💰 quantitative finance

"Rich-Get-Richer"? Platform Attention and Earnings Inequality using Patreon Earnings Data

Using Patreon earnings data, this paper demonstrates that platform attention algorithms drive a "rich-get-richer" dynamic where earnings follow a highly concentrated Pareto distribution ( $\alpha \approx 2$ ), with algorithmic shifts disproportionately harming the creator middle class and causing inequality across different social media platforms to converge toward increasingly heavy-tailed distributions.

Original authors: Ilan Strauss, Jangho Yang, Mariana Mazzucato

Published 2026-07-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ilan Strauss, Jangho Yang, Mariana Mazzucato

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the internet as a giant, bustling digital city where everyone is trying to get noticed. In this city, the "mayors" aren't elected by votes or prices; they are the algorithms—secret, invisible rules that decide which street corner gets the most foot traffic and which one stays empty. This paper lives in the world of digital economics, a field that studies how money flows in these online spaces. To understand the story, you need to know two things: first, that attention is the new currency (getting seen is how you get paid); and second, that many things in nature and society follow a "power law." Think of a power law like a pyramid where the bottom is huge and the top is tiny, but the very top is so high it touches the sky. In a normal pyramid, the steps are even. In a power-law pyramid, the bottom is a massive crowd, and the top is a single, lonely giant. This matters because if the rules of the city favor that giant too much, the middle class of the city disappears, and only the superstars survive.

The authors of this paper, Ilan Strauss, Jangho Yang, and Mariana Mazzucato, decided to peek behind the curtain of these digital cities using a special tool: Patreon. Patreon is a website where fans pay creators directly for extra content, but here's the catch: Patreon doesn't really have its own way to show you new creators. To get paid on Patreon, you have to be famous somewhere else first, like on YouTube, Instagram, or Twitter. The creators then put a link to their Patreon on their main profile. Because of this, the money a creator makes on Patreon is a perfect mirror of how much attention they got on their main platform. If a YouTuber gets a million views, they likely get a lot of Patreon fans. If they get zero views, they get zero Patreon fans.

The researchers looked at data from over 100,000 creators across three different years (2018, 2021, and 2024) to see how the money was distributed. They found something startling: the earnings on almost every platform follow a "rich-get-richer" pattern, described mathematically by a number called alpha (α). In their world, a lower alpha means a steeper, more unfair pyramid. They found that for most platforms, this number is around 2.0. This is a very specific number that suggests the system is heavily stacked against the average person. It means that for every step up the ladder you take, the rewards don't just get a little bigger; they explode. The top creators are pulling in money at a rate that looks more like a billionaire's investment portfolio than a regular worker's paycheck.

When they compared the different cities, the results were like a report card on fairness. YouTube and Instagram had the lowest numbers (around 1.8 and 1.84), meaning they are the most unfair. On these platforms, the "middle class" of creators—those who are decently popular but not superstars—are getting squeezed out. Their earnings are lower, and the gap between the rich and the rest is wider. In contrast, Twitter/X and Patreon-only creators had higher numbers (around 2.35 and 2.24), suggesting a slightly flatter, fairer pyramid where more people can make a living, though the top still gets the biggest slice.

The most worrying part of the story is the time travel. When the authors looked at the data from 2018 and compared it to 2024, they saw the pyramids getting steeper. The numbers for Instagram and YouTube dropped, meaning the "rich-get-richer" effect is getting stronger every year. It's as if the city mayor is slowly turning off the lights in the neighborhoods and shining a giant spotlight only on the one house at the very top. The authors suggest this is because the algorithms are becoming better at finding viral content and pushing it to everyone, while ignoring the "long tail" of smaller, niche creators.

However, the authors are careful not to say they have solved the mystery. They admit that while the patterns look like the algorithms are the culprit, other things could be at play. Maybe the people on Twitter just happen to be more generous with their money, or maybe the creators on YouTube are just more competitive. They can't prove with 100% certainty that the algorithm is the only reason, but the evidence strongly suggests that the way these platforms distribute attention is the main driver of this inequality. They also note that as Artificial Intelligence gets better at making content, it might make this problem even worse, as the algorithms could become even more obsessed with the "viral" stuff that already exists, leaving less room for new, different voices to be heard.

In short, this paper paints a picture of a digital world where the rules of the game are slowly changing to favor the superstars even more than before. The "middle class" of creators is shrinking, and unless the invisible rules of the algorithms are tweaked to give a chance to the little guys, the future of online creativity might look less like a diverse city and more like a single, towering skyscraper with everyone else living in the shadows.

Technical Summary: "Rich-Get-Richer"? Platform Attention and Earnings Inequality using Patreon Earnings Data

Problem Statement
Online social media markets are increasingly governed by algorithmic attention allocation rather than price mechanisms. Unlike classical multi-sided market models where prices balance supply and demand, platforms utilize non-price algorithms to process vast data on content and user preferences to allocate visibility. While these algorithms enable market efficiencies, they also function as distributional devices that determine economic outcomes for participants. A critical gap exists in the literature regarding the economic impact of these mechanisms: platforms rarely disclose creator earnings or algorithmic optimization details, making it difficult to empirically test whether algorithmic attention fosters unequal ecosystems or supports a healthy "middle class" of creators. This paper addresses this gap by investigating earnings inequality within platform markets, specifically examining how third-party social media algorithms influence monetization outcomes.

