What does Hard Gates and Soft-Prompt Decay conclude, in numbers, about “Closing the Loop: the Weekly Interception Report, the Residual Calculator and a Pre-Registered Test”?
Answer: Chapter 8 gives a weekly interception report (in an invented example the block rate falls from 42.39 to 12.71), the Gate Residual Violation Rate Calculator and a pre-registered test (power 0.916). Marked boxes separate published evidence, runnable Python calculations (21 programs), classical sources and the author's hypotheses.
Scope of the numbers: Soft reminders work in proportion; when layers are independent the residual violation rate is the product of their miss rates, and with illustrative parameters four layers still leave 0.0048, not zero. Its parameters are illustrative; all calculations are generated by its programs.
Data table
The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 12.71, 21, 42.39
| wk saves BAN ELL SPC SUM WID per1000 reblock leak/1000 | |
| 1 | 1840 21 9 30 6 12 42.39 0.141 1.087 |
| 2 | 2210 17 7 22 3 10 26.70 0.119 0.452 |
| 3 | 1975 12 4 15 2 6 19.75 0.103 0.000 |
| 4 | 2360 10 3 11 1 5 12.71 0.100 0.000 |
| week 4: zero leaks in 2360 saves, 95% bound 0.00127 per save | |
| turn 0-9 saves | 3120 blocks 45 per1000 14.42 |
| turn 10-19 saves | 2650 blocks 64 per1000 24.15 |
| turn 20-29 saves | 1610 blocks 55 per1000 34.16 |
| turn 30+ saves | 1005 blocks 42 per1000 41.79 |
| late over early block rate | 2.90, log ratio 1.064 |
Show the full program
# Chapter 8. p81: the weekly report of a gate's interception log.
# Invented log of four weeks (not real data). Per week: saves; first
# blocks by rule (BAN banned words, ELL ellipsis, SPC digit-hanzi
# spacing, SUM float sums, WID line width); re-blocks (the same file
# blocked again right after a fix: the fix order was not obeyed);
# leaks (violations the publish gate found that the save hook had
# passed: hook down or an unhooked channel). Last lines: blocks per
# save by the turn at which the save happened, the raw sign of decay.
import math
WEEKS = [(1840, {"BAN": 21, "ELL": 9, "SPC": 30, "SUM": 6, "WID": 12},
11, 2),
(2210, {"BAN": 17, "ELL": 7, "SPC": 22, "SUM": 3, "WID": 10},
7, 1),
(1975, {"BAN": 12, "ELL": 4, "SPC": 15, "SUM": 2, "WID": 6},
4, 0),
(2360, {"BAN": 10, "ELL": 3, "SPC": 11, "SUM": 1, "WID": 5},
3, 0)]
RULES = ["BAN", "ELL", "SPC", "SUM", "WID"]
print("wk saves BAN ELL SPC SUM WID per1000 reblock leak/1000")
for i, (n, b, rb, lk) in enumerate(WEEKS, 1):
tot = 0
for r in RULES:
tot += b[r]
row = "%2d %5d " % (i, n)
for r in RULES:
row += "%4d" % b[r]
print(row + " %7.2f %7.3f %9.3f" % (1000 * tot / n, rb / tot,
1000 * lk / n))
n4 = WEEKS[3][0]
print("week 4: zero leaks in %d saves, 95%% bound %.5f per save" % (
n4, 1 - 0.05 ** (1 / n4)))
BUCKETS = [("turn 0-9", 3120, 45), ("turn 10-19", 2650, 64),
("turn 20-29", 1610, 55), ("turn 30+", 1005, 42)]
for name, n, b in BUCKETS:
print("%-11s saves %5d blocks %3d per1000 %6.2f" % (
name, n, b, 1000 * b / n))
lo = BUCKETS[0][2] / BUCKETS[0][1]
hi = BUCKETS[3][2] / BUCKETS[3][1]
print("late over early block rate: %.2f, log ratio %.3f" % (
hi / lo, math.log(hi / lo)))Single source and citation
Fan, Y. (2026). Hard Gates across the Client Lifecycle: the Decay of Soft-Prompt Compliance over Turns and State Machines for Interception before Publication (Version 1.0.0), Chapter 8: Closing the Loop: the Weekly Interception Report, the Residual Calculator and a Pre-Registered Test. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-256 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026256,
author = {Fan, Yafu},
title = {Hard Gates across the Client Lifecycle: the Decay of Soft-Prompt Compliance over Turns and State Machines for Interception before Publication},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · The Product Account: Why Stacked Soft Layers Never Leave a Zero Residual
- Chapter 2 · Holes That Line Up: the Floor That Common-Cause Failure Puts under Stacked Soft Layers
- Chapter 3 · Decay over Turns: an Exponential Model of Soft-Prompt Compliance, Re-Sending Periods and Reading the Log
- Chapter 4 · Where to Intercept: before the Write, after It or before Publication, and Whether a Broken Gate Passes or Refuses
- Chapter 5 · Exhausting the State Machine: Proving a Dirty Publication Unreachable and Finding the Shortest Counterexamples of Three Faulty Designs
- Chapter 6 · The Price of Hardness: Pattern Width, Wrong Blocks and Gates That Get Switched Off
- Chapter 7 · Classical Sources East and West: Law and the Inked Line in the Han Feizi, and Reason's Classification of Error