Developers in the ecash community say they have made progress against a risk that gets less attention than hacks of individual wallets: too many custody systems depending on the same code. At Ecash Hackday in Berlin, the Fedimint team ran a single federation on three separate software implementations. Separately, a developer named Calle presented Federated Cashu at the bitcoin++ event.
What was demonstrated
Fedimint is a protocol in which a group of operators, called a federation, jointly look after users' bitcoin. Until now, federations have relied on one shared codebase. That creates what the source calls a software monoculture: a single bug or exploit could affect many federations at once.
At the Berlin event, one federation operated across three independent implementations:
- The original Fedimint implementation, built by the Fedimint team
- An implementation based on Cashu, built by thesimplekid
- A third implementation, built by Eric Sirion
Federated Cashu
Calle's Federated Cashu uses a new blind BLS signature scheme. Under it, a federation can keep running as long as any four of its five operators are active. The design is also described as keeping user privacy data out of the operators' reach.
The five layers of independence
The source argues that custody should not be judged as simply trusted or trustless. A better question is how many independent failures would have to happen before funds are at risk. It lists five layers where diversity is needed:
| Layer | Goal |
|---|---|
| Protocol implementations | Several implementations of every protocol that holds funds |
| Wallet software | Several wallet implementations |
| Hardware | Keys generated on devices from different vendors |
| Human operators | Multiple independent, trustworthy people whose privacy is protected |
| Jurisdictions | Spread across different places and legal systems |
Why it matters
When you learn about crypto custody, it is easy to focus on whether a system is run by a company or by code. This story shows another angle: how many separate things can break at once. If every federation uses the same software, one flaw can hit them all. Using different codebases, hardware makers, people and countries means one failure is less likely to take down everything. The Berlin test and the 4-of-5 design are early examples of that approach, and the five layers give you a checklist for judging any custody setup.




