A single statistical coefficient is enough to describe what is happening to the creator economy. The study published in Industrial and Corporate Change by Oxford Academic tracks Patreon revenues over several years and compares creators associated with multiple platforms, including Instagram. For creators affiliated with Instagram, a lower coefficient indicates greater concentration at the top of the distribution; it has tightened between 2021 and 2024, a sign that concentration is increasing. Platforms with the most concentrated distributions show a weaker middle class of creators.
The essentials
- Among 104,719 Patreon creators, revenues follow a power law close to Pareto: an infinitesimal minority captures the lion’s share.
- On Instagram, the concentration coefficient fell from approximately 2.1 in 2021 to 1.84 in 2024: the lower the coefficient, the tighter the distribution becomes at the top.
- The results are consistent with cumulative amplification linked to recommendations, without causal proof.
- Cultural diversity depends on the economic viability of a middle class of creators exposed to erosion by current dynamics.
- Rules for deliberate exposure to emerging creators exist: their adoption remains at this stage voluntary and marginal.
The actual distribution of revenues according to the coefficient
A Pareto coefficient equal to 2 means, in plain language, that doubling the number of creators at the top does not double the group’s revenues: the gaps between distribution levels are stable. When this coefficient drops toward 1.84, as on Instagram between 2021 and 2024, the gaps widen. Creators at the top earn proportionally more compared to those in the middle. Middle creators, for their part, move closer to the revenues of creators at the bottom.
Patreon revenues remain unequal, but their distribution is less concentrated than those associated with Instagram or YouTube in this study. The study published in Industrial and Corporate Change establishes that revenues there follow a power law close to 2: a handful of creators captures a disproportionate fraction of paid subscriptions. This figure applies to a platform where revenues come directly from subscribers, with no advertising intermediary. On Instagram, advertising is added, and trends are consistent with growing importance of algorithmic recommendations.
The evolution of the coefficient between 2021 and 2024 is precisely what the study allows us to track. Over these three years, concentration increased measurably. Behind the statistical movement lies a human reality: thousands of creators who produce regularly, build an audience, but see their revenues stagnate or decline while large accounts progress.
The algorithm rewards what has already succeeded
Content platforms do not distribute exposure by chance. They distribute it according to recommendation rules that favor content that has already generated engagement. A creator with a million followers sees their new posts pushed to new audiences. A creator with twenty thousand followers depends more on the loyalty of their existing audience and organic discovery.
This mechanism of cumulative feedback—the more visibility you have, the more you get—is documented in the literature on markets with network effects. The structure observed is consistent with a cumulative dynamic of the preferential attachment type. This is sometimes called a winner-take-most market, as opposed to winner-take-all: the gains do not go to a single actor, but to a very small number, with exponential gaps between layers.
The results are coherent with a possible role of algorithms in concentration, without proving it causally; the question concerns the speed of this process and its effects by revenue stratum. The observed evolution indicates increased concentration between the measured dates. A particularly unfavorable effect for the middle class is observed in the most concentrated distributions, among creators who live partially from their activity without having reached the threshold from which automatic recommendations take over. This phenomenon is linked to what the absence of public statistics on culture renders invisible: lacking independent data on revenue distribution by platform, regulators have only partial information.
The middle class of creators, first victim of cumulative dynamics
Speaking of a “middle class of creators” might seem metaphorical. The study gives it empirical content. Between the first tenth of the distribution, the least well-paid creators, and the top 10 percent, the gap is such that speaking of a continuous distribution becomes misleading. These are two distinct economies coexisting on the same platform.
Intermediate creators are those who generate enough revenue to justify serious activity, but not enough to live solely from creation. They invest time, equipment, sometimes collaborators. They produce a significant share of cultural diversity: thematic niches, experimental formats, content that does not seek immediate virality. The most concentrated dynamics are associated with a more fragile middle class of creators.
The period 2021-2024 coincides with increased reliance on recommendations by the platform. Meta reoriented Instagram’s feed from a chronological flow of subscriptions to an algorithmic flow of recommended content, including accounts the user does not follow. The objective was to increase time spent on the platform. The study interprets the trends as consistent with attention allocation more favorable to viral content. Algorithmic surveillance at work produces an analogous dynamic in other sectors: when an algorithm ranks and hierarchizes, those already well-ranked consolidate their position.
Patreon data on the subscription economy
Patreon offers a particularly clear case study because its data reflects a direct relationship between creators and audience. Subscribers pay a monthly amount in exchange for exclusive content or privileged access. Revenues depend on the loyalty of existing subscribers and the creator’s ability to attract new ones, not on an advertising algorithm or a platform recommendation system.
