Why I built Twerkhub.

People ask me this every week. Sometimes a member in the Discord. Sometimes a creator we're courting. Sometimes a journalist doing a piece on creator platforms. The short answer is the one we put on the home page — "if you know, you know." The long answer is below. Eight years compressed into eleven hundred words.

2018 · the boredom that started it.

Alexia and I were in a department in Palermo, Buenos Aires. She had just come back from a residency in Taipei. I had just quit a frontend job. We were spending three or four hours a night going through YouTube — looking for the dancers we cared about. The good ones. The studio cuts from iDance Taipei. The MDC NRG heels work. The street fancams from Caracas before they got buried under a hundred tutorials.

The problem wasn't lack of content. The problem was that YouTube's algorithm couldn't show us what we wanted. Not even close. We'd find one good clip, click "see more from this channel," and get fed five algorithmic guesses that were nothing alike. The dancers we loved were posting weekly — and we could never find them on the second visit.

So we started a private playlist. Just for us. Two people, one document, manually curated. By March it had 80 videos. By June, 240. By the end of 2018, we'd shared the link with about 30 friends and the playlist had 600 videos in it.

The platform existed before we knew we were building one. It started as a list to fix our own problem.

2019-2021 · the years where it was just a folder.

For three years there was no business. There was a YouTube playlist, a Telegram chat for the people who used it, and a notebook where Alexia kept a list of dancers she wanted to track. We weren't trying to monetize. The friends-of-friends list grew to about 240. We added Russian dancers (because she had Lada's account and started cross-pollinating). We added Korean studio work. The 7-playlist split started here — different vibes deserved different lists.

What the three years gave us: tradecraft. We learned which thumbnails meant good fancam work, which channels were uploading reuploads vs originals, how to spot a studio-translation cut from a club fancam in the first 4 seconds. That tradecraft is the moat now. You can't shortcut it. You either spent the years or you didn't.

2022 · the moment it had to become a thing.

Two things happened in 2022. One: YouTube tightened community guidelines and started shadow-banning a lot of the dancers we cared about — work that was perfectly legal, just outside the algorithm's comfort. We started losing access to videos we'd indexed. Two: the Telegram had grown to 1,400 people and was unmanageable. People wanted access, we couldn't filter, we couldn't keep the energy of the early small group.

We had to decide: shut it down, or build it properly. We chose properly. The first paid tier launched in October 2022 — $5/mo, mostly to filter for people who actually cared. By December we had 380 paying members.

2023 · what we got wrong first.

The first version of the platform looked like every other creator tool. Algorithm feed. Recommendations sidebar. "Trending now." Notifications every time anyone uploaded. Within four months retention was below 40% at the 90-day mark. The platform was teaching members the same behavior YouTube had — scroll, scroll, dopamine hit, leave.

I tore it out. We spent three weeks in early 2023 stripping every algorithmic surface. No For-You. No infinite scroll. No personalized recommendations. Just seven curated playlists, updated weekly, manually. People were furious for about two weeks. Then retention shot up. By Q3 2023 we were at 71% 90-day retention — roughly twice the creator-platform median.

This is the lesson I keep coming back to: members don't want what they think they want. They want what makes them feel like they have taste. Algorithms make people feel processed. Curation makes people feel chosen.

2024 · the token economy.

By 2024 we had 1,200 members and a problem. Subscription churn was healthy but we wanted longer engagement loops. Pure subscription = members log in, watch what's new, log out. We wanted them to build something on the platform.

The 4-tier token economy launched in March 2024. Basic, Medium, Premium, VIP Top. Tokens earned through engagement (page exploration, full-clip watches, daily logins, streaks). Tokens spent unlocking deeper archive access. The full math is in the token economy playbook.

Two unexpected outcomes. (1) About 40% of VIP Top members got there through earned tokens, not direct purchase. They built their access. They cared more about it. (2) The streak mechanic created accidental community — members started checking in daily, posting their streak counts in the lounge, helping each other not break.

2025 · scaling without breaking the philosophy.

The hardest year. We grew from 1,200 to about 1,900 members. Three things almost broke us:

• Curator overload. Just Alexia and one assistant, watching 40+ hours of YouTube weekly to curate the 7 playlists. We added two paid curators in mid-2025. They live in Taipei and Seoul, give us geographic depth we couldn't have.
• Creator pitch volume. 80+ pitches a month. We had no system. Built one — full process is in how to pitch Twerkhub as a creator.
• Discord drama. A community of 1,500+ generates internal politics. Two big incidents that year forced us to rewrite moderation philosophy. The full structure is in the Discord playbook.

2026 · where we are now.

2,178 members. 38 verified creators across Taipei, Moscow, Seoul, Medellín, Buenos Aires, Bangkok. 7 playlists, 1,500+ archived videos. Thursday drops at 00:00 ART. The brand voice is sharper. The platform is faster than it's been in three years.

I'm writing this from the same Palermo apartment. Different desk, same window. Alexia is in Mexico City this week filming a residency with a creator we've been courting for six months. The Discord is buzzing because tomorrow's drop has the first BEFOX studio cut we've ever indexed.

What the eight years taught me.

Curation is a real job. Not in the loose sense everyone uses now. Real curation means watching 10 hours to surface 30 minutes. The math doesn't work as a side project. We pay our curators because anything less makes the playlist worse.

Specific beats broad. We never tried to be a general dance platform. We picked a niche — twerk, fancam, heels choreo, the studios behind those — and went deep. Our 38 creators are not the most famous dancers. They're the ones we care about. Members who care about them stay.

The algorithm makes you forgettable. Every algorithmic platform converges on the same content because the same engagement signals apply. The only way to build a community with identity is to refuse the convergence. That's expensive. It's also irreplaceable.

Slow infrastructure outlasts fast trends. We've watched four "next OnlyFans" platforms launch and shut down since 2022. They optimized for growth velocity. We optimized for being around in five years. Different game.

What I tell people who want to build something similar.

1. Start with your own boredom. The thing you can't find on existing platforms is your unfair advantage.
2. Don't monetize for at least 18 months. Build the thing first. Find the people who want it. Money comes after the answer is yes.
3. Pick the niche you'd defend in a fight. If you wouldn't argue for your niche at a dinner party, you'll quit the first hard year.
4. Manual is a feature, not a bug. Algorithms are commodity. Taste is rare.

Tag line, in case you missed it.

If you know, you know. We don't pitch you. We don't optimize for the average. We don't show you what an algorithm thinks you'll click. We show you what we'd show our closest friend who asked "what's the good stuff this week."

That's the whole platform. That's why I built it.

—Anti · firestarter · April 2026