5D Library › Short-Video Completion Hazards, Recommender Exploration Loops and Automated Acquisition Funnels · 中文
Chapter 3. Feedback Loops, the Matthew Effect and Gameable Metrics
The Matthew effect in a recommender loop
A recommender gives exposure to videos that already have likes, and likes bring more exposure: a self-reinforcing loop. Merton used “the Matthew effect” for the greater recognition that already eminent scientists receive (Merton 1968); in network science, preferential attachment, “the rich get richer”, produces power laws (Barabási and Albert 1999); the simplest mathematical model is the Pólya urn (Eggenberger and Pólya 1923). Salganik, Dodds and Watts showed in an experimental music market that letting listeners see others’ downloads made success both more unequal and less predictable (Salganik et al. 2006).
The program spreads the appeal of 48 videos evenly from 0.2 to 0.8; each of 10,000 viewers is shown one video, chosen with probability proportional to (likes + 1) to the power ; is a world without feedback, a Pólya urn, and lets popularity compound. Each runs in eight independent “worlds”.
Calculation
# Chapter 3. p05: The Matthew effect in a recommender feedback loop.
# 48 videos with fixed appeal q_i spread evenly from 0.2 to 0.8. Each
# of 10,000 viewers is shown one video, chosen with probability
# proportional to (likes_i + 1)^b, and likes it with chance q_i.
# b = 0 is a world with no feedback (independent exposure); b = 1 is a
# Polya urn; b = 1.5 lets popularity compound. For each b we run eight
# independent worlds with the same videos (as Salganik et al. did with
# songs) and report: the Gini coefficient of likes, the share of likes
# held by the top video, the rank correlation of likes with appeal,
# and in how many of the eight worlds the best video came first.
import random
def gini(xs):
xs = sorted(xs)
n = len(xs)
cum = sum((i + 1) * x for i, x in enumerate(xs))
return (2 * cum) / (n * sum(xs)) - (n + 1) / n
def ranks(xs):
order = sorted(range(len(xs)), key=lambda i: xs[i])
r = [0.0] * len(xs)
for pos, i in enumerate(order):
r[i] = float(pos)
return r
def spearman(a, b):
ra, rb = ranks(a), ranks(b)
n = len(a)
ma, mb = sum(ra) / n, sum(rb) / n
cov = sum((x - ma) * (y - mb) for x, y in zip(ra, rb))
va = sum((x - ma) ** 2 for x in ra)
vb = sum((y - mb) ** 2 for y in rb)
return cov / (va * vb) ** 0.5
def world(b, rng, n=48, viewers=10000):
q = [0.2 + 0.6 * i / (n - 1) for i in range(n)]
likes = [0] * n
for _ in range(viewers):
w = [(x + 1) ** b for x in likes]
i = rng.choices(range(n), weights=w)[0]
if rng.random() < q[i]:
likes[i] += 1
return q, likes
if __name__ == "__main__":
rng = random.Random(1968)
print(" b Gini top share rank corr best video first")
for b in (0.0, 1.0, 1.5):
g = t = c = 0.0
wins = 0
for _ in range(8):
q, likes = world(b, rng)
g += gini(likes) / 8
t += max(likes) / sum(likes) / 8
c += spearman(q, likes) / 8
wins += likes.index(max(likes)) == len(q) - 1
print(" {:3.1f} {:8.3f} {:10.3f} {:11.3f} {:10d} of 8".format(
b, g, t, c, wins))
print("\nReading: without feedback, likes track appeal closely")
print("(rank correlation 0.970), and the best video wins in half the")
print("worlds only because its neighbours are almost as good. With")
print("feedback, early luck is amplified: inequality rises, the link")
print("between appeal and success weakens, and one video can take")
print("most of the likes. Past view counts measure luck as well as")
print("quality.")
# b Gini top share rank corr best video first
# 0.0 0.214 0.035 0.970 4 of 8
# 1.0 0.702 0.186 0.794 1 of 8
# 1.5 0.964 0.853 0.621 2 of 8
#
# Reading: without feedback, likes track appeal closely
# (rank correlation 0.970), and the best video wins in half the
# worlds only because its neighbours are almost as good. With
# feedback, early luck is amplified: inequality rises, the link
# between appeal and success weakens, and one video can take
# most of the likes. Past view counts measure luck as well as
# quality.Without feedback the Gini coefficient of likes is 0.214 and the rank correlation of likes with appeal 0.970; the best video wins in only half the worlds, simply because its neighbours are almost as good. At the Gini coefficient rises to 0.702 and the rank correlation falls to 0.794; at the Gini coefficient reaches 0.964, the winner takes 0.853 of all likes, and the rank correlation is only 0.621. In a world with feedback, past views measure luck as well as quality.
Which metric can editing length game?
Goodhart observed that a statistical regularity tends to collapse once it is used for control (Goodhart 1975). Short-video metrics are no exception. The program takes five videos that differ in length and in “holding power” (the Weibull scale ) and compares four rankings: completion rate, mean watch time, completion normalised by the baseline for the same length ( seconds), and holding power recovered directly from completion.
