What Happened Between 4.5 and 4.7?
At 14:00 Thursday, Tang Ke plots the original h log as a timeline: 4.2, 4.3, 4.5, 4.7, 4.9, 5.2, 5.6, 5.8, 6.1. The numbers move gradually, and no material shows an instantaneous jump in working conditions.
Another table contains only pass and fail. G1's Monday d=1.6 passed; Wednesday's 2.3 and 2.2 failed; Thursday's low readings passed and high readings failed. Classification looks like a sudden switch.
The source of suddenness requires layers. Conditions may change gradually and d may move with them, while the checklist places a line at 2.0 and the action rule connects classification to release or pause. Natural process, measurement, classification, and consequence each have their own shape.
This chapter explains only the fictional structure. h, d, and 2.0 are not real material laws. The few readings cannot fit a complete response curve.
The First Layer Contains Only Observed h
The bottom layer places the nine h records at their times. Neighboring differences include 0.1, 0.2, and 0.3, with an overall move upward. No per-second values are invented between records.
Connecting points helps show order without proving a straight trajectory. h may have risen, fallen, or fluctuated briefly between observations.
This layer supports “observed times moved gradually,” not strict monotonicity or constant speed. Visual smoothness cannot exceed the record.
The Second Layer Contains Only Observed d
d is not paired with all nine h points. The mainline has G1's 1.6 and Wednesday's 2.3 and 2.2. Q1 adds a's 1.7, 2.2, 1.8 and b's 1.8, 2.3, 1.9.
The points differ in object, time, and condition. Shapes distinguish mainline from Q1, and segments connect low-high-low readings of the same piece without merging them into one time series.
Sorting every d by h and drawing one curve would hide subitem, Q1, and repeated-measure relations. Gu Ning keeps scatter points and correspondence, without claiming d is known for arbitrary h.
Only the Third Layer Adds the 2.0 Classification Line
V2's checklist says d≤2.0 passes and d>2.0 fails. The case function is:
$$ C(d)= \begin{cases} \text{pass}, & d\leq 2.0\ \text{fail}, & d>2.0 \end{cases} $$
The line does not change d. A value of 1.9 remains 1.9 and 2.2 remains 2.2; the function maps a continuous scale to two labels. Boundary inclusion comes from the rule, so exactly 2.0 passes.
The Fourth Layer Contains Action Results
The stage gate connects labels to action: an entry pass merely permits further continuation checks, while failure pauses resource release. Budget, version, and node conditions must still hold.
From d=2.0 to slightly above, the raw quantity may change little while action changes from possible entry to pause. The discontinuity belongs to the rule's consequence and need not imply an equally large jump in the measured object.
The pause then prevents Wednesday 14:00–16:00 receipt, and window closure changes final-payment conditions and future paths. One threshold can amplify into several discrete consequences along an action chain.
An Illustrative Line Must Not Impersonate an Observed Curve
For explanation, Gu Ning draws a dashed line rising gradually from lower h and d to higher positions, crossing 2.0 into a differently colored background.
The legend states “structural illustration, not a fit.” Existing points cannot establish linearity, monotonicity, or the precise h crossing and cannot predict new objects from the dashed slope.
The image explains how a continuous input can acquire a discrete label through a threshold. Without the legend, an explanatory structure may be mistaken for an empirical model and a precise crossing becomes fabricated evidence.
Two Endpoints Cannot Locate the Actual Crossing
Linear interpolation between low and high d could calculate an h crossing, but assumes a line and comparable objects and conditions, none established here.
The roughly 0.5 low-high differences for both a and b look orderly but come from one Q1 cycle. The relation may curve, lag, or depend on another condition. Interpolation can be an assumption demonstration, not E0/E1's natural boundary.
The 4.7–5.6 range therefore remains transitional. A clean vertical line on a chart cannot supply an unknown crossing.
The Threshold Is Not the Cause of Change
d above 2.0 receives a failure label because of the rule; the rule did not move d from 1.6 to 2.2. Explaining the reading still requires measurement, material, execution, and conditions.
Treating the threshold as cause invites changing the number instead of investigating process. Treating natural change as threshold makes the world appear to decide at 2.0 by itself.
The diagram colors generating relations and adjudicative relations differently. The former remains incompletely identified; the latter comes from the stored checklist.
A Threshold Is Not Arbitrarily Wrong Either
A human rule can have grounds. The parties need a jointly adjudicable receipt condition, and 2.0 supplies one. Without a boundary, dispute may be deferred until after the outcome.
