Everyone has a convergence story for OnlyFans. This report runs the branches instead: five claims screened against sources, two natural experiments read for magnitude and latency, one condition table pre-registered. The verdict vocabulary is structure or scaffolding.
| Claim in the draft | Status | What the sources say |
|---|---|---|
| Creators grew "7.5M in 2019 to 85M by 2022" | Fabricated | 348K (2019) → 1.6M (2020) → 2.16M (2021) → 3.18M (2022). Off by ~25×. SignHouse |
| US savings rate hit 33% in April 2020 | Correct | BEA record, up from 12.7% in March; previous record 17.3% (1975). CNBC |
| Women lost "1.4M more jobs than men" | Conflated | Two real stats spliced. Women were 55% of April 2020's 20.5M job losses; the 1.4M was the education-and-health sector count. NBC · BLS |
| Aug 2021 ban came from Mastercard/Visa pressure | Misattributed | The CEO named banks: JPMorgan, BNY Mellon, Metro. Mastercard publicly denied. The Aug 17 → Aug 25 timeline is accurate. CNN |
| OnlyFans pioneered the creator economy | Misattributed | Patreon proved creator-direct subscription in 2013, three years before OnlyFans launched. Contrary Research |
The claim under test is an only-way claim: that OnlyFans, or something exactly like it, had to happen. Three systems are said to have converged and made it so. A movement that installed the vocabulary of bodily sovereignty. A pandemic that concentrated savings and displaced women from service work. Payment infrastructure that had quietly matured. Remove any one, the story says, and the explosion never comes.
That story is unfalsifiable as written, and this trade is chronically suspected of telling exactly that kind of story. So the claim is restated as something attackable: for each proposed cause, is it structure — remove it and the thing stops — or scaffolding, loud while the thing went up, gone when you look back? The instrument that decides is the DeStrange, and its standing question is: how many branches did you run?
The comparison set: OnlyFans, Patreon, Substack, Twitch, Cameo, Fansly, JustForFans, ManyVids, Chaturbate, MyFreeCams, Pornhub's model program, Tumblr, Gumroad, Ko-fi, and AVN Stars — which folded in 2022 and is the most valuable row in the table, because a necessity test with no failures in it is worthless. Case list and coding rules are committed with a timestamp before any outcome is coded; that one paragraph of pre-registration is the cheapest possible cure for a story assembled around its ending.
You cannot switch off a platform's banking to see what happens. But reality ran that experiment twice, and the responses have a magnitude and a latency.
| Case | A · adult ok | B · rails | C · audience | D · shock | E · vocabulary | F · rails held | Y · breakout |
|---|---|---|---|---|---|---|---|
| OnlyFans | ● | ● | ● | ● | ● | ● | ● |
| Patreon | ○ | ● | ● | ● | ? | ● | ? |
| Substack | ○ | ● | ● | ● | ○ | ● | ? |
| Chaturbate | ● | ● | ○ | ○ | ○ | ● | ? |
| Fansly | ● | ● | ? | ● | ● | ● | ? |
| Tumblr (post-2018) | ○ | ● | ● | ○ | ○ | ○ | ○ |
| Pornhub models (post-2020) | ● | ○ | ● | ● | ● | ○ | ○ |
| AVN Stars (folded 2022) | ● | ● | ○ | ● | ● | ? | ○ |
The opening hypothesis does not survive its own test. Three co-equal systems goes in; one structure and two scaffolds come out. Remove the vocabulary and the cam economy is already there, profitable, six years early — MeToo downgrades from cause to legitimacy accelerant for the mainstream wave. Remove the pandemic and the curve keeps its slope and loses its spike. Remove the rails and the thing stops: minus 78 percent in a day next door, eight days from policy to surrender at home.
Scaffolding fools retrodiction because it is loud while the building goes up, and gone by the time you look back.And the pioneer claim inverts. OnlyFans did not invent the creator economy; Patreon had proved the mechanism in 2013. What OnlyFans did was apply a solved mechanism to the one category every incumbent's terms of service had fenced off. The inevitability, such as it was, belonged to the gap — a proven model, a excluded population, and the only question being who would be willing.
