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。

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: 12, 18571, 24, 36, 40, 49308

A. monthly views; A12 clips, B = 2 long videos, floor 40
month1: A 49,766 B 18,571
month6: A 51,937 B 40,323
month 12: A51,937 B 49,308
month 24: A51,937 B 60,149
month 36: A51,937 B 65,568
B. first month in which B passes A
floor/day monthly cumulative weighted (wA1, wB = 2)
20 none none3
4015 34 3
807 13 2
1604 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}
}

本书其他章节结论