Calculation
# Chapter 3. p06: Which metric can be gamed by video length?
# Five videos of different lengths D and different "holding power" L
# (the Weibull scale of p01, shape k = 0.6; larger L = viewers stay
# longer at every moment). The platform can rank them by:
# completion rate S(D);
# mean watch time (seconds);
# length-normalised completion S(D) / S0(D), where S0 is the
# platform-wide baseline curve with L = 20 s for that length;
# fitted holding power D / (-ln S(D))^(1/k), which inverts the curve.
# True quality here is L alone. A metric is gameable if a creator can
# raise it by changing length rather than holding power.
import math
K = 0.6
def surv(t, L):
return math.exp(-(t / L) ** K)
def mean_watch(D, L, n=2000):
return sum(surv((i + 0.5) * D / n, L) for i in range(n)) * D / n
VIDEOS = [("A", 7, 16.0), ("B", 15, 24.0), ("C", 30, 20.0),
("D", 60, 28.0), ("E", 90, 18.0)]
def order(vals):
return "".join(n for n, _ in sorted(vals, key=lambda x: -x[1]))
if __name__ == "__main__":
print(" video D (s) L (s) completion watch (s) normalised")
rows = []
for name, D, L in VIDEOS:
c, w, z = surv(D, L), mean_watch(D, L), surv(D, L) / surv(D, 20.0)
rows.append((name, L, c, w, z))
print(" {:5s} {:6d} {:7.1f} {:12.3f} {:11.1f} {:12.3f}".format(
name, D, L, c, w, z))
print("\n ranking by true holding power L : " +
order([(r[0], r[1]) for r in rows]))
print(" ranking by completion rate : " +
order([(r[0], r[2]) for r in rows]))
print(" ranking by mean watch time : " +
order([(r[0], r[3]) for r in rows]))
print(" ranking by normalised completion: " +
order([(r[0], r[4]) for r in rows]))
print(" ranking by fitted holding power : " + order(
[(n, D / (-math.log(surv(D, L))) ** (1 / K)) for n, D, L in VIDEOS]))
print("\n video A cut from 7 s to 5 s: completion {:.3f} -> {:.3f},"
.format(surv(7, 16.0), surv(5, 16.0)))
print(" normalised {:.3f} -> {:.3f}".format(
surv(7, 16.0) / surv(7, 20.0), surv(5, 16.0) / surv(5, 20.0)))
print("\nReading: completion rewards short videos and watch time")
print("rewards long ones, whatever their quality; each can be raised")
print("by editing length alone. Comparing each video with the")
print("baseline for its own length nearly recovers the true order")
print("(only the two weakest swap) and barely moves when a video is")
print("trimmed; fitting the holding power itself recovers it fully.")
# video D (s) L (s) completion watch (s) normalised
# A 7 16.0 0.544 4.8 0.927
# B 15 24.0 0.470 9.5 1.091
# C 30 20.0 0.279 14.3 1.000
# D 60 28.0 0.206 24.3 1.424
# E 90 18.0 0.072 22.0 0.851
#
# ranking by true holding power L : DBCEA
# ranking by completion rate : ABCDE
# ranking by mean watch time : DECBA
# ranking by normalised completion: DBCAE
# ranking by fitted holding power : DBCEA
#
# video A cut from 7 s to 5 s: completion 0.544 -> 0.608,
# normalised 0.927 -> 0.940
#
# Reading: completion rewards short videos and watch time
# rewards long ones, whatever their quality; each can be raised
# by editing length alone. Comparing each video with the
# baseline for its own length nearly recovers the true order
# (only the two weakest swap) and barely moves when a video is
# trimmed; fitting the holding power itself recovers it fully.The true order of holding power is D, B, C, E, A. Completion ranks the videos exactly by length from short to long (A, B, C, D, E): the 7-second video A has the weakest holding power yet the highest completion, 0.544. Mean watch time favours long videos: the 90-second video E is second weakest yet second in watch time (22.0 seconds). Completion normalised by the same-length baseline almost recovers the true order, with only the two weakest swapped; recovering holding power directly restores it fully. Cutting A from 7 to 5 seconds raises its completion from 0.544 to 0.608 but its normalised value only from 0.927 to 0.940.
Hypothesis · not established
Testable consequence. If a platform or team rewards creators on raw completion, the distribution of video lengths should shift towards the short end after the reward starts, while the same-length normalised metric should not rise with it. What would show this wrong: if lengths do not change, or the normalised metric rises together with completion, the higher completion comes from better content rather than from editing length.
Summary
Recommender feedback magnifies early luck into lasting gaps, so past view counts are not direct evidence of quality. Completion and watch time can both be manipulated by editing length; videos should be compared against the baseline for their own length, or by fitting holding power directly.
References
Fan, Y. (2026). Short-Video Completion Hazards, Recommender Exploration Loops and Automated Acquisition Funnels: Weibull Swipe-Away Kinetics, Multi-Stage Pool Promotion Thresholds and Traffic-to-Cashflow Conversion Bounds [短视频完播率风险函数、推荐反馈环与自动化获客漏斗] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090213 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).