Methodology
The study utilizes a novel, large-scale dataset of monthly Patreon earnings for content creators across three cross-sections: March 2018, 2021, and 2024 ( $n = 104,719$ ). Patreon serves as an ideal proxy for external platform attention because it offers minimal native distribution; creators must link their Patreon pages from external platforms (YouTube, Instagram, Twitch, Twitter/X, Facebook) to drive traffic. Consequently, Patreon earnings largely reflect the attention captured on these primary external platforms.

To isolate platform-specific dynamics, the authors restrict the dataset to creators affiliated with a single external platform (or Patreon-only), removing those with multiple affiliations to control for cross-platform sorting. Missing earnings data, particularly among high earners who tend to keep earnings private, are imputed linearly based on paid members, total members, content type, NSFW status, and year.

The core analytical framework employs power-law distributions ( $p(x) \sim x^{-\alpha}$ ) to characterize earnings. The authors fit these distributions using the maximum likelihood estimation method proposed by Clauset et al. (2009), employing the Kolmogorov-Smirnov (KS) statistic to assess goodness-of-fit. The exponent $\alpha$ serves as the primary metric for inequality: a lower $\alpha$ indicates a heavier tail and higher inequality (consistent with concentrated capital income), while a higher $\alpha$ suggests a more equitable distribution. The analysis interprets these exponents through the lens of preferential attachment models (Barabási and Albert 1999), where $\alpha$ is theoretically linked to the balance between random exploration ( $\gamma$ ) and exploitation of existing popularity ( $1-\gamma$ ).

Key Contributions

Empirical Evidence on Algorithmic Inequality: The paper provides one of the few systematic analyses of content creator earnings distributions, moving beyond anecdotal evidence to quantify the "rich-get-richer" dynamic across major platforms.
Cross-Platform Comparison: By leveraging Patreon data as a proxy for external attention, the study offers a comparative framework to assess how different platform algorithms (e.g., YouTube's recommendation engine vs. Twitter's social graph) correlate with earnings concentration.
Temporal Dynamics: The study documents the evolution of earnings inequality over time (2018–2024), revealing trends in how algorithmic shifts impact the distribution of creator income.
Methodological Rigor: The authors explicitly address confounding variables, such as platform-to-Patreon conversion rates and audience composition, treating the observed patterns as consistent with algorithmic mechanisms while acknowledging alternative explanations like creator sorting.

Results

Power Law Dynamics: Across most platforms, Patreon earnings follow a power-law distribution with an exponent $\alpha \approx 2.0$ . This places creator earnings closer to concentrated capital income ( $\alpha \approx 1.5$ ) than to labor income, suggesting multiplicative, compounding effects driven by algorithmic attention.
Platform Variation: Significant variation exists in the degree of inequality:
- High Concentration (Low $\alpha$ ): YouTube ( $\alpha = 1.8$ ) and Instagram ( $\alpha = 1.84$ ) exhibit the most extreme "rich-get-richer" dynamics. These platforms show lower mean and median earnings and a weaker "middle class" of creators.
- Lower Concentration (High $\alpha$ ): Twitter/X ( $\alpha \approx 2.35$ ) and Patreon-only creators show higher exponents, indicating lighter tails and more equitable distributions. The authors note that Twitter/X's historical reliance on social graphs rather than pure algorithmic recommendation may contribute to this.
Temporal Trends: Between 2018, 2021, and 2024, $\alpha$ values generally declined for YouTube, Instagram, Facebook, and Twitch, indicating increasing earnings concentration. Instagram showed the most dramatic shift, with $\alpha$ dropping from $\approx 2.1$ in 2021 to $1.84$ in 2024, coinciding with a shift toward algorithmic recommendations.
Middle-Class Erosion: Platforms with stronger power laws (lower $\alpha$ ) correlate with lower median earnings and narrower interquartile ranges, suggesting that the "rich-get-richer" dynamic disproportionately extracts gains from mid-tier creators.
Content Categories: Earnings inequality varies by content type. Animation and Podcasts exhibit the lowest $\alpha$ (highest inequality), while Music, Comics, and Writing show higher $\alpha$ values, suggesting more equitable distributions in these categories.

Significance and Claims
The paper posits that algorithmic attention allocation is a plausible primary driver of the observed earnings concentration. The findings suggest that current platform algorithms, which prioritize existing popularity (low $\gamma$ ), create ecosystems where a small elite captures the majority of revenue, eroding the creator middle class.

The authors explicitly state that their findings are descriptive and suggestive rather than definitively causal. They acknowledge that alternative mechanisms—such as audience composition, creator self-selection, and platform-specific conversion norms—could contribute to the observed patterns. However, the consistency of the power-law dynamics across different platforms and time periods supports the hypothesis that algorithmic design plays a central role in shaping earnings inequality.

The study concludes that without changes to the underlying algorithmic mechanisms (specifically increasing the "exploration" component to surface long-tail content), earnings inequality is likely to persist. The authors suggest that policy interventions, such as mandating transparency in recommendation systems (e.g., via the EU Digital Services Act) or requiring algorithms that incentivize long-tail content, could foster more equitable digital economies. They caution that simply increasing market competition may not suffice if platforms remain locked into engagement-maximizing algorithms that reinforce the "rich-get-richer" dynamic.

More like this