关于「四条公理与一张空表:在三评审三选项上穷举阿罗定理」,《不存在完美裁决器》给出的定量结论是什么?

结论:第一章在三评审三选项上穷举所有满足帕累托与无关选项独立的 262,144 种规则,能保证传递的只有独裁者。

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。

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: 262144

n = 2 reviewers, verdicts strict      :      64 rules
   transitive on every profile: 2
   survivor: reviewer 1 dictates
   survivor: reviewer 2 dictates
n = 2 reviewers, verdicts ties allowed:     729 rules
   transitive on every profile: 2
   survivor: reviewer 1 dictates
   survivor: reviewer 2 dictates
n = 3 reviewers, verdicts strict      : 262,144 rules
   transitive on every profile: 3
   survivor: reviewer 1 dictates
   survivor: reviewer 2 dictates
   survivor: reviewer 3 dictates
only dictators survive Pareto + IIA + ORD (Arrow, small n)
查看完整验算程序
# Chapter 1. p12: exhaustive search of all IIA + Pareto rules (small n).
# By IIA a rule is a separate verdict table for each pair {x, y}: it
# maps the reviewers' x-vs-y votes to a social verdict. Pareto fixes
# the two unanimous rows; every other row is free. We enumerate every
# such rule and keep those whose verdicts are transitive on all
# profiles of strict rankings over 3 options (Arrow's ORD axiom).
# Exact enumeration; nothing is sampled.
import itertools

OPTS = "ABC"
ORDERS = ["".join(p) for p in itertools.permutations(OPTS)]
PAIRS = [("A", "B"), ("B", "C"), ("A", "C")]


def pattern(prof, x, y):
    return tuple(o.index(x) < o.index(y) for o in prof)


def survivors(n, verdicts):
    pats = list(itertools.product((True, False), repeat=n))
    free = [p for p in pats if any(p) and not all(p)]
    profiles = list(itertools.product(ORDERS, repeat=n))
    pp = [[pattern(pr, x, y) for x, y in PAIRS] for pr in profiles]

    def tables():
        for vals in itertools.product(verdicts, repeat=len(free)):
            t = dict(zip(free, vals))
            t[tuple([True] * n)] = 1
            t[tuple([False] * n)] = -1
            yield t
    tabs = list(tables())
    keep = []
    for tab_ab in tabs:
        for tab_bc in tabs:
            need = {}
            ok = True
            for ab, bc, ac in pp:
                a, b = tab_ab[ab], tab_bc[bc]
                # A?B = a, B?C = b; transitivity pins A?C in some cases
                if a >= 0 and b >= 0:
                    req = {1} if a > 0 or b > 0 else {0}
                elif a <= 0 and b <= 0:
                    req = {-1} if a < 0 or b < 0 else {0}
                else:
                    req = {1, 0, -1}
                cur = need.get(ac, {1, 0, -1}) & req
                if not cur:
                    ok = False
                    break
                need[ac] = cur
            if not ok:
                continue
            for tab_ac in tabs:
                if all(tab_ac[k] in v for k, v in need.items()):
                    if check(pp, tab_ab, tab_bc, tab_ac):
                        keep.append((tab_ab, tab_bc, tab_ac))
    return len(tabs) ** 3, keep


def check(pp, t1, t2, t3):
    for ab, bc, ac in pp:
        r = {("A", "B"): t1[ab], ("B", "C"): t2[bc], ("A", "C"): t3[ac]}
        g = dict(r)
        for (x, y), v in r.items():
            g[(y, x)] = -v
        for x, y, z in itertools.permutations(OPTS, 3):
            a, b, c = g[(x, y)], g[(y, z)], g[(x, z)]
            if a >= 0 and b >= 0 and c < 0:
                return False
            if a > 0 and b >= 0 and c <= 0:
                return False
    return True


def who_dictates(rule, n):
    ds = []
    for i in range(n):
        if all(t[p] == (1 if p[i] else -1) for t in rule for p in t):
            ds.append(i + 1)
    return ds


if __name__ == "__main__":
    for n, verdicts, label in ((2, (1, -1), "strict"),
                               (2, (1, 0, -1), "ties allowed"),
                               (3, (1, -1), "strict")):
        total, keep = survivors(n, verdicts)
        ds = [who_dictates(r, n) for r in keep]
        print("n = {} reviewers, verdicts {:<12}: {:>7,} rules".format(
            n, label, total))
        print("   transitive on every profile: {}".format(len(keep)))
        for d in ds:
            print("   survivor: reviewer {} dictates".format(d[0])
                  if d else "   survivor: no dictator")
    print("only dictators survive Pareto + IIA + ORD (Arrow, small n)")

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 1 章「四条公理与一张空表:在三评审三选项上穷举阿罗定理」(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}
}

本书其他章节结论