In this research article i show that consensus proofs when naively used by on-chain consensus clients leaves them vulnerable to byzatine attacks that are cheap to perform and carry no risks. However by introducing fraud proofs to these consensus clients they can become resilient to byzantine behaviour, enabling the first of it’s kind byzantine fault tolerant cross chain bridges.
This research can serve as the foundation for any kind of cross-consensus bridge including the DOT <> KSM bridge.
An optimistic approach makes sense here, broadly. Although once you have the ability to interpret consensus proofs between chains, an optimistic approach is useful as a time/space optimization, just to avoid eagerly verifying signatures.
I’m wondering about potentially exploring this approach for Ethereum bridging, i.e. since its light client protocol isn’t particularly secure (ref: Altair is no Light Client - by James Prestwich). It might make sense to just reimplement the entirety of CasperFFG on a parachain and replay all necessary attestations to interpret finality.
This actually crosses into the territory of my next article on how parachains can scale secure interoperability to all chains.