Crime by Design: RPC Centralization in Decentralized Networks
Sandhya Saravanan
Session summary
The presenter argues that, despite blockchain’s decentralization goals, many users now rely on centralized mediators such as RPC providers, wallets, bridges, and exchanges. The authors claim that leading RPC providers have visibility into network-layer data and sign‑up details, making their position comparable to traditional financial intermediaries. They link this visibility to risks seen in real‑world violent attacks on crypto holders, emphasizing that RPC insiders could, in principle, access both user identity and balance information.
The talk surveys 20 RPC providers and reports that nearly all collect personal details such as email, billing information, and addresses by default. The presenter states that centralized RPCs can restrict access by country, potentially forge wallet balances because responses are not verifiable, and front‑run users by observing transaction simulations. To reason about defenses, the authors propose a leakage framework that separates network and application‑layer leakage, and distinguishes reads from writes. They map existing tools like encrypted mempools, Tor, PIR, and stealth addresses onto this framework, arguing that multiple techniques must be composed to approximate “private RPC” and that the framework can extend to other blockchain mediators.
Key takeaways
- The authors argue that major RPC providers occupy a bank‑like position by observing network traffic and user sign‑up data, giving them broad insight into user activity.
- A survey of 20 RPC providers reportedly found only one that does not, by default, collect email, billing information, and similar identifiers, raising privacy and censorship concerns.
- The proposed leakage framework decomposes network and application‑layer leakage for reads and writes, and is used to show that multiple privacy tools must be combined to significantly reduce RPC‑level leakage.