关于「原典考辨:《孙子·谋攻》的「知彼知己」与莱斯维茨 1824 年的兵棋」,《红蓝军合成客群沙盘》给出的定量结论是什么?
结论:第七章考辨《孙子·谋攻》的「知彼知己」与莱斯维茨 1824 年的兵棋,知情裁判会让毫无优势的话术在 0.5215 的对比中胜出。
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。
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: 0.5215
| umpire | fav cut | P(fav ahead by 3 pts) | median gap |
| blind | 4 | 0.2223 | +0.0000 |
| leaning | 3 | 0.3633 | +0.0167 |
| strong lean | 2 | 0.5215 | +0.0333 |
| expected lean: 0.0167 per buyer (ambiguous share x 1/6) |
查看完整验算程序
# Chapter 7. p72: dice are honest; an umpire who knows the favourite
# is not. In the 1824 Kriegsspiel an umpire resolved combat with dice
# and simple arithmetic on a map at scale 1:8000. Sandbox analogue
# (illustrative): two scripts with the same true quality face 240
# synthetic buyers each; 20 per
# cent of buyers clearly buy, 10 per cent are ambiguous transcripts
# that the umpire resolves with a six-sided die. A blind umpire counts
# an ambiguous case as a sale on 4 to 6; an umpire who knows which
# script is the house favourite counts it on 3 to 6 for that script.
# We count how often the favourite is declared better by at least 3
# points, over 4000 seeded duels.
import random
N, CLEAR, AMBIG, DUELS = 240, 0.20, 0.10, 4000
def sales(rng, cut):
k = 0
for _ in range(N):
u = rng.random()
if u < CLEAR:
k += 1
elif u < CLEAR + AMBIG and rng.randint(1, 6) >= cut:
k += 1
return k
rng = random.Random(20261113)
print("umpire fav cut P(fav ahead by 3 pts) median gap")
for label, cut in (("blind", 4), ("leaning", 3), ("strong lean", 2)):
wins, gaps = 0, []
for _ in range(DUELS):
a = sales(rng, cut) / N
b = sales(rng, 4) / N
gaps.append(a - b)
wins += a - b >= 0.03
gaps.sort()
print("%-12s %7d %21.4f %+10.4f"
% (label, cut, wins / DUELS, gaps[DUELS // 2]))
print("expected lean: %.4f per buyer (ambiguous share x 1/6)"
% (AMBIG / 6))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 7 章「原典考辨:《孙子·谋攻》的「知彼知己」与莱斯维茨 1824 年的兵棋」(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}
}