关于「破绽发现曲线:未见概率、物种数估计与停测规则」,《红蓝军合成客群沙盘》给出的定量结论是什么?

结论:第五章画破绽发现曲线,用古德估计与赵氏下界给出停测规则,24 位买家零发现只能把破绽率上限压到 0.1173。

数据口径:本书检验相反的立项命题:只用平均买家会漏掉尾部破绽,拉丁超立方合成客群在同样测试量下覆盖更多人格组合(McKay、Beckman 与 Conover 1979 年刊于《技术计量学》),多轮砍价有 Rubinstein 1982 年给出的贴现均衡。 书中参数为示意;全部计算由书中程序生成,不调用任何模型接口;本书是非个人化经营建议。

English version

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: 24

modes 40; largest rate 0.150; smallest 0.00259
nmodes seensharemissing mass
62.880.0720.40049
248.420.2110.24178
4813.290.3320.17034
9619.780.4940.10753
19227.300.6820.05680
38434.230.8560.02209
76838.530.9630.00494
153639.870.9970.00039
buyers to expect 50 per cent of modes: 99
buyers to expect 80 per cent of modes: 302
buyers to expect 90 per cent of modes: 483
modes one random buyer trips on average: 0.552
查看完整验算程序
# Chapter 5. p51: the failure-mode discovery curve.
# 40 leak modes (illustrative) with Zipf-like trigger rates per
# synthetic buyer, p_i = 0.15 / i^1.1. After n independent buyers the
# expected number of modes seen is E(n) = sum_i [1 - (1 - p_i)^n] and
# the chance that the next buyer trips a mode not yet seen (missing
# mass) is M(n) = sum_i p_i (1 - p_i)^n. Sums use math.fsum.
import math

K = 40
P = [0.15 / i ** 1.1 for i in range(1, K + 1)]


def seen(n):
    return math.fsum(1 - (1 - p) ** n for p in P)


def missing(n):
    return math.fsum(p * (1 - p) ** n for p in P)


print("modes %d; largest rate %.3f; smallest %.5f" % (K, P[0], P[-1]))
print("    n   modes seen   share   missing mass")
for n in (6, 24, 48, 96, 192, 384, 768, 1536):
    e = seen(n)
    print("%5d   %9.2f    %.3f   %.5f" % (n, e, e / K, missing(n)))
for goal in (0.5, 0.8, 0.9):
    n = 1
    while seen(n) < goal * K:
        n += 1
    print("buyers to expect %.0f per cent of modes: %d"
          % (goal * 100, n))
print("modes one random buyer trips on average: %.3f" % seen(1))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《红蓝军合成客群对抗沙盘、参数化挑剔买家多轮砍价博弈与漏单破绽预检母典》第 5 章「破绽发现曲线:未见概率、物种数估计与停测规则」(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-259 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026259,
  author    = {Fan, Yafu},
  title     = {Red-Blue Synthetic-Customer Sandboxes: Multi-Round Bargaining against Parameterised Demanding Buyers and Pre-Launch Leak Detection},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论