关于「当评分变成目标」,《结构洞、函子迁移与灵感共振》给出的定量结论是什么?

结论:第五章说明评分成为目标之后,前十名的真实价值从 1.88 个标准差降到 0.48;真实对应每 500 对只有 1 对时,一道测试的精确率只有 0.031。

数据口径:书中的网络、领域、评分与势阱均为示意模型;全部计算由书中程序生成;本书不是投资、医疗或心理建议。 第六章用模拟检验克莱默斯公式(相差不超过 1.32 倍),并发现噪声太少是主要风险:低噪声下只有 0.017 的运行找到更好的状态,而先孵化后专注的节奏(0.730)并不比稳定的中等噪声(0.750)更好。

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: 0.48, 1.88

gaming meantop-10 score (sd)top-10 true value (sd)
0.002.651.88
0.252.761.61
0.503.200.99
1.003.910.48
best possible top-10 true value 2.65
查看完整验算程序
# Chapter 5. p09: A four-part score as a target. 1,000 candidate ideas
# have a true value V (standard normal). Four measured metrics (span,
# spectral match, predictive gain, verifiability) each equal 0.5 V plus
# noise of spread 1. The score is their sum. Then a generator that can
# raise the measured metrics without raising V: each idea gets a hidden
# "gaming" term G (exponential, mean m) added to every metric. We pick
# the top 10 by score and report their mean score and mean true value,
# in units of the population's spread. 500 repetitions, seed 102.
import math
import random

rng = random.Random(102)

def trial(m):
    ideas = []
    for _ in range(1000):
        v = rng.gauss(0, 1)
        g = rng.expovariate(1 / m) if m > 0 else 0.0
        s = math.fsum(0.5 * v + g + rng.gauss(0, 1) for _ in range(4))
        ideas.append((s, v))
    ss = [s for s, _ in ideas]
    mu = math.fsum(ss) / len(ss)
    sd = (math.fsum((x - mu) ** 2 for x in ss) / len(ss)) ** 0.5
    top = sorted(ideas, reverse=True)[:10]
    best = sorted((v for _, v in ideas), reverse=True)[:10]
    return (math.fsum((s - mu) / sd for s, _ in top) / 10,
            math.fsum(v for _, v in top) / 10, math.fsum(best) / 10)

if __name__ == "__main__":
    print("   gaming mean   top-10 score (sd)   top-10 true value (sd)")
    for m in (0.0, 0.25, 0.5, 1.0):
        a = b = c = 0.0
        for _ in range(500):
            x, y, z = trial(m)
            a, b, c = a + x / 500, b + y / 500, c + z / 500
        print("   {:11.2f}   {:17.2f}   {:22.2f}".format(m, a, b))
    print("   best possible top-10 true value {:.2f}".format(c))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《跨学科结构洞桥接、函子知识迁移与灵感随机共振》第 5 章「当评分变成目标」(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-102 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026102,
  author    = {Fan, Yafu},
  title     = {Cross-Disciplinary Structural Holes, Functorial Knowledge Transfer and Inspiration Stochastic Resonance},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论