关于「谢林聚点:多重均衡里怎样想到一处,又何时失灵」,《施为言语与聚点定框》给出的定量结论是什么?

结论:第二章讨论谢林聚点,复述克劳福德等人公布的协调率从 0.82 降到 0.52,示意模型里一点点收益不对称让协调率从 0.889 跌到 0.497。

数据口径:书中对话全部合成,参数为示意,全部计算由书中程序生成;本书是非个人化经营建议,不是法律意见;分润只写单层、只给直接推荐人。 第一章用奥斯汀与塞尔的施为句理论,把报价拆成六个适切条件:示意模型里结构化报价把适切概率从 0.305 提到 0.734,一份完全可信的报价让对方期望熵下降 0.918 比特。

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: 0.497, 0.52, 0.82, 0.889

eps P1 picks X P2 picks X coordination
0.000.941 0.941 0.889
0.010.459 0.541 0.497
0.050.460 0.540 0.497
0.100.461 0.539 0.497
0.500.474 0.526 0.499
1.000.510 0.490 0.500
split the surplus first (eps0): coordination 0.889
published coordination rates (2008):
sym 100/100 (hypothetical)0.82
asym 100/101 (hypothetical)0.52
X-Y sym 5/5 (paid)0.64
X-Y asym 5/5.1 (paid)0.38
查看完整验算程序
# Chapter 2. p22: a minute payoff asymmetry breaks a focal point.
# Two labels X (salient) and Y. Matching on X pays player 1 the amount
# 1 + eps and player 2 the amount 1; matching on Y pays the reverse.
# Level-k account in the spirit of Crawford, Gneezy and Rottenstreich
# (2008): when eps = 0 the naive level-0 player follows label salience
# (picks X with 0.5 + sig); as soon as eps > 0, level-0 follows payoff
# salience instead (own favourite with 0.5 + pi). Level-1 and level-2
# players give logit best responses (precision lam) to the level below.
# Population: shares f0, f1, f2 of levels 0, 1, 2. Model parameters are
# illustrative; the published rates are copied from the 2008 paper.
import math

SIG, PI, LAM = 0.25, 0.25, 8.0
F0, F1, F2 = 0.2, 0.5, 0.3
PUBLISHED = [("sym 100/100 (hypothetical)", 0.82),
             ("asym 100/101 (hypothetical)", 0.52),
             ("X-Y sym 5/5 (paid)", 0.64),
             ("X-Y asym 5/5.1 (paid)", 0.38)]


def logit_x(ux, uy):
    return 1.0 / (1.0 + math.exp(-LAM * (ux - uy)))


def level0(eps, me):
    # probability that player `me` (1 or 2) picks X at level 0
    if eps == 0:
        return 0.5 + SIG
    return 0.5 + PI if me == 1 else 0.5 - PI


def best(eps, me, q_other_x):
    # payoffs of matching on X / Y for player `me`
    px, py = (1 + eps, 1.0) if me == 1 else (1.0, 1 + eps)
    return logit_x(q_other_x * px, (1 - q_other_x) * py)


def rates(eps):
    l1 = {m: best(eps, m, level0(eps, 3 - m)) for m in (1, 2)}
    l2 = {m: best(eps, m, l1[3 - m]) for m in (1, 2)}
    mix = {m: math.fsum([F0 * level0(eps, m), F1 * l1[m], F2 * l2[m]])
           for m in (1, 2)}
    return mix[1] * mix[2] + (1 - mix[1]) * (1 - mix[2]), mix


print("eps     P1 picks X  P2 picks X  coordination")
for eps in (0.0, 0.01, 0.05, 0.10, 0.50, 1.00):
    c, mix = rates(eps)
    print("%.2f     %.3f       %.3f       %.3f" % (eps, mix[1], mix[2], c))
c_fix, _ = rates(0.0)
print("split the surplus first (eps -> 0): coordination %.3f" % c_fix)
print("published coordination rates (2008):")
for name, r in PUBLISHED:
    print("  %-28s %.2f" % (name, r))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《施为性言语行为、谢林聚点定框与高维语言坍缩商业现实母典》第 2 章「谢林聚点:多重均衡里怎样想到一处,又何时失灵」(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-428 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026428,
  author    = {Fan, Yafu},
  title     = {Performative Speech Acts, Schelling Focal-Point Framing and Verbal Reality-Collapse in High-Stakes Negotiation},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论