The book does not show that 2.0 is an optimal real-world standard. It is a case rule inherited by V2, with a preserved scope and version. Modification requires reopening the checklist together; an adverse 2.2 does not permit unilateral relaxation.
A rule can be useful and limited. Its human origin does not remove its force within the adopted version.
Boundary Inclusion Must Be Specified in Advance
“No greater than 2.0 passes” and “less than 2.0 passes” differ at exactly 2.0. Choosing the sign after seeing the boundary object makes the rule manipulable by outcome preference.
The original checklist includes 2.0, so this chapter does not reconsider its side. A later version may change it while preserving old results under the old rule.
Boundary precision must also match recording resolution. d is stored to one decimal; no hidden 2.04 or 1.96 may be invented to change classification.
Measurement Limits Matter More Near a Boundary
Far from 2.0, small reading differences may not change the label. Near it, the same difference can flip the outcome. Classification is more sensitive near the boundary.
This does not mean natural variation becomes larger there. What changes is the rule's response. When measurement uncertainty is unknown, preserve raw readings, repeat relations, and a pause exit.
Wednesday's 2.2 and 2.3 remain on the failing side, so rereading does not change classification. That does not prove exact measurement accuracy.
A Pass Label Hides Distance
Values 1.6, 1.8, and 2.0 all pass but lie at different distances from the boundary. A label alone hides which is more easily crossed by a small change.
Likewise, 2.2 and an extreme value of ten both fail but can imply very different urgency and recovery. Binary labels serve adjudication, not preservation of full state.
The diagram retains both values and labels so classification does not delete continuous information.
An Independent Sensitivity Demonstration
Outside D17, suppose three readings are 1.9, 2.0, and 2.1. Under d≤2.0, the first two pass and the third fails. Change the boundary to 1.8 and all fail; change it to 2.2 and all pass.
The data stay fixed while classification changes with the rule. This shows threshold sensitivity, not which threshold is correct.
Sensitivity cannot select the most favorable threshold over the agreed 2.0. D17 remains governed by its checklist.
Separate Process Nonlinearity from Rule Nonlinearity
A generating process may itself have thresholds, saturation, or phase transitions where a small input change produces a large output change. Such claims need corresponding continuous material or theory.
Current Chengwan evidence contains limited conditions and a crossed rule. It does not establish whether d's generating process is nonlinear. Inferring a material phase change from a sudden action consequence exceeds the data.
The possibility remains open without becoming mainline fact. Rule nonlinearity is known; process shape is not.
A Labor Boundary Also Creates Discrete Consequences
After Thursday's investigation, the ten slots comprise six existing commitments, one G1, one G2 first segment, one investigation, and only one uncommitted recovery slot.
Releasing it to G2 moves uncommitted recovery capacity from one to zero. Quantity falls by one, but action state changes from “recovery entrance retained” to “no uncommitted slot this week.”
The discrete consequence comes from the survival rules of Chapters 5 and 9, not a natural discontinuity in labor at zero. Capacity thresholds require the consequence they protect.
The Eight-Unit Line Is Also an Action Boundary
Cash is thirteen: eight protected, four originally unallocated, one conditional V2 unit. If an action made only 7.9 available when protection is needed rather than 8.1, it would cross the line despite a small difference.
The decimals illustrate and are not new Chengwan accounts; actual records use integer units. The eight-unit protection is a case arrangement.
A hard constraint turns a continuous balance into feasible and infeasible. It can update with grounds when protected obligations change, but not after the fact to approve a favored action.
A Deadline Likewise Discretizes Continuous Minutes
V2's original receipt window ended at 16:00 Wednesday. A qualifying handoff before and after that point receives different labels even if separated by one minute.
The physical product does not disappear at 16:01. What changes is fulfillment of the original window. A later product may enter a new object, not rewrite B_D.
Time thresholds make coordination adjudicable and create boundary consequences. Actors should arrange margin before a deadline rather than treating every minute as equivalent at the end.
Multiple Thresholds Can Cascade
Wednesday's d crossing 2.0 triggered stage pause; pause prevented 14:00 handoff; after 16:00 B_D occurred; the final-payment trigger failed, and V2 entered reopening.
Each step has its own rule. The 0.2 excess did not mathematically become every later loss; it propagated through established action relations.
