Abstract
Patreon allows content creators to monetize additional content from loyal fans. Because it offers minimal native distribution, Patreon earnings largely reflect creators' popularity on their primary external platforms (Instagram, Twitch, YouTube, Twitter/X, Facebook, or "Patreon-only"), making them a useful proxy for platform-level algorithmic attention dynamics. Fitting power-law tails to Patreon earnings by primary platform affiliation, we find three key results. First, platforms exhibit "rich-get-richer" earnings dynamics (Barab & aacute;si and Albert, 1999 Science, 286, 509-512), reflected in a Pareto exponent $\alpha \approx 2$, which is closer to concentrated capital income than labor income. Second, platforms with more concentrated earnings (lower $\alpha$) have lower mean and median earnings, and thus an eroded creator "middle class." Third, across most platforms, $\alpha$ values decline and converge over time (based on three cross-sections: 2018, 2021, and 2024), meaning earnings become increasingly concentrated among top creators-consistent with algorithmic recommendations rising in importance. Algorithmic attention allocation is a plausible driver of these patterns, though platform-specific conversion rates and audience willingness to pay may be just as important.