed
Someone proposes: since every item has gaps, Gu Ning had better say nothing at all. This advice protects against hasty guarantees, but it may conflate limited judgment with omniscience into a single standard. Actual matters keep proceeding, and waiting itself will not automatically obtain all the material.
Gu Ning can state what she knows, what remains unchecked, what assumptions she adopts, and which changes would reopen the judgment. Such a statement need not guarantee the outcome, nor make the unknown vanish from the record. It marks the current boundary rather than declaring observation complete.
Conversely, limited judgment cannot escape calibration by being forever "limited." If an estimate is adopted, its applicable object and outcomes should be saved, and later material may check it. Admitting limits lets judgment be returned to; it does not make judgment forever immune to being shown unsuitable.
Nor does this treat the inexhaustibility of the Ground of Possibility as granting Gu Ning unlimited time to wait. Her verification time, obtainable material, and costs of expression remain finite; they must be organized around the current question, not treated as if the theoretical margin were a resource guarantee.
The list of unknowns should therefore support limited delivery. Judgment may be provisionally adopted, gaps clearly retained, outcomes entering later. Only with these three connected does not-knowing become more than a phrase that stops thinking.
A Later Outcome Cannot Convert the Original Unknown into the Known
Suppose R17 is later confirmed within the window. Gu Ning may feel that the receiving period was never really a problem. The outcome shows the event held; it does not directly prove that at 11:20 she already knew the other side would accept the period. The outcome obtained later and the original gap are different facts.
If confirmation later fails, one still cannot conclude from the outcome that every unknown at the time played the same role. The four historical records' missing material may have limited transfer, but that does not mean they caused the current non-confirmation. Insufficient data, prediction mismatch, and actual cause must be kept apart.
After-the-fact explanation may propose new objects of inquiry. The client's supplementary note about a conflict may help check the conditions; yet Gu Ning cannot relocate the new statement back to 10:00, rewriting facts not yet obtained as things deliberately ignored earlier.
Saving the list of unknowns and its versions protects precisely this time relation. If a later blank is filled, it can show how the gap was reduced; if it is never filled, its original state need not be arbitrarily altered according to whether the outcome went well.
The learning direction of this chapter: bring every kind of unknown into contact with the action of obtaining or revising. Success also needs this calibration, and failure cannot have all its explanation monopolized by a single outcome.
Even When Material Cannot Be Obtained, Its Provenance Is Saved
Sometimes Gu Ning knows what to ask, but the other party temporarily cannot answer, or declines to provide details concerning private arrangements. Unobtainable differs from "no relevant condition exists," and the absence of an answer cannot prove a negative decision has been made. The gap shows only that the current observation has not completed the corresponding check.
One may ask for the limited state sufficient to support this judgment—for example, whether the proposed period is acceptable—without requiring the client to disclose their entire personal schedule. If even that state is not obtained, the unknown and the current limits are retained, rather than inferring the answer from the hoped-for outcome. Material demands should connect to the question; prediction needs must not acquire unlimited rights of explanation.
Waiting too must have boundaries. If nothing can be obtained by a certain node, one states which assumption the judgment continues to depend on, which uses are temporarily unsuitable, and that it reopens when new messages arrive. No answer cannot forever maintain an outward appearance of complete assurance, yet it also need not cancel the known material unrelated to that gap. A limited statement keeps "unobtainable" from being just one more sourceless blank cell.
Writing the Unknown List with Objects and Next Steps
Gu Ning finally records the unknowns separately. At the historical layer, four items of original material are still missing, and records with different conditions are not directly entered into the baseline. At the current layer, the client's reply says the list was received, the receiving period still must be checked, and full acceptance of version one has not been obtained.
At the representation layer, the transfer from eight historical records to R17 still requires checking key differences, and the existing fields may not cover important conditions. At the outcome layer, even with conditions supplemented, one still cannot promise in-window confirmation without grounds. Each layer points to different work; none is collapsed into the single phrase "there is risk."
The list also states the corresponding actions: continued limited search for the historical originals; confirmation of current questions within scope; model limitations left to the next chapter; outcomes saved under the window and adjudication conditions. The new reply already obtained keeps its original wording and the 11:20 time, answering for no other blank.
These divisions can help readers organize their own judgments. First ask which object an unknown belongs to and whether existing material really addresses it; then ask what can be obtained and which step changes once obtained. The parts with no way to obtain are clearly retained, not hidden behind a deceptively neutral number.
The judgment this chapter yields: not-knowing needs to be made concrete, and uncertainty needs a range. Information gaps can be reduced through corresponding material; model omissions may require reopening the representation; and under current conditions the outcome is still not uniquely determined—it cannot automatically be named proven randomness.
The Chengwan main line obtained one limited reply at 11:20; R17 remains not fully confirmed. The next chapter continues examining Gu Ning's model: which conditions are compressed into a single label, which are temporarily left outside, and how to let the representation provisionally support judgment without blocking new material in the name of its own completeness.