What does Context-Parametric Rule Distillation and Online Calibration conclude, in numbers, about “Classical Sources: Wheelwright Bian and Aristotle on Equity”?
Answer: Chapter 7 reads Wheelwright Bian in the Zhuangzi, Mencius on law that cannot carry itself out, and Aristotle on equity; the unspoken residual share in the illustrative model is 0.291.
Scope of the numbers: Parameters are illustrative and all calculations are generated by the book's programs. This is general business analysis, not investment, legal or medical advice, and it describes no real customer or sale.
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: 0.291
| total variance of the right speed: 0.86 | ||
| factors written down | loss | share of loss left |
| none (one speed) | 0.86 | 1.000 |
| hardness | 0.50 | 0.581 |
| hardness+moisture | 0.34 | 0.395 |
| hardness+moisture+grain | 0.25 | 0.291 |
| tacit share that no written rule removes: 0.291 | ||
| apprentice learning the full written rule from n wheels: | ||
| n = | 5: extra loss 1.2447 | n x extra 6.22 |
| n = | 10: extra loss 0.3421 | n x extra 3.42 |
| n = | 40: extra loss 0.0601 | n x extra 2.41 |
| n = 160: extra loss 0.0144 | n x extra 2.31 |
Show the full program
# Chapter 7. p71: Wheelwright Bian as a variance budget (illustrative).
# The right chisel speed v* for a wheel = hardness + moisture + grain +
# a tacit part felt in the hand. Components are independent with
# variances 0.36, 0.16, 0.09 and 0.25 (arbitrary units). Loss of a
# wheel = (v - v*)^2. A written rule can condition on the recorded
# factors only; the tacit part stays as residual loss however good the
# rule is. An apprentice who estimates the written rule from n wheels
# (one slope per factor and an intercept, by least squares) pays an
# extra loss on top; Monte Carlo shows it shrinks roughly as 1 / n.
import math
import random
VAR = (("hardness", 0.36), ("moisture", 0.16), ("grain", 0.09))
TACIT = 0.25
def main():
total = math.fsum([v for _, v in VAR] + [TACIT])
print("total variance of the right speed: {:.2f}".format(total))
print(" factors written down loss share of loss left")
for k in range(len(VAR) + 1):
left = math.fsum([v for _, v in VAR[k:]] + [TACIT])
names = "+".join(n for n, _ in VAR[:k]) or "none (one speed)"
print(" {:<28} {:.2f} {:.3f}".format(names, left, left / total))
print("tacit share that no written rule removes: {:.3f}".format(
TACIT / total))
print("apprentice learning the full written rule from n wheels:")
rng = random.Random(7101)
for n in (5, 10, 40, 160):
extra = []
for _ in range(2000):
xs = [[rng.gauss(0, math.sqrt(v)) for _, v in VAR]
for _ in range(n)]
ys = [math.fsum(x) + rng.gauss(0, math.sqrt(TACIT))
for x in xs]
# independent factors: per-factor slope, centred
mx = [math.fsum(x[j] for x in xs) / n for j in range(3)]
my = math.fsum(ys) / n
coef = []
for j in range(3):
sxy = math.fsum((x[j] - mx[j]) * (y - my)
for x, y in zip(xs, ys))
sxx = math.fsum((x[j] - mx[j]) ** 2 for x in xs)
coef.append(sxy / sxx)
err = [(c - 1.0) ** 2 * v for c, (_, v) in zip(coef, VAR)]
icpt = my - math.fsum(c * m for c, m in zip(coef, mx))
extra.append(math.fsum(err) + icpt ** 2)
mean = math.fsum(extra) / 2000
print(" n = {:>3}: extra loss {:.4f} n x extra {:.2f}".format(
n, mean, n * mean))
if __name__ == "__main__":
main()Single source and citation
Fan, Y. (2026). Next-Generation Knowledge Distillation, Context-Parametric Compilation and Live-Conversion Feedback Calibration (Version 1.0.0), Chapter 7: Classical Sources: Wheelwright Bian and Aristotle on Equity. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-377 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026377,
author = {Fan, Yafu},
title = {Next-Generation Knowledge Distillation, Context-Parametric Compilation and Live-Conversion Feedback Calibration},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · Simpson's Reversal: Why a Rule Without Its Context Can Point the Wrong Way
- Chapter 2 · Context-Parametric Compilation: Squeezing a Decision Table into Decision Lists and Ordered Decision Diagrams
- Chapter 3 · Resolving Conflicts Across Volumes: Do More Rules Make a Pack More Accurate?
- Chapter 4 · Online Bayesian Calibration: Giving Every Rule a Beta Distribution That Grows
- Chapter 5 · Threshold Drift: Discounted Beta Updates, Sliding Windows and a Cumulative-Sum Alarm
- Chapter 6 · Calibration and the Whole Pipeline: the Brier Partition, Binned Calibration Error and a Net-Benefit Simulation
- Chapter 8 · Closing the Loop: Field Tools and Checker Specifications, the Risk Ledger and a Pre-registered Test