关于「实战:一张样盘,三种人生」,《同盘异象:身份调制》给出的定量结论是什么?

结论:第八章让三位同一个时辰出生的虚构人物共用一张财帛大限样盘,逐年调制,十年里规则分别压掉 4、1 与 2 档,并给出带巴纳姆检查的四选一盲测:真实挑中率 0.35 时需要 129 人。书中以标注框区分已发表的证据、可运行的计算、原典与作者的假说,每条假说都写明什么结果会推翻它。

数据口径:本书不是投资、法律、税务、医疗或心理意见,全部计算由书中程序生成。 书中以标注框区分已发表的证据、可运行的计算、原典与作者的假说,每条假说都写明什么结果会推翻它。

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: 0.35, 129

H1: own page picked, chance 0.25
    true rate 0.35: 129 participants, 43 triads of three (reject chance if picks >= 41)
    true rate 0.40: 62 participants, 21 triads of three (reject chance if picks >= 22)
    true rate 0.45: 36 participants, 12 triads of three (reject chance if picks >= 14)
null simulation, 129 participants, 20000 runs, seed 131, critical value 41:
    H1 fires by chance 0.0500; H2 alarm fires by chance 0.0486
    if a flattering generic page is picked 0.40 of the time, H2 alarm fires 0.9765
查看完整验算程序
# Chapter 8. p83: A pre-registered blind test of persona modulation, with a Barnum control.
# Each participant, at the end of a decade year, reads four anonymised one-page write-ups for the same
# reading cell (decade palace x wave): the one modulated for their own role, stage and track, two
# modulated for other personas, and one generic text with no persona at all. They pick the page that
# fits their year best. Chance of picking their own page: 1/4.
#   H1 (modulation): own-page picks above 1/4.   H2 (Barnum check): generic-page picks NOT above 1/4.
# Exact one-sided binomial, alpha 0.05, power 0.80. Then a null simulation (seed 131): with no real
# skill, how often does each test fire, and how often does a flattering generic page win H2's alarm?
import random
from math import exp, lgamma, log

ALPHA, POWER, SEED, SIMS = 0.05, 0.80, 131, 20000


def tail(n, c, p):
    lp, lq = log(p), log(1 - p)
    return sum(exp(lgamma(n + 1) - lgamma(x + 1) - lgamma(n - x + 1) + x * lp + (n - x) * lq)
               for x in range(c, n + 1))


def crit(n, p0):
    c = int(n * p0)
    while tail(n, c, p0) > ALPHA:
        c += 1
    return c


def n_exact(p0, p1):
    n = 8
    while True:
        c = crit(n, p0)
        if tail(n, c, p1) >= POWER:
            return n, c
        n += 1


if __name__ == "__main__":
    print("H1: own page picked, chance 0.25")
    for p1 in (0.35, 0.40, 0.45):
        n, c = n_exact(0.25, p1)
        print("    true rate %.2f: %d participants, %d triads of three (reject chance if picks >= %d)"
              % (p1, n, -(-n // 3), c))
    n = 3 * -(-n_exact(0.25, 0.35)[0] // 3)          # whole triads for the 0.35 design
    c = crit(n, 0.25)
    rng = random.Random(SEED)
    fire_own = fire_gen = 0
    for _ in range(SIMS):
        own = gen = 0
        for _ in range(n):
            u = rng.random()
            own += u < 0.25
            gen += 0.25 <= u < 0.50
        fire_own += own >= c
        fire_gen += gen >= c
    print("null simulation, %d participants, %d runs, seed %d, critical value %d:" % (n, SIMS, SEED, c))
    print("    H1 fires by chance %.4f; H2 alarm fires by chance %.4f" % (fire_own / SIMS, fire_gen / SIMS))
    rng = random.Random(SEED)
    alarm = 0
    for _ in range(SIMS):
        gen = sum(1 for _ in range(n) if rng.random() < 0.40)
        alarm += gen >= c
    print("    if a flattering generic page is picked 0.40 of the time, H2 alarm fires %.4f" % (alarm / SIMS))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《同盘异象:事业角色、婚育阶段与行业赛道断语调制母典》第 8 章「实战:一张样盘,三种人生」(版本 0.1.0). 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-131 · v0.1.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026131,
  author    = {Fan, Yafu},
  title     = {The Same Chart, Different Lives Codex},
  year      = {2026},
  publisher = {5D Library},
  edition   = {0.1.0},
  note      = {DOI pending}
}

本书其他章节结论