关于「字典序优先级裁决器:它放弃了哪条公理,换来了什么」,《不存在完美裁决器》给出的定量结论是什么?
结论:第五章审计字典序优先级与分档规则,并用操纵普查说明吉巴德与萨特思韦特定理:四选项时博尔达在 0.5122 的剖面可被操纵,优先级规则为零。
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: 0.5122
| 3 options, 216 profiles | |
| plurality manipulable | 36 share 0.1667 |
| Borda manipulable | 51 share 0.2361 |
| Black manipulable | 24 share 0.1111 |
| priority manipulable | 0 share 0.0000 |
| 4 options, 13,824 profiles | |
| plurality manipulable | 4,068 share 0.2943 |
| Borda manipulable | 7,080 share 0.5122 |
| Black manipulable | 3,816 share 0.2760 |
| priority manipulable | 0 share 0.0000 |
查看完整验算程序
# Chapter 5. p52: who can game the panel? (Gibbard 1973; Satterthwaite
# 1975). Every rule here picks one winner; ties go to the alphabetically
# first option. A profile is manipulable if some reviewer, by reporting
# a false ranking, gets a winner she truly prefers. Exact census over
# all profiles of 3 reviewers with 3 options (216) and 4 options
# (13,824). The theorem says only a dictatorship, or a rule with at
# most two possible winners, is immune once there are 3 or more
# options; the priority rule below is such a dictatorship.
import itertools
from fractions import Fraction
def scores(prof, w):
s = {}
for o in prof:
for pos, c in enumerate(o):
s[c] = s.get(c, 0) + w[pos]
return s
def pick(s):
return min(s, key=lambda c: (-s[c], c))
def plurality(prof):
m = len(prof[0])
return pick(scores(prof, [1] + [0] * (m - 1)))
def borda(prof):
m = len(prof[0])
return pick(scores(prof, list(range(m - 1, -1, -1))))
def black(prof):
opts = prof[0]
for x in opts:
if all(2 * len([o for o in prof if o.index(x) < o.index(y)])
> len(prof) for y in opts if y != x):
return x
return borda(prof)
def priority(prof):
return prof[0][0]
def manipulable(rule, prof, orders):
w = rule(prof)
for i, true in enumerate(prof):
for lie in orders:
if lie == true:
continue
alt = list(prof)
alt[i] = lie
if true.index(rule(alt)) < true.index(w):
return True
return False
if __name__ == "__main__":
for opts in ("ABC", "ABCD"):
orders = ["".join(p) for p in itertools.permutations(opts)]
profs = list(itertools.product(orders, repeat=3))
print("{} options, {:,} profiles".format(len(opts), len(profs)))
for name, rule in (("plurality", plurality), ("Borda", borda),
("Black", black), ("priority", priority)):
k = 0
for pr in profs:
if manipulable(rule, pr, orders):
k += 1
print(" {:<9} manipulable {:>6,} share {:.4f}".format(
name, k, float(Fraction(k, len(profs)))))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 5 章「字典序优先级裁决器:它放弃了哪条公理,换来了什么」(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}
}