What does Short-Video Traffic Pools conclude, in numbers, about “Branching Processes: Galton–Watson Cascades, Criticality and Overdispersion”?

Answer: Chapter 2 uses branching processes to show that raising transmission strength by a tenth can multiply the chance of a large cascade by 4.1.

Scope of the numbers: Chapter 3 shows that a hard threshold is a ramp in the data, whose width narrows from 0.124 to 0.023 as the batch grows from 100 to 3,000 viewers; the illustrative base video breaks out with probability 0.03585, with the bottleneck at the fourth gate. Marked boxes separate published evidence, runnable Python calculations (18 programs), classical sources and the author's hypotheses.

中文版

Data table

The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 4.1

mP(die out)mean sizeP(N>=100)P(N>=1000)
0.501.00002.00.00000.00000
0.801.00005.00.00140.00000
0.901.000010.00.01740.00001
0.951.000020.00.04130.00152
0.991.0000100.00.07100.01663
1.001.0000inf0.08000.02524
1.050.9063inf0.13200.09526
1.200.6863inf0.31570.31370
1.500.4172inf0.58280.58281
near m = 1 the tail of N falls like n^(-1/2) for P(N >= n);
a 10 percent rise in m from 0.9 to 0.99 multiplies P(N>=100)
by 4.1

Key formula

\[P(N = n) = \frac{e^{-mn}(mn)^{n-1}}{n!}.\]
Show the full program
# Chapter 2. p21: Galton-Watson cascades with Poisson offspring.
# Each viewer who shares brings in a Poisson(m) number of new sharers
# (m = mean reshares per sharer, illustrative). The cascade dies out
# with probability q, the smallest root of q = exp(m (q - 1)), found
# by iterating from 0 (the Watson iteration of Chapter 7). The total
# number of sharers from one seed follows the Borel law
#     P(N = n) = exp(-m n) (m n)^(n - 1) / n!,
# with mean 1 / (1 - m) when m < 1. For m > 1 the Borel terms
# leave out the mass of never-ending cascades, so the tail includes
# it. We report P(N >= 100) exactly as one minus the sum of the first
# 99 terms, added with math.fsum.
import math


def extinction(m, steps=20000):
    # m <= 1: extinction is certain (theorem); iteration would creep up
    if m <= 1.0:
        return 1.0
    q = 0.0
    for _ in range(steps):
        q = math.exp(m * (q - 1.0))
    return q


def borel(m, n):
    lg = -m * n + (n - 1) * math.log(m * n) - math.lgamma(n + 1)
    return math.exp(lg)


print("    m  P(die out)  mean size  P(N>=100)  P(N>=1000)")
for m in (0.5, 0.8, 0.9, 0.95, 0.99, 1.0, 1.05, 1.2, 1.5):
    q = extinction(m)
    mean = "%9.1f" % (1.0 / (1.0 - m)) if m < 1.0 else "      inf"
    t100 = 1.0 - math.fsum(borel(m, n) for n in range(1, 100))
    t1000 = 1.0 - math.fsum(borel(m, n) for n in range(1, 1000))
    print("%5.2f  %10.4f  %s  %9.4f  %10.5f"
          % (m, q, mean, t100, t1000))
print("near m = 1 the tail of N falls like n^(-1/2) for P(N >= n);")
print("a 10 percent rise in m from 0.9 to 0.99 multiplies P(N>=100)")
print("by %.1f" % ((1.0 - math.fsum(borel(0.99, n) for n in range(1, 100)))
                    / (1.0 - math.fsum(borel(0.9, n)
                                       for n in range(1, 100)))))

Single source and citation

Fan, Y. (2026). Cold-Start Dynamics of Short-Video Traffic Pools: Step Thresholds in Completion, Engagement and Replay and a Zero-Ad-Spend Breakout Test Matrix (Version 1.0.0), Chapter 2: Branching Processes: Galton–Watson Cascades, Criticality and Overdispersion. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 5DL-B-273 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026273,
  author    = {Fan, Yafu},
  title     = {Cold-Start Dynamics of Short-Video Traffic Pools: Step Thresholds in Completion, Engagement and Replay and a Zero-Ad-Spend Breakout Test Matrix},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

Other chapter answers in this book