What does The Nightly Funnel conclude, in numbers, about “Classical Sources East and West: "At the Limit, Change" in the Great Commentary, the Sunzi on Weak and Strong Points, and Random Allocation in Peirce and Jastrow”?
Answer: Chapter 7 examines the Great Commentary on the Book of Changes on change at the limit, the Sunzi on weak and strong points, and the random allocation of Peirce and Jastrow in 1885, and gives the distillation spec for the four kits.
Scope of the numbers: Its parameters are illustrative; all calculations are generated by its 23 programs; it is general business analysis, not business, legal or financial advice, and it promises no rise in conversion.
Data table
The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python only
| policy | changes tried | kept | harmful kept | share harmful | uplift |
|---|---|---|---|---|---|
| one | 26 | 3.0 | 0.01 | 0.002 | +0.093 |
| uplift P5 +0.000 | P50 +0.082 | P95 +0.217 | |||
| two | 52 | 9.2 | 0.11 | 0.012 | +0.238 |
| uplift P5 +0.040 | P50 +0.217 | P95 +0.480 | |||
| 2x2 | 52 | 6.0 | 0.01 | 0.002 | +0.195 |
| uplift P5 +0.040 | P50 +0.170 | P95 +0.369 |
Show the full program
# Chapter 7. p72: one change per cycle versus two changes bundled.
# Each candidate change has a true relative effect on net contribution
# per visitor: 0 (prob 0.60), +4 percent (0.25) or -4 percent (0.15).
# A cycle is one week of testing; the estimate of a variant's effect
# has standard error 3.2 percent (about 42,000 visitors per arm, see
# chapter 8). Keep the variant if estimate / SE > 1.645. Policy "one"
# tests one change per cycle; policy "two" ships two changes as one
# variant, kept or dropped together. Policy "2x2" runs the two changes
# as a factorial design: each main effect is read on all traffic with
# the same SE (no interaction assumed) and kept on its own merit.
# 26 cycles, 4,000 runs.
import math
import random
SE, ZBAR, CYCLES, REPS = 0.032, 1.645, 26, 4000
rng = random.Random(40572)
def effect():
u = rng.random()
return 0.0 if u < 0.60 else (0.04 if u < 0.85 else -0.04)
def run(policy):
kept, harm, total = 0, 0, []
for _ in range(CYCLES):
if policy == "one":
groups = [[effect()]]
elif policy == "two":
groups = [[effect(), effect()]]
else:
groups = [[effect()], [effect()]]
for fx in groups:
est = math.fsum(fx) + SE * rng.gauss(0, 1)
if est / SE > ZBAR:
for f in fx:
kept += 1
harm += 1 if f < 0 else 0
total.append(math.log1p(f))
return kept, harm, math.expm1(math.fsum(total))
print("policy changes tried kept harmful kept share harmful uplift")
for policy in ("one", "two", "2x2"):
per = 1 if policy == "one" else 2
k, h, u = 0, 0, []
for _ in range(REPS):
a, b, c = run(policy)
k += a
h += b
u.append(c)
u.sort()
print("{:<7} {:>13} {:>5.1f} {:>13.2f} {:>14.3f} {:>+7.3f}".format(
policy, CYCLES * per, k / REPS, h / REPS,
h / k, math.fsum(u) / REPS))
print(" uplift P5 {:+.3f} P50 {:+.3f} P95 {:+.3f}".format(
u[REPS // 20], u[REPS // 2], u[19 * REPS // 20]))Single source and citation
Fan, Y. (2026). Autonomous Nightly Funnel Micro-Surgery, Bandit Traffic Allocation and Zero-Touch Conversion Self-Improvement (Version 1.0.0), Chapter 7: Classical Sources East and West: "At the Limit, Change" in the Great Commentary, the Sunzi on Weak and Strong Points, and Random Allocation in Peirce and Jastrow. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-405 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026405,
author = {Fan, Yafu},
title = {Autonomous Nightly Funnel Micro-Surgery, Bandit Traffic Allocation and Zero-Touch Conversion Self-Improvement},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · The Funnel Check-up: Six-Step Beta Posteriors, Nightly Change Priorities and a Compliant Candidate Pool
- Chapter 2 · Thompson Sampling and Regret: from Two Small Samples in 1933 to the Even Split
- Chapter 3 · The Reward Function: Net Contribution per Visitor, Not the Conversion Rate
- Chapter 4 · Non-stationary Traffic: Weekday Effects, Discounted Memory and Sliding Windows
- Chapter 5 · Elimination Rules: Minimum Exposure, Posterior Win Probability and Time-uniform Confidence Sequences
- Chapter 6 · Trustworthy Data: Negative Bias in Adaptive Data, Sample Ratio Mismatch and Guardrail Rollback
- Chapter 8 · The Nightly Funnel Ledger: Field Tools and Checker Specifications, a Pre-registered Test and the Conditions of Refutation