关于「实践与证伪方案:文案检查器、预注册检验与推翻条件」,《梅花易数的哈希与基率》给出的定量结论是什么?
结论:第八章给出经营者的自查清单、八条规则的攻心文案检查器与预注册检验方案(盲法匹配需要 325 人)。书中以标注框区分已发表的证据、可运行的 Python 计算(18 段程序)、原典与作者的假说。
数据口径:书中的人群分布、注意率与记忆率均为示意;全部计算由书中程序生成;本书分析结构与概率,不做任何个人占断,也不构成商业、法律或理财建议。 第六章用似然比为「事后对上」定价,并模拟选择性记忆如何把一半对一半的运气变成 0.795 的回忆相符率。
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: 325
test p0 p1 n reject if >= T1 blind three-way match 0.3333 0.4000 325 123 size 0.0488, power 0.8018 T2 strict omen, 1/8 0.1250 0.2000 149 26 size 0.0490, power 0.8095 T1 normal approximation: 317.4 T3 collision audit: any n; one mismatching pair refutes p11
查看完整验算程序
# Chapter 8. p81: sample sizes for three pre-registered tests, exact.
# T1 blind matching: each client sees the reading cast for their own
# question and two decoys cast for other times, and picks one; chance
# is 1/3, the claim to detect 0.40. T2 strict omen: a trigram and one
# dimension named in advance, chance 1/8, claim 0.20. T3 collision
# audit: n pairs who cast in the same double-hour; the rule predicts
# all identical, so one mismatch refutes it. One-sided alpha 0.05,
# power 0.80; exact binomial tails; smallest n that works.
import math
def upper_tail(n, p, c):
lp, lq = math.log(p), math.log(1 - p)
terms = [math.exp(math.lgamma(n + 1) - math.lgamma(k + 1)
- math.lgamma(n - k + 1) + k * lp + (n - k) * lq)
for k in range(c, n + 1)]
return math.fsum(terms)
def smallest_n(p0, p1, alpha=0.05, power=0.80):
n = 5
while True:
c = math.ceil(n * p0)
while upper_tail(n, p0, c) > alpha:
c += 1
if upper_tail(n, p1, c) >= power:
return n, c
n += 1
def main():
print("test p0 p1 n reject if >=")
for name, p0, p1 in (("T1 blind three-way match", 1 / 3, 0.40),
("T2 strict omen, 1/8", 0.125, 0.20)):
n, c = smallest_n(p0, p1)
print("{:<28} {:.4f} {:.4f} {:>5} {}".format(
name, p0, p1, n, c))
print(" size {:.4f}, power {:.4f}".format(
upper_tail(n, p0, c), upper_tail(n, p1, c)))
z_a, z_b = 1.6449, 0.8416
p0, p1 = 1 / 3, 0.40
approx = ((z_a * math.sqrt(p0 * (1 - p0))
+ z_b * math.sqrt(p1 * (1 - p1))) / (p1 - p0)) ** 2
print("T1 normal approximation: {:.1f}".format(approx))
print("T3 collision audit: any n; one mismatching pair refutes p11")
if __name__ == "__main__":
main()唯一数据源与引用
范雅福(Fan, Y.). (2026). 《梅花易数体用生克自动机、年月日时哈希碰撞与外应注意捕获基率》第 8 章「实践与证伪方案:文案检查器、预注册检验与推翻条件」(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-156 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026156,
author = {Fan, Yafu},
title = {Plum-Blossom Numerology as a Body-Use Automaton, Hash Collisions in Date-Time Casting and Base Rates of Attention-Captured Omens},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}