Continuing the journey through layer-two technologies, here is a summary of the paper “eltoo: A Simple Layer2 Protocol for Bitcoin” by Christian Decker et al (see also: a summary in the Blockstream blog).
Eltoo proposes a new construction for payment channels.
It is not a fully-fledged protocol, rather, it only describes one crucial building block – state revocation mechanism.
As you might remember from my summary of “SoK: Off the chain transactions”, the crucial challenge in L2 protocol design is old state invalidation.
Lightning uses replace by revocