关于「深度信任:Beta 信念更新、可核验证据片段与预售的分母」,《长视频半衰期与深度信任》给出的定量结论是什么?
结论:第五章用 Beta 信念更新描述深度信任,高质量证据下 12 分钟视频每千位起播观众约 13.2 位准备购买。
数据口径:示意曲线下,短视频头 48 小时拿到三年观看的 0.750,带搜索地板的长视频只拿到 0.059,第 1,095 天仍有每天 5.00 次观看;半衰期翻十倍时,短内容峰值要高出 9.96 倍才打平现值。 书中参数为示意,全部计算由书中程序生成。
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: 12, 13.2
| ready viewers per 1,000 starters | ||
| length | q = 0.9 | q = 0.6 |
| 4 min | 0.0 | 0.0 |
| 8 min | 0.0 | 0.0 |
| 12 min | 13.2 | 1.4 |
| 16 min | 27.4 | 3.2 |
| 20 min | 38.2 | 4.5 |
| share of starters still watching at 12 min: 0.091 |
查看完整验算程序
# Chapter 5. p52: ready viewers per 1,000 starters, by video length.
# 20,000 simulated starters (fixed seed). Leave time ~ Weibull k = 0.6
# with S(30 s) = 0.70 (p41). Prior ~ one of the four p51 priors with
# weights 0.3 / 0.2 / 0.3 / 0.2. A checkable segment arrives every 2
# minutes; each confirms with probability q (evidence quality). A viewer
# who has seen n segments is "ready" by the p51 rule (mean >= 0.75,
# sd <= 0.12). Lengths 4 to 20 minutes; q = 0.9 or 0.6. Illustrative.
import math
import random
PRIORS = [(1.0, 3.0), (2.0, 6.0), (2.0, 2.0), (4.0, 2.0)]
WEIGHTS = [0.3, 0.2, 0.3, 0.2]
K = 0.6
LAM = 30.0 / (-math.log(0.70)) ** (1.0 / K)
N = 20000
def ready(a, b, n, s):
a2, b2 = a + s, b + n - s
m = a2 / (a2 + b2)
sd = math.sqrt(a2 * b2 / ((a2 + b2) ** 2 * (a2 + b2 + 1.0)))
return m >= 0.75 and sd <= 0.12
def run(length_min, q):
rng = random.Random(20261120)
hits = 0
for _ in range(N):
a, b = rng.choices(PRIORS, WEIGHTS)[0]
leave = LAM * (-math.log(1.0 - rng.random())) ** (1.0 / K)
watched = min(leave, length_min * 60.0)
n = int(watched // 120.0)
s = 0
for _ in range(n):
s += rng.random() < q
hits += ready(a, b, n, s)
return 1000.0 * hits / N
if __name__ == "__main__":
print("ready viewers per 1,000 starters")
print(" length q = 0.9 q = 0.6")
for L in (4, 8, 12, 16, 20):
print(" {:>3} min {:>6.1f} {:>6.1f}".format(
L, run(L, 0.9), run(L, 0.6)))
stay = math.exp(-(720.0 / LAM) ** K)
print("share of starters still watching at 12 min: {:.3f}".format(stay))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《长视频搜索半衰期红利、深度信任预售与多年期被动获客资产母典》第 5 章「深度信任:Beta 信念更新、可核验证据片段与预售的分母」(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-370 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026370,
author = {Fan, Yafu},
title = {Long-Form Video Search Half-Lives, Deep-Trust Pre-Selling and Multi-Year Passive Acquisition Assets},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}