关于「实战工具与检查器规格:先手探针台账、预注册检验与证伪方案」,《先手探针与隐藏事实》给出的定量结论是什么?

结论:第八章给出先手探针台账、蒸馏小工具「先手扰动探针设计与信息增益试算器」、按四件套分组的检查器规格与预注册证伪方案。所有探针真实可买、如实交付、预售如实标注,零虚假稀缺、零会重置的倒计时,不冒充客户询价,不按受保护特征差别定价。书中以标注框区分已发表的证据、可运行的 Python 计算(21 段程序)、原典与作者的假说。

数据口径:书中参数为示意,全部计算由书中程序生成,数据全部是合成数据。 本书是非个人化经营建议,不构成法律、税务或投资意见,不向任何人承诺收入。

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

idtypepostEIGrealnetverdict
T1limited tier 0.0690.950.848.9GREEN
T2waitlist0.1270.910.7117.8GREEN
T3preorder0.0800.630.24-4.2RED:M
T4open call0.6001.161.8018.9GREEN
T5rate card0.4090.570.1914.1GREEN
T6limited tier 0.0430.380.10-30.5RED:H
T7limited tier 0.0500.270.06-33.1YELLOW
T8quote ask0.3750.340.098.4RED:P
查看完整验算程序
# Chapter 8. p81: scoring a week of the first-move probe ledger.
# Eight synthetic ledger rows (no real customer, order or counterpart).
# Each row: id, site, probe type, prior Beta(a, b) on the response rate,
# visitors or invitees n, responses k, set-up cost, value of a bit, and
# compliance flags: H = hypothesis registered before launch, R = price
# really payable and delivered, S = no invented stock or deadline,
# P = public channel only (no posing as a client), M = sample at least
# the registered minimum before any price change.
# Computed: posterior mean, expected information gain before launch
# (Lindley, bits), realised gain (KL posterior || prior, bits), net
# value, and a red / yellow / green verdict.
import math

ROWS = [
    ("T1", "academy", "limited tier", 2, 38, 120, 9, 15, "HRSPM"),
    ("T2", "academy", "waitlist", 3, 27, 80, 11, 5, "HRSPM"),
    ("T3", "homie-x", "preorder", 2, 18, 30, 2, 20, "HRSP"),
    ("T4", "aix", "open call", 4, 6, 40, 26, 10, "HRSPM"),
    ("T5", "aix", "rate card", 4, 6, 12, 5, 0, "HRSPM"),
    ("T6", "homie-x", "limited tier", 2, 38, 30, 1, 40, "RSPM"),
    ("T7", "academy", "limited tier", 2, 38, 20, 1, 40, "HRSPM"),
    ("T8", "aix", "quote ask", 4, 6, 6, 2, 0, "HRSM"),
]
BIT, MIN_N = 25.0, 50


def digamma(x):
    acc = []
    while x < 6.0:
        acc.append(-1.0 / x)
        x += 1.0
    f = 1.0 / (x * x)
    acc.append(math.log(x) - 0.5 / x - f * (1.0 / 12 - f * (
        1.0 / 120 - f * (1.0 / 252 - f / 240))))
    return math.fsum(acc)


def lbeta(a, b):
    return math.lgamma(a) + math.lgamma(b) - math.lgamma(a + b)


def h_beta(a, b):
    return (lbeta(a, b) - (a - 1) * digamma(a) - (b - 1) * digamma(b)
            + (a + b - 2) * digamma(a + b))


def eig(a, b, n):
    terms = []
    for k in range(n + 1):
        lp = (math.lgamma(n + 1) - math.lgamma(k + 1)
              - math.lgamma(n - k + 1) + lbeta(a + k, b + n - k)
              - lbeta(a, b))
        terms.append(math.exp(lp) * h_beta(a + k, b + n - k))
    return (h_beta(a, b) - math.fsum(terms)) / math.log(2)


def kl_bits(a1, b1, a0, b0):
    nats = (lbeta(a0, b0) - lbeta(a1, b1) + (a1 - a0) * digamma(a1)
            + (b1 - b0) * digamma(b1)
            + (a0 - a1 + b0 - b1) * digamma(a1 + b1))
    return nats / math.log(2)


def verdict(flags, n, net):
    missing = [f for f in "HRSP" if f not in flags]
    if missing:
        return "RED:" + "".join(missing)
    if "M" not in flags and n < MIN_N:
        return "RED:M"
    return "YELLOW" if net < 0 else "GREEN"


if __name__ == "__main__":
    print("id  type          post  EIG   real  net    verdict")
    for rid, site, kind, a, b, n, k, cost, flags in ROWS:
        post = (a + k) / (a + b + n)
        g = eig(a, b, n)
        real = kl_bits(a + k, b + n - k, a, b)
        net = BIT * g - cost
        print("{:<3} {:<12} {:>5.3f} {:>5.2f} {:>5.2f} {:>6.1f}  {}".format(
            rid, kind, post, g, real, net, verdict(flags, n, net)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《高能动性主动推理、意念强度先手扰动与商业隐藏事实勘探母典》第 8 章「实战工具与检查器规格:先手探针台账、预注册检验与证伪方案」(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-395 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026395,
  author    = {Fan, Yafu},
  title     = {High-Agency Active Inference, Intentional Perturbation Probes and Hidden Market-Truth Exploration},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论