The Myth of the Billion-User Toy: Fiatjaf on Key Sharding, Dumbphones, and Scaling Illusions
As alternative networks either stall at pet-project scale or centralize to survive, protocol architect fiatjaf challenges key custody orthodoxy and calls for vibecoded hardware over endless software churn.
The recurring tragedy of decentralized protocols is not that they break under load, but that their creators rarely intend for them to reach load at all. In a sharp series of reflections across the network, protocol architect fiatjaf [s236] confronted the self-limiting delusions of the developer ecosystem: if these new inventions have such obvious flaws that make them impossible to scale to a billion users, why haven’t their own inventors realize it?
The answer, he argued, is an unspoken comfort with insularity. Now I think they are not really thinking about changing anything, they just want to see their pet project running, and they are happy if it runs with a few thousand users and they’re idolized by some of them.
While enthusiasts celebrate early adoption, foundational limitations are swept aside: someone invents some new thing that is sold as having the potential to “build a new internet”. And the thing actually works, wow! Everybody is excited, people start to put time into it, use it in their lives, tell others, convince businesses to accept payments using that because it is instant and free, you name it.
Yet without structural viability, the trap snaps shut: The most likely outcome is that the new technology never grows beyond a million users, so the obvious issues never materialize. The second most likely outcome is that the thing grows and slowly turns into the worst version of itself and most people don’t realize.
Nowhere is this tension sharper than in key management and user onboarding. Turning to the community’s insistence on raw private keys, fiatjaf [s367] dismantled the purist dogma: I think every person should have a phone with Amber installed and only use Nostr apps from that phone and the absolute requirement for ever thinking about using Nostr is that you have to first install Zapstore then Amber and promise you’ll never uninstall that, never lose your phone to someone that will steal your key, and that you’ll keep a paper or steel backup of your nsec at home at all times.
In response, he defended Pomegranate, a scheme where keys are split and coordinated via remote signers: You shard your key in many parts and give it to semi-trusted operators to sign stuff in your behalf.
Sharded signers, he insisted, allow ordinary participants to use decentralized applications across devices without treating every login as a high-stakes operational drill.
At the root of the problem lies physical hardware. Modern smartphones, he observed in fiatjaf [s331], are fundamentally hostile to intentional thought: It’s ridiculous how ineffective and unnerving phones are, they are optimized for browsing garbage and shutting down your brain.
Rather than building more ephemeral software layers on compromised platforms, he issued a challenge: When are we going to be able to vibecode a dumbphone with exactly the features we need? Vibecoding software is useless, give me vibecoded hardware, please.
