关于「作品引力:优先连接、同质筛选与不可消除的随机项」,《零解释税与作品引力》给出的定量结论是什么?

结论:作品引力由两个机制构成:同质筛选决定来的是谁,口味相近者占比从随机的 0.263 升到 0.849;累积优势决定来多少,同时把质量相同者的盟友数拉开到 8.43 倍。社会影响放大不平等,也放大不可预测性:影响 0.9 时不可预测性是独立条件的两倍有余。把解释时间改投作品,在低重叠受众下盟友中位数从 125 升到 263,在高重叠受众下却未必占优。

数据口径:书中参数一律为示意,全部计算由书中程序生成;本书是非个人化经营建议,不是心理咨询或临床判断,不作任何收益承诺,「显化」只指把构想做成可交付、可核验的作品。 第三章求出解释的盈亏阈值,示意参数下为 0.464,只对阈值以上受众解释的净值是 320.0,对所有人解释是负 203.8。

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: 0.263, 0.849, 8.43

A. focus creator's allies out of 800 (fair share 40)
settingP10P50P90P90/P10near share
attachment only718598.430.263
attach + homophily .10722669.430.629
attach + homophily .051733553.240.849
homophily only .103238471.470.684
B. focus quality is 1.0 in every run: the spread is path noise
查看完整验算程序
# Chapter 6. p61: "gravity" as preferential attachment plus homophily
# plus noise. 20 creators sit at evenly spaced points of a taste circle;
# creator 0 is the focus, with quality 1.0, the rest have quality
# exp(N(0, 0.3)). 800 allies arrive one by one, each with a taste bin.
# With probability eps = 0.1 an ally picks a creator at random (noise);
# otherwise picks creator j with weight (k_j + a) q_j exp(-dist / h),
# k_j = allies so far (Price, 1976; Barabasi and Albert, 1999), h the
# homophily width (McPherson et al., 2001). 100 runs per setting, fixed
# seed; we report quantiles of the focus creator's ally count and the
# share of its allies whose taste lies within 0.1 of its own.
import math
import random

NC, T, RUNS, EPS = 20, 800, 100, 0.1


def circ(a, b):
    d = abs(a - b) % 1.0
    return min(d, 1.0 - d)


def run(rng, h, a, pa):
    q = [1.0] + [math.exp(rng.gauss(0.0, 0.3)) for _ in range(NC - 1)]
    pos = [j / NC for j in range(NC)]
    if h is None:
        aff = [[1.0] * NC for _ in range(NC)]
    else:
        aff = [[math.exp(-circ(pos[b], pos[j]) / h) for j in range(NC)]
               for b in range(NC)]
    k = [0] * NC
    near = 0
    for _ in range(T):
        b = rng.randrange(NC)
        if rng.random() < EPS:
            j = rng.randrange(NC)
        else:
            w = [((k[i] if pa else 0) + a) * q[i] * aff[b][i]
                 for i in range(NC)]
            j = rng.choices(range(NC), weights=w)[0]
        k[j] += 1
        if j == 0 and circ(pos[b], 0.0) <= 0.1 + 1e-9:
            near += 1
    return k[0], near / k[0] if k[0] else 0.0


def q_at(xs, p):
    xs = sorted(xs)
    return xs[int(p * (len(xs) - 1))]


if __name__ == "__main__":
    print("A. focus creator's allies out of {:,} (fair share {})".format(
        T, T // NC))
    print("   setting                 P10  P50  P90  P90/P10  near share")
    for name, h, a, pa in (("attachment only", None, 1.0, True),
                           ("attach + homophily .10", 0.10, 1.0, True),
                           ("attach + homophily .05", 0.05, 5.0, True),
                           ("homophily only .10", 0.10, 1.0, False)):
        rng = random.Random(43261)
        res = [run(rng, h, a, pa) for _ in range(RUNS)]
        ks = [r[0] for r in res]
        nr = math.fsum(r[1] for r in res) / RUNS
        p10, p50, p90 = q_at(ks, 0.1), q_at(ks, 0.5), q_at(ks, 0.9)
        print("   {:<22}  {:>3}  {:>3}  {:>3}  {:>7.2f}  {:>10.3f}".format(
            name, p10, p50, p90, p90 / max(p10, 1), nr))
    print("B. focus quality is 1.0 in every run: the spread is path noise")

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《零解释税生存动力学:放弃低维认同求证、全功率自我显化与作品引力场母典》第 6 章「作品引力:优先连接、同质筛选与不可消除的随机项」(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-432 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026432,
  author    = {Fan, Yafu},
  title     = {Zero Explanation-Tax Dynamics: Abandoning Validation-Seeking, Full-Power Reality Rendering and Proof-by-Artifact Gravity},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论