关于「半衰期:把「48 小时速朽」与「三年持续进客」写成参数」,《长视频半衰期与深度信任》给出的定量结论是什么?
结论:A 段是地板为每天 40 次时的月观看。第一个月,短视频 49,766,长视频 18,571,前者是后者的两倍多。到第 12 个月,短视频是 51,937,长视频 49,308,几乎追平。第 24 个月,长视频 60,149,第 36 个月 65,568,短视频始终停在 51,937。
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, 18571, 24, 36, 40, 49308
| A. monthly views; A | 12 clips, B = 2 long videos, floor 40 |
| month | 1: A 49,766 B 18,571 |
| month | 6: A 51,937 B 40,323 |
| month 12: A | 51,937 B 49,308 |
| month 24: A | 51,937 B 60,149 |
| month 36: A | 51,937 B 65,568 |
| B. first month in which B passes A | |
| floor/day monthly cumulative weighted (wA | 1, wB = 2) |
| 20 none none | 3 |
| 40 | 15 34 3 |
| 80 | 7 13 2 |
| 160 | 4 6 2 |
| C. 36-month totals at floor 40: A 1,867,562 B 1,904,421 |
查看完整验算程序
# Chapter 1. p13: equal production hours, 36 months, stock versus flow.
# Strategy A posts 12 short clips a month (2 hours each, curve S of p11).
# Strategy B posts 2 long videos a month (12 hours each): a push burst
# 1000 * 2^(-t/10) plus a search floor f * 2^(-t/365). Both spend 24
# hours a month. Uploads are spread evenly over each month. We count
# monthly views and value-weighted views, where w is the number of
# qualified actions (description clicks that reach a landing page) per
# 1,000 views: w = 1.0 for clips and w = 2.0 for long videos. Every
# number is illustrative and must be replaced by the reader's own data.
import math
DPM = 30.4375
MONTHS = 36
def cum(comps, t0, t1):
out = []
for a, h in comps:
k = math.log(2.0) / h
out.append(a / k * (math.exp(-k * max(t0, 0.0))
- math.exp(-k * max(t1, 0.0))))
return math.fsum(out)
def monthly(comps, per_month, months=MONTHS):
res = []
for m in range(months):
t0, t1 = m * DPM, (m + 1) * DPM
parts = []
for u in range(m + 1):
for j in range(per_month):
up = (u + (j + 0.5) / per_month) * DPM
if up < t1:
parts.append(cum(comps, t0 - up, t1 - up))
res.append(math.fsum(parts))
return res
def first_cross(a, b):
ca = cb = 0.0
for m in range(len(a)):
ca, cb = ca + a[m], cb + b[m]
if cb > ca:
return m + 1
return None
if __name__ == "__main__":
A = monthly([(3000.0, 1.0)], 12)
print("A. monthly views; A = 12 clips, B = 2 long videos, floor 40")
B40 = monthly([(1000.0, 10.0), (40.0, 365.0)], 2)
for m in (1, 6, 12, 24, 36):
print(" month {:>2}: A {:>7,.0f} B {:>7,.0f}".format(
m, A[m - 1], B40[m - 1]))
print("B. first month in which B passes A")
print(" floor/day monthly cumulative weighted (wA = 1, wB = 2)")
for f in (20.0, 40.0, 80.0, 160.0):
B = monthly([(1000.0, 10.0), (f, 365.0)], 2)
mm = next((m + 1 for m in range(MONTHS) if B[m] > A[m]), None)
cv = first_cross(A, B)
cw = first_cross(A, [x * 2.0 for x in B])
print(" {:>6.0f} {:>6} {:>6} {:>6}".format(
f, "none" if mm is None else mm, "none" if cv is None else cv,
"none" if cw is None else cw))
print("C. 36-month totals at floor 40: A {:,.0f} B {:,.0f}".format(
math.fsum(A), math.fsum(B40)))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 1 章「半衰期:把「48 小时速朽」与「三年持续进客」写成参数」(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}
}