关于「孔多塞循环的精确概率:216 种剖面、评审人数与选项个数」,《不存在完美裁决器》给出的定量结论是什么?
结论:第二章精确算出孔多塞循环的概率:三评审为 0.0556,评审增多时趋向 0.0877,选项增多时迅速上升。
数据口径:多道检查意见不一时,经营者最常见的想法是:只要合成规则设计得足够聪明,总能得到一个公正、一致、无可挑剔的裁决。 阿罗 1950 年与 1951 年证明,选项不少于三个时,不存在同时满足无限制域、弱帕累托、无关选项独立与非独裁的排序合成规则。
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: 0.0556, 0.0877
| n | P(no Condorcet winner) | exact fraction |
|---|---|---|
| 3 | 0.0556 | 1/18 |
| 5 | 0.0694 | 5/72 |
| 7 | 0.0750 | 875/11664 |
| 9 | 0.0780 | 65485/839808 |
| 11 | 0.0798 | 134057/1679616 |
| 13 | 0.0811 | 4901897/60466176 |
| 15 | 0.0820 | 89225851/1088391168 |
| 17 | 0.0827 | 1439664941/17414258688 |
| 19 | 0.0832 | 2112831602453/25389989167104 |
| 21 | 0.0837 | 76462477749515/914039610015744 |
| 23 | 0.0840 | 153584540033093/1828079220031488 |
| 25 | 0.0843 | 99880830340054985/1184595334580404224 |
| limit n -> infinity: 0.0877 |
查看完整验算程序
# Chapter 2. p22: exact cycle probability as the panel grows (3 options).
# Impartial culture with n reviewers (n odd). A profile is summarised
# by how many reviewers hold each of the 6 rankings; its probability is
# the multinomial n! / (k1! k2! k3! k4! k5! k6!) / 6 ** n. We enumerate every
# composition of n into 6 parts and add the weight of those with no
# Condorcet winner. Integer weights, exact Fraction result.
# Limit as n grows (Guilbaud 1952, as reported by Gehrlein 2002):
# 1/4 - (3 / (2 pi)) asin(1/3)
import itertools
import math
from fractions import Fraction
OPTS = "ABC"
ORDERS = ["".join(p) for p in itertools.permutations(OPTS)]
def compositions(n, k):
for cut in itertools.combinations(range(n + k - 1), k - 1):
prev = -1
parts = []
for c in cut:
parts.append(c - prev - 1)
prev = c
parts.append(n + k - 1 - prev - 1)
yield parts
PREF = {(x, y): [o.index(x) < o.index(y) for o in ORDERS]
for x in OPTS for y in OPTS if x != y}
def no_winner(counts, n):
for x in OPTS:
ok = True
for y in OPTS:
if y == x:
continue
k = 0
for c, f in zip(counts, PREF[(x, y)]):
if f:
k += c
if 2 * k <= n:
ok = False
break
if ok:
return False
return True
def p_cycle(n):
bad = 0
fn = math.factorial(n)
for parts in compositions(n, 6):
if no_winner(parts, n):
w = fn
for c in parts:
w //= math.factorial(c)
bad += w
return Fraction(bad, 6 ** n)
if __name__ == "__main__":
print(" n P(no Condorcet winner) exact fraction")
for n in range(3, 26, 2):
p = p_cycle(n)
fr = str(p)
if len(fr) > 40:
fr = "(long fraction)"
print(" {:>2} {:.4f} {}".format(n, float(p), fr))
lim = 0.25 - 1.5 / math.pi * math.asin(1 / 3)
print("limit n -> infinity: {:.4f}".format(lim))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《阿罗不可能定理与多评审规则聚合、五件套拒绝码冲突时为何不存在完美裁决器母典》第 2 章「孔多塞循环的精确概率:216 种剖面、评审人数与选项个数」(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-310 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026310,
author = {Fan, Yafu},
title = {Arrow's Impossibility Theorem and the Aggregation of Multiple Reviewers: Why No Perfect Arbiter Exists When Kit Refusal Codes Conflict},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}