关于「距离买到的是方差,不是更好的平均」,《结构洞、函子迁移与灵感共振》给出的定量结论是什么?
结论:第二章说明距离提高的是方差而不是平均:模型中突破的机会在距离 5 处最高(0.02275),距离 4 处每次突破伴随 23.46 次失败;最好的项目只把 0.3 的部件放在远处。
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。
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: 0.02275, 23.46
| distance | mean | spread | P(failure) | P(breakthrough) |
|---|---|---|---|---|
| 1 | -0.06 | 0.80 | 0.1200 | 0.00069 |
| 2 | -0.24 | 1.10 | 0.2448 | 0.00637 |
| 3 | -0.54 | 1.40 | 0.3712 | 0.01495 |
| 4 | -0.96 | 1.70 | 0.4906 | 0.02091 |
| 5 | -1.50 | 2.00 | 0.5987 | 0.02275 |
| 6 | -2.16 | 2.30 | 0.6930 | 0.02138 |
| 7 | -2.94 | 2.60 | 0.7722 | 0.01821 |
| 8 | -3.84 | 2.90 | 0.8363 | 0.01440 |
| breakthrough chance peaks at distance 5 (0.02275) | ||||
| at distance 4: 23.46 failures per breakthrough |
查看完整验算程序
# Chapter 2. p03: Distance raises variance, not the average. Following
# Fleming's reading of patent data, let the log-impact of a combination
# of two fields at semantic distance d (shortest path in a field map) be
# normal with mean mu(d) = -0.06 d^2 and spread s(d) = 0.5 + 0.3 d.
# Illustrative values. A "breakthrough" is log-impact above 2.5; a
# "failure" is below -1. Exact normal probabilities, no sampling.
from math import erf, sqrt
def cdf(x):
return 0.5 * (1 + erf(x / sqrt(2)))
def mu(d):
return -0.06 * d * d
def s(d):
return 0.5 + 0.3 * d
if __name__ == "__main__":
print(" distance mean spread P(failure) P(breakthrough)")
best = None
for d in range(1, 9):
pf = cdf((-1 - mu(d)) / s(d))
pb = 1 - cdf((2.5 - mu(d)) / s(d))
best = max(best or (pb, d), (pb, d))
print(" {:8d} {:5.2f} {:7.2f} {:11.4f} {:16.5f}".format(
d, mu(d), s(d), pf, pb))
print(" breakthrough chance peaks at distance {} ({:.5f})".format(
best[1], best[0]))
print(" at distance 4: {:.2f} failures per breakthrough".format(
cdf((-1 - mu(4)) / s(4)) / (1 - cdf((2.5 - mu(4)) / s(4)))))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 2 章「距离买到的是方差,不是更好的平均」(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}
}