The load-bearing verdict rides on one eight-day crisis that was reversed: a near-death, observed once, is inference, not proof. The creator counts are aggregator-sourced until the Companies House triangulation lands. The trend-break projection is illustrative until the pre-2020 series is harvested. And a condition table with ? cells has, by definition, not yet been run — the falsification sentence above is live, and it may fire.
| Pattern | Verdict | Where it stands |
|---|---|---|
| Named instrument | Pass | The DeStrange, run 001 — named, repeatable, standing question stated, specimen numbering implies the cadence. |
| Coinage | Pass | Structure vs scaffolding — the verdict vocabulary travels without the site. Collision-checked (jenga: taken; load-bearing: becoming LLM-speak; DeStrange: clear). |
| Linked deliverables | Partial | Sources linked where they stand, but the repo — data, coding sheet, script — publishes with the first harvest. Until it exists, "check me" is a promise. |
| Declared lens | Pass | The deck declares it — runs the branches instead — and six chapters demonstrate it. The author's personal lens belongs to the essay this report backs. |
| Open door | Partial | The tally widget orients a cold reader in ten seconds, but the plain-text method note that doubles as a start-here page is still owed. One page, no design, hardest-working thing on the site. |
The sign says: your body, your terms, your business — the studio, the agency, and the algorithm all fired at once. The myth sold to participants and audience alike is disintermediation as empowerment.
The rule set producing the sign is not feminism or lockdown boredom; it is which work a bank will process. Every incumbent's ToS fenced off the category; the code stays invisible because processors prefer it that way.
Sign promise: ownership without gatekeepers. System delivery: access contingent on banking tolerance, decided in eight days in Aug 2021, at 20% rent. Gap reframe: OnlyFans is not replacing the gatekeepers. It is replacing the studio's cut with the processor's veto.
Cam-adjacent professionals (the pre-2016 base) → service and gig workers displaced by the she-cession → influencers with importable Instagram/TikTok audiences (the Spring 2020 social-proof wave) → mainstream celebrity stunts, 2021.
Primary: lockdowns concentrated spendable money (savings rate 33%, April 2020, a BEA record) while displacing the female-majority service workforce (women were 55% of April's 20.5M job losses). Secondary: lockdown privatized consumption — the social-surveillance cost of subscribing collapsed. Tertiary: the rails were already frictionless.
Complex contagion on the creator side — joining required multiple reinforcing exposures and social proof (Centola 2010) — with winner-take-all outcome distribution on the earnings side (Salganik, Dodds & Watts 2006). Identity split: informational adoption (income) and identity-performative adoption (the destigmatized-entrepreneur persona; Van Bavel & Packer).
Network: hub amplification via celebrity joins, winner-take-all topology. Contagion: complex, bridge-width dependent — each join widened the bridge for second-degree ties. Statistical physics: homophily bred creator subcultures that later seeded the counter-signal. Micro-motives: a Schelling cascade — mild individual decisions, non-linear collective outcome, tipping ~Spring 2020. Generative test: the micro-rules (direct pay, subscription, 20% cut, no feed) grow the macro outcome; the near-ban shows the rules' emergent behavior surprising the platform itself.
Within-frame: creator disavowal and migration toward owned infrastructure — private Discords, direct subscriptions — after the 2021 rent-and-rails scare. Exit: the platform-labor reframe: union talk, gig-work analogies, rights language reconstructing the employment protections the empowerment frame skipped. Exit is the stronger bet, and it is the one the forward run measures.
| Cam | 2018 | Platform dependency and performer identity, two years before the mainstream conversation. |
| Normal People | 2018 | Female desire narrated from the inside — the literary parallel to what the platform commercialized. |
| Hustlers | 2019 | Monetizing male desire as financial reclamation, released the year before breakout. |
| WAP | 2020 | August 2020, peak growth: explicit female sexual agency as mainstream entertainment. |
| I May Destroy You | 2020 | Bodily sovereignty and consent, aired simultaneously with the platform's peak growth. |
Strategist: position against rails-risk, not on empowerment copy — the gap is where the durable story is. Creator: the exit bet beats the within-frame bet; own the rail relationship or own nothing. Researcher: the leading indicator is processor policy language and the L:E framing ratio — both move before the metrics do.
QCA — Qualitative Comparative Analysis; codes cases as condition sets to test necessity/sufficiency in small samples. Consistency — how reliably a condition accompanies the outcome (≥0.90 ≈ necessary). Coverage — how much of the outcome the condition explains; guards against trivial necessity. Natural experiment — reality removes a variable so you don't have to. Interrupted time series — fit the pre-shock trend, project, subtract. Complex contagion — adoption requiring multiple reinforcing exposures. Schelling cascade — mild local preferences, extreme collective outcome. Structure / scaffolding — remove it and the thing stops, versus remove it and the thing merely slows. Pre-registration — freezing rules before outcomes are looked at. Sign / code / gap — what it signals, the rule set producing it, and the distance between promise and delivery.