A cascade diagram reveals which rule may change and which fact cannot. A later renegotiated window does not erase Wednesday B_D; a revised internal check does not alter the original 2.2.
Thresholds Can Invite Borderline Manipulation
When passage has large benefits, participants may repeat until a favorable value appears, select the best sample, change rounding, or revise the boundary after the result.
Chengwan reduces these openings by freezing S-3, preserving every reading, and limiting the reread object, without claiming complete elimination. Q1 retains both favorable and adverse readings rather than reporting only low-position passes.
The more consequential a rule, the more process records matter. A threshold is not a reason to distrust everyone; it is a reason to design reviewable interfaces in advance.
What a Gray Zone Can Solve
Some systems add a review zone around a formal threshold: ordinary handling far from it and extra readings or human review near it. This can prevent one small fluctuation from triggering an expensive action.
A gray zone does not cancel the external 2.0 standard or make an inside-boundary result better. It adds an intermediate internal action state with time and resource cost.
Chengwan adopts no numerical gray zone. The sample is too small and the window closed. It remains a candidate rule design for a later version.
Dual Thresholds Create Hysteresis
An independent example requires d≤1.8 for entry but pauses an entered process only when d>2.0. Different entry and exit lines reduce repeated switching near one boundary.
The cost is that the same d produces different actions under different histories, so path must be recorded. If parties see only one passing standard, dual thresholds may also create misunderstanding.
The demonstration does not enter V2. Use would require authority, relation to checklist acceptance, and consequences in the intermediate region.
A Threshold Cannot Replace Continuous Monitoring
Checking only whether final d exceeds 2.0 loses the path approaching it. Continuous or staged records can show shrinking margin early and reopen action before formal failure.
Denser monitoring has cost and may reveal more brief crossings. Sampling times, review, and action rules must be defined so every fluctuation does not cause switching.
The dense h and sparse d records expose this asymmetry. A new version should state which process points genuinely serve decisions.
How to Read the Threshold Diagram
Layer one shows actual timed h without interpolating a true path. Layer two shows d by object. Layer three is the 2.0 line. Layer four contains entry, pause, window closure, and financial state.
Solid lines indicate actual records and adopted rules; dashed lines are structural illustrations; gray marks the h transition region without paired d. No exact crossing is mainline fact.
Read from facts to rules to consequences. Reasoning backward from window failure can make every earlier number look inevitably predictive.
How Rule Sensitivity Enters Review
Gu Ning preserves two questions: how would classification change under another threshold, and why was 2.0 used in V2? The first concerns sensitivity; the second concerns provenance.
Sensitivity reveals dependence on the boundary without changing history. The provenance page shows 2.0 came from the retrievable joint checklist, not optimization on current results.
More objects may later test alignment between threshold and goal consequences. Selecting a new line from current results and using those same results to prove it works would mix design and validation.
First Audit the Commitment Behind Each Threshold
One number may serve very different functions: a known natural boundary, a contractual acceptance line, an internal alert, or a temporary queue grouping. Different functions support different conclusions after crossing.
Chengwan traces 2.0 to the uniquely identifiable V1 checklist referenced by V2. It directly governs whether the parties classify an item as receivable, while S-3 converts that external condition into an internal release gate. Thus 2.0 is first a versioned acceptance rule, not a proven natural critical point of the material.
The eight units and one margin slot arise from other commitments. They protect existing fulfillment and recovery under adverse paths and do not participate in client acceptance. The Wednesday 16:00 line comes from receipt scheduling. All three can affect one action but require separate provenance.
Gu Ning adds four fields to every line: proposer, protected object, applicable version, and action after crossing. A line unable to answer them may remain an observational prompt but cannot silently become a stage gate. Changing an internal alert does not rewrite an external checklist; renegotiating receipt does not relax d.
Action Must Remain Executable Under Reading Uncertainty
“The true value may be on the other side” authorizes neither normal release nor permanent rejection. Limited resolution and reading variation require an interim action under incomplete information.
S-3's current action is pause. It preserves 2.3 and 2.2 rather than averaging them into 2.25 and pretending to gain precision, and it does not call measurement error zero because both share a side. Pause means only that release is unmet; record, device, procedure, and condition can still be investigated.
If a later reread crosses from 1.9 to 2.1, the team needs a prior rule: whether to reread, by whom, how many times, how to aggregate, and which resources remain unreleased. Choosing minimum, mean, or majority only after seeing results hides preference in the calculation.
