关于「信任棘轮两个方向都会转」,《计算著作学与自校验专著》给出的定量结论是什么?
结论:第四章说明发现一处错误抵消 17.8 次核对无误。
数据口径:书中的读者、书、错误率与成本均为示意模型;全部计算由书中程序生成;本书不是法律、财务或出版建议。 书中以标注框区分已发表的证据、可运行的 Python 计算(16 段程序)、原典与作者的假说。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 17.8
| good checks | P(careful) | after one error |
|---|---|---|
| 0 | 0.200 | 0.016 |
| 5 | 0.349 | 0.034 |
| 10 | 0.535 | 0.071 |
| 20 | 0.841 | 0.260 |
| 40 | 0.991 | 0.881 |
| 80 | 1.000 | 1.000 |
| one found error cancels 17.8 good checks | ||
| 20 good checks, then 2 errors: 0.023 | ||
| 40 good checks, then 2 errors: 0.331 |
查看完整验算程序
# Chapter 4. p07: The trust ratchet runs both ways. A reader checks claims
# one at a time. A careful author is wrong on 1 claim in 100, a careless
# one on 15 in 100. Prior P(careful) = 0.20. Each claim that checks out
# multiplies the odds by 0.99 / 0.85; each error found multiplies them by
# 0.01 / 0.15. How many good checks does one found error cancel?
from math import log
E_C, E_S, PRIOR = 0.01, 0.15, 0.20
def p_after(good, bad, prior=PRIOR):
odds = prior / (1 - prior)
odds *= ((1 - E_C) / (1 - E_S)) ** good * (E_C / E_S) ** bad
return odds / (1 + odds)
if __name__ == "__main__":
print(" good checks P(careful) after one error")
for g in (0, 5, 10, 20, 40, 80):
print(" {:11d} {:10.3f} {:15.3f}".format(
g, p_after(g, 0), p_after(g, 1)))
lr_good = log((1 - E_C) / (1 - E_S))
lr_bad = log(E_S / E_C)
print(" one found error cancels {:.1f} good checks".format(
lr_bad / lr_good))
for g in (20, 40):
print(" {} good checks, then 2 errors: {:.3f}".format(
g, p_after(g, 2)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《计算著作学、先证伪后立界修辞与自校验专著编译器》第 4 章「信任棘轮两个方向都会转」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-101 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026101,
author = {Fan, Yafu},
title = {Computational Bibliometrics, Falsification-First Rhetoric and Self-Verifying Monograph Compilers},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}