Yet even in this framework, the distribution follows a power law close to 2. The 104,719 creators studied form a very tapered pyramid. Creators at the top, those who have reached critical mass of subscribers, generate monthly revenues that represent very high multiples of the median. The majority of creators active on Patreon receive modest revenues, compatible with supplementary income but rarely with a primary activity.
This result on Patreon illuminates the Instagram analysis. Concentration exists also in Patreon data, without allowing us to isolate the effect of platform design choices. Gaps between platforms are consistent with a role for recommendation systems, among other possible explanations. The regulatory issue then concerns calibrating this amplification, concentration being inherent to any market where attention is the scarce resource.
The ability of recommendation rules to preserve diverse cultural supply
The Oxford Academic study identifies available levers without settling them. The study documents distributions consistent with cumulative dynamics and suggests a possible role for recommendations; the means of action remain to be specified.
A first avenue is technical: require platforms exceeding a certain audience threshold to reserve a measurable share of exposure to creators who have not yet reached a defined popularity threshold. This type of discovery obligation exists in other cultural sectors—quotas of recent titles in online music, exposure requirements for European works on video platforms. Its application to creation on social networks faces two difficulties. First, the definition of “emerging creator” is hard to stabilize regulatorily. Second, platforms can circumvent the obligation by reducing overall engagement rather than increasing exposure for intermediate creators.
A second avenue is informational: a right of public audit of recommendation algorithms entrusted to an independent authority, with annual publication of a report on the evolution of revenue concentration by platform. This approach forces platforms to make their design choices legible, without constraining them on content. Regulators like the European Commission, with the Digital Services Act adopted in 2022, have committed in this direction for the very largest platforms. Implementation remains partial in 2024, and published data do not yet allow tracking the evolution of creator revenue concentration with the precision the Oxford Academic study achieves.
Two scenarios sketch possible trajectories toward 2030. In the first, platforms adopt, under regulatory pressure or strategic calculation, exposure rules that slow cumulative dynamics. The decline in the concentration coefficient observed between 2021 and 2024 could then slow, preserving an intermediate stratum of creators. Cultural diversity would be maintained, at the price of less optimization of engagement time for platforms. In the second, algorithms continue to maximize engagement without discovery constraints.
Concentration increases, cultural creation reorganizes around two poles: large commercial accounts and amateur creators without economic ambition. The intermediate stratum, less well-equipped economically, empties for lack of viability.
The most useful signal to distinguish these trajectories is not the concentration coefficient itself, which measures a state, but the evolution of the share of total revenues captured by creators between the tenth and ninetieth percentile of the distribution. If this share stabilizes or rises, the middle class of creators resists. If it continues to fall, cumulative dynamics prevail.
Platforms’ interest in preserving the status quo
It would be inaccurate to present platforms as indifferent to the diversity of their supply. Their short-term economic interest is to optimize engagement, which favors popular content. But their medium-term interest is to maintain supply diverse enough to retain audiences with varied tastes. A platform showing only homogeneous mega-creators loses differentiation against its competitors.
TikTok built part of its competitive advantage on a promise of discovery: its algorithm deliberately exposes users to creators they don’t yet know. This promise attracted emerging creators in mass and produced rapid-growth trajectories that other platforms struggle to replicate. Meta responded by developing similar features on Instagram and investing in Reels, a short-form format where algorithmic recommendation takes priority over subscription. Between 2021 and 2024, measured data show increased concentration on Instagram, without allowing evaluation of the effectiveness of these adjustments on this trend.
Platforms know how to produce discovery when they choose to; their constraint is economic. Very popular content generates significant advertising revenues and mobilizes more screen time. The incentive to push it more responds to this structure. Without external rules that modify this incentive, platforms have no autonomous reason to arbitrate against concentration.
This is why publishing recommendation criteria and transparency on concentration effects appear as more robust levers than calls for self-regulation. An attention market where the rules of the game are known allows creators, regulators, and researchers to measure the real effects of algorithmic choices and adjust. An opaque market reproduces existing dynamics without friction.
The Oxford Academic study does not exhaust the subject. It opens a precise window on a dynamic that deserves annual tracking and comparison between platforms. Data declared by platforms under the European Digital Services Act could, if sufficiently granular, provide material for this tracking. This is an open project, for regulators, for researchers, and for creators who have an interest in the rules of the game being legible.
Sources
- Industrial and Corporate Change, Oxford Academic, “Platform attention and earnings inequality using Patreon earnings data”
- Journal d’un Progressiste, “Culture without public statistics leaves the field to platforms”
- Journal d’un Progressiste, “Algorithmic surveillance deepens digital hierarchy at work”