No cross-line reread occurred here, so history is not supplemented with a fictional treatment. Gu Ning records it as a gap for the next version. An executable pause lets the team admit uncertainty without making informational absence an automatic pass.
Freeze the Rule Version Before the Reading
If one set of readings can choose after the fact among 2.0, 2.2, or a gray zone, the diagram becomes a menu for favorable answers. Freezing tells the data which line adjudicates before they arrive.
V2 references the old checklist and S-3 operates before Wednesday release, so 2.0 was fixed when 2.2 and 2.3 appeared. Thursday's gray-zone and dual-threshold ideas can enter only candidate versions and cannot alter B_D. Even if both parties later judge 2.0 too strict, history remains “failed under the then-current version,” with a separate effective time for the new rule.
Freezing does not make rules eternal. It places change after a reading and before the next action, with new material, tradeoffs, and authority stated. Learning occurs without rewriting prior adjudication.
Retain replaced rules. Keeping only the latest number could make a later reader believe Wednesday used it and misjudge execution. Threshold history is part of the evidence chain just like reading history.
Recovery After Crossing Requires More Than Returning to the Passing Side
If a later d reads 1.9, that trial returns to the passing side only. G2 recovery still needs a new client window, usable material version, a decision about the final slot, and treatment of conditional funds.
Let internal pass be R, formation of a new window W, and resource permission K. A recovery authorization under the new version requires at least all three, not R alone:
$$ \text{recovery authorization}=R\land W\land K $$
This expression organizes D17 and is not a universal project formula. It prevents one improved signal from erasing a closed window, consumed labor, and new coordination requirements.
A new window with d still failing cannot replace quality entry either. Each threshold protects a different object, and recovery rebuilds each. Chapter 19 develops these exit and reopening rules; here, recrossing one line does not reverse the entire action chain.
Threshold-Near Prediction Needs an Object
Someone seeing h near the transition region may ask for the probability of crossing 2.0. Answering requires an object, time, material, procedure, and distribution of conditions, none obtainable from the illustrative line.
There is no comparable d batch near h=5.0 and no measurement-error distribution. Gu Ning draws no probability band and does not fill the unknown region with fifty percent.
The threshold diagram explains classification and cannot generate probability inputs. The two can connect later only with adequate material for each.
Threshold as a Rule Framework's Reorganization of Certainty
In RC's order theory, consensus formed through repeated observation settles into coordinatable, operational rule frameworks. The 2.0 line realizes such a rule at D17's scale: it organizes continuous readings into jointly adjudicable states, and the stage gate distributes release authority from those states.
A rule's consensual construction makes it neither arbitrary nor false, nor a natural entity prior to observation. Separating reading, classification, and action shows how a secondary construction gains real efficacy and can be renegotiated through versioning when new observation arrives.
The D17 Threshold Diagram
At 15:00 Thursday, Gu Ning saves the diagram. h moves gradually; limited d points lie on both sides of 2.0; the classification function and stage gate are explicit; Wednesday's window and the eight-unit and one-slot boundaries appear separately.
Main conclusion: observed conditions and readings may change continuously, rules map them into discrete states, and the action chain amplifies state differences into pause, missed windows, and changed cash paths. Discrete consequences do not establish an abrupt environmental rupture.
Unresolved are the actual d-h curve, transition readings, measurement distribution, long-term suitability of 2.0, and whether to adopt an internal gray zone. Completing the diagram answers none of them.
Resource and V2 State Remain
Organizing the diagram uses fifty minutes of management time without another half-day production slot or working-capital expense. One recovery slot remains this week and cash remains thirteen.
V2 stays paused, no new receipt window has formed, one conditional unit remains, and the final payment has not arrived. Threshold interpretation cannot reopen Wednesday's window.
The diagram structures the next judgment but is not G2 recovery authority. Any continuation faces this week's capacity and the path already taken.
Handing Off to the Path Ledger at 15:20 Thursday
At 15:20, Gu Ning connects the threshold diagram to the stage record. G1's pass, use of G2's first segment, Wednesday's failed entry, one investigation slot, and window closure retain their times.
These choices have changed the present. Some cash and labor cannot return, the first-payment balance remains conditional, while the old checklist and S-3 records create learning material. Action has not returned to Saturday's zero-input state.
The next chapter places these changes in a path ledger. It will not use prior input to prove continuation, but distinguish locked choices, recoverable resources, learned capabilities, and future branches opened or closed as a result.