You, Hollywood & the Gaming Industry Skipped Italian Brainrot. 23 Million Players Didn’t.

Invisible To Adults, Unmissable To Teens: Italian Brainrot Went Platform-Native and has Made Millions. That Was My Point.

I get it

Sometimes you open a drop from me and think, what fresh hell is he on about now? Italian brainrot? Be serious. You came for cinema thinkpieces and I’ve handed you a cat-prawn in hotel slippers yelling “cappuccino”.

Awkward twist

That cat-prawn went platform-native and broke records: ~23–24m concurrent users inside Steal a Brain Rot, a Roblox platform-wide all-time high ~47.4m platform-wide, and ~500k on Fortnite Creative.

Whether you love it or loathe it, the cat-prawn is now a cash register. Audited totals aren’t public, but on standard Roblox economics a conservative back-of-the-envelope puts gross takings in the low seven figures for the record window (e.g., even 0.5–1.0% of peak CCU spending US$10–$20 across a week), with eight-figure upside if engagement holds. So, you know, it matters.

Confession time

My original Italian Brainrot essay (also below) was one of the worst-performing posts I’ve ever published. You all took one look at a shrimp in a tutu and punted me into the sea. And yet, here we are.

Instead of me gloating

(but I mean, kinda gloating), let me break it down for you and why it matters (and mattered then but no one cared. Fine.)

And if you haven’t yet, consider a full subscription to Strange Loop. So you don’t miss the next cat-prawn stupid thing from the internet that is making millions.

What I argued (and why it sounded insane at the time)

• Mechanism over message. Culture has shifted from “what does it mean?” to “how fast can I recognise it?” We live by feedback loops, not narrative arcs. Or, as I put it then: “We don’t experience culture, we loop through it … until absurdity becomes architecture.”

• AI is the accelerant. Generative tools don’t fail at realism; they succeed at intentional weirdness, producing what I called post-coherence aesthetics: content meant to be felt before it’s understood.

• Sound is syntax. TikTok’s audial turn recoded culture so that a noise, a timbre, a cadence is the payload. Recognition travels at the speed of a hook.

• Recognition > coherence (measure it). I proposed WTF-per-second (WPS) as a proxy for recognition velocity: the faster the micro-shocks, the better the engagement.

• Post-ownership folk culture → platform capture. The most potent artefacts are owned by no one and made by everyone; platforms harvest the surplus.

• Cohort split is structural. The format lands with under-25s and collapses with older audiences, a civic design problem masquerading as taste.

To put very simply

Culture now runs on speed and recognition, not meaning: we don’t follow stories, we ride loops. AI pours fuel on this by pumping out deliberate weirdness that hits first and only makes sense later. On TikTok and its clones, sound does the heavy lifting. One hook and you’re in. What wins is quick recognition shocks per second (call it WPS): the more little jolts, the more engagement. These memes are “owned” by everyone and no one, so platforms capture most of the money. And the split is generational: under-25s get it instantly; older audiences mostly don’t — that’s a design issue, not just taste.

At the time, this read like I’d inhaled too much shrimp. Then Roblox happened.

Meme Velocity by Platform (a.k.a. Why You Feel Like You’re Drowning in Shrimp)

This chart compares how fast memes evolve across platforms — and spoiler: TikTok is not playing. Based on a strange-loop-inspired adaptation of Douglas Hofstadter’s recursive systems theory, it tracks the average time it takes for a meme to go from emergence to self-referencing mutation. In 2015, it took ~12 hours for a meme on Facebook to start looping itself. On TikTok in 2025? It takes under 30 minutes — a 96% reduction in cycle time. Not only is TikTok feeding us memes, it’s digesting and remixing them before we’ve even finished laughing. You don’t scroll past the loop. You scroll into it.

What just happened (and why it matters)

In Steal a Brain Rot, you buy characters with Robux, then steal other players’ characters. This is simple Pokémon logic for an AI-born meme mythos. The cast: Ballerina Cappuccino, UNG Tung Soho, Nini Pini, all characters you have never heard of (I know. You’re over 25.) The punchline? A character called TTT Saur was pulled from the game during a licensing dispute. The game about stealing characters ran head-first into ownership. Peak 2025.

Now line that up with the theses I covered:

• The loop became the gameplay. This wasn’t IP being “adapted” for a game; the feedback loop itself is the core game mechanic: collect, recognise, repeat, escalate. That’s mechanism over message in the wild.

• AI → scale. The post-coherence aesthetic didn’t just vibe on TikTok; it cleared tens of millions concurrent in a natively social sandbox. The ceiling on weird is higher than your brand safety deck.

• Recognition beats lore. Players aren’t chasing canon; they’re chasing instant legibility; the jolt of seeing Ballerina Cappuccino sprint past your HUD. That’s WPS as design spec, not a metaphor.

• Platforms capture value. The TTT Saur wrinkle is the collision I flagged: post-ownership folk IP smashing into legacy rights while the platform sits between the crowd and the creator, skimming the energy.

• Cohort truth. This phenomenon is “essentially invisible to people over 18” — but enormous for teens. That’s a format divide with downstream consequences for politics, markets, and culture.

To put very simply (again)

The meme didn’t get adapted into a game, it became the game: collect, recognise, repeat, escalate. AI then blew it up, taking that “post-coherence” look from TikTok to tens of millions playing together in a social sandbox. Players aren’t chasing lore; they want instant legibility, the jolt of seeing Ballerina Cappuccino flash past. So WPS (little recognition shocks per second) is effectively the design spec. When folk, ownerless IP hits old-school rights (see the TTT Saur takedown), the platform sits in the middle and skims the value. And the split is stark: largely invisible to most over-18s, enormous for teens, a format divide with real spillover into politics, markets, and culture.

Why I wrote it anyway (despite the flop)

Because Italian Brainrot is not an outlier; it’s the singularity where meme history folds back on itself until nonsense becomes the most legible thing on your screen. “You’re not seeing a chart. You’re seeing the shape of meaning collapsing and becoming culture anyway.” That’s the line I buried in a piece the algorithm ignored and it’s the only way to understand why a shrimp-adjacent universe is now a blockbuster game.

I’ll end on the line I meant as a joke but actually meant: “This is not a joke about memes. It’s a taxonomy of a distributed, AI-accelerated, anti-narrative mythos… They aren’t characters. They’re signals.”

If they make a movie, it’ll likely bomb, and yes, I will absolutely watch it. Because in 2025, the funniest thing on the internet keeps turning into infrastructure. And infrastructure always wins, even when your Substack doesn’t.

Operator notes (for adults who still have jobs and need to understand what to do about this)