Last night, after finishing my backtest, I scrolled past Papi Jiang's Angry. My first reaction wasn't laughter, but calculating its propagation efficiency. 244,000 people filmed along with it, and the Douyin hashtag views exceeded 100 million. This is no longer an entertainment event; it's a strategy with an extremely high "Sharpe Ratio" for AI content on social networks.

I pulled a set of data overnight, covering content production costs, propagation decay coefficients, and user behavior conversion rates, attempting to deconstruct the quantitative logic behind this phenomenon. The conclusion is direct: Papi Jiang using AI to create the #1 hit song of 2026 is essentially finding an arbitrage window with "low slippage, high return."

1. Production Side: Cost Structure Approaching Zero

First, look at production costs. Traditional hit songs require arrangement, mixing, and post-production, but AI music generation tools (such as Suno, Udio) have compressed marginal costs to near zero. Papi Jiang's team only needs to do two things:

• Input prompts: Similar to "angry, repetitive, catchy, female voice"
• Overlay her performance talent (expressions, sense of rhythm)

This production function is no longer linear growth but exponential decline. We can model it:

Production Efficiency = (Output Quality) / (Input Cost)
Traditional Mode: 10-point hit song / 1 million cost = 0.0001
AI Mode: 8-point hit song / 0.01 yuan cost = 800

Although AI-generated quality hasn't reached the level of top producers, Papi Jiang's secondary interpretation (performance + editing) compensates for this. The final product's propagation power on social media is far higher than pure AI works of equivalent quality.

[!note] Key Metric: Unit Cost Propagation Value (UCPV)

Traditional Hit Song UCPV ≈ 5,000 views/10k yuan

AI Hit Song UCPV ≈ 50 million views/10k yuan

Difference of 4 orders of magnitude

2. Propagation Side: Algorithm Dividend of Viral Diffusion

Douyin's recommendation algorithm is essentially a time-series propagation model. Papi Jiang's Angry hit three key nodes:

• Emotional Anchor: Anger, helplessness, humor—these emotions belong to high-forwarding tags in the algorithm.
• Secondary Creation Threshold: Simple melody, and since it uses AI generation, anyone can imitate, adapt, or even use the same AI tool to generate their own version (Real-world case: A large number of users have already used AI to generate variants of Angry, tagging them with Papi Jiang).
• Time Window: July 2026, peak summer traffic, and the AI music topic has not yet been over-consumed.

I performed a backtest using a simple SIR model:

S (Susceptible Population) = Douyin DAU 800 million
I (Infected) = Initial Papi Jiang Fans 50 million
R (Recovered) = Viewed but not forwarded Assumed 30%

Propagation Rate β = 0.0001 (Probability of forwarding after viewing per user)
Recovery Rate γ = 0.02 (Probability of losing interest daily)

Basic Reproduction Number R0 = β / γ = 5.0

The R0 for traditional hit songs usually falls between 2-3, while Angry's R0 exceeds 5, meaning each user propagates to an average of 5 new users. What does this number mean quantitatively? It means if Papi Jiang's team invested 1 million in buying traffic during the initial release, the ROI could reach over 10x.

3. Core Viewpoint: The Essence of AI Music is "Low Slippage Trading"

In live trading, what I fear most is slippage—the gap between expected returns and actual execution prices. In content propagation, slippage is the decay of "users seeing but not forwarding."

The biggest slippage for traditional music producers comes from two places:

• High production costs leading to few trial-and-error attempts.
• Limited distribution channels preventing rapid iteration.

AI music eliminates the first slippage, and Papi Jiang's performance ability eliminates the second slippage (she herself is a super channel). The effect of this combination punch is: You can test a large number of content variants at extremely low cost and in a very short time until you find the version with R0 > 5.

Comparison:

• Traditional Mode: 1 song, 5 million production cost, betting on a hit.
• AI Mode: 100 songs, 5,000 yuan cost, where 1 hit covers all costs.

This is the "high-frequency sampling" logic in quantitative hedging strategies. Papi Jiang didn't make 100 songs, but she achieved the effect of "one sample, law of large numbers taking effect" through AI generation + performance.

4. Risk Warning: Decay Factors to Consider in Live Trading

Of course, this strategy is not risk-free. Several regression factors need attention:

• User Fatigue: When AI music floods the market, algorithms will lower the weight of such content, manifested as a steeper "click-through rate decay curve."
• Copyright Disputes: The copyright ownership of AI-generated content is not yet clear. If legal disputes arise, propagation will be interrupted (similar to "slippage penalty").
• Imitator Effect: Papi Jiang's success will be heavily copied, leading to oversupply and lowering average propagation efficiency.

From a quantitative perspective, this is already a "crowded trade." But as the first person to eat the crab, Papi Jiang's Sharpe Ratio is extremely high. Subsequent followers face lower R0 and higher traffic-buying costs.

5. Code Snippet: Simulating the Propagation Path of AI Hit Songs

```python

import numpy as np

import matplotlib.pyplot as plt

Parameters

N = 800_000_000 # Douyin DAU

I0 = 50_000_000 # Initial Fans

S0 = N - I0

R0 = 0

beta = 0.0001

gamma = 0.02

days = 30

SIR Model

S, I, R = [S0], [I0], [R0]

for t in range(1, days):

dS = -beta S[-1] I[-1] / N

dI = beta S[-1] I[-1] / N - gamma * I[-1]

dR = gamma * I[-1

Original link: https://www.tmtpost.com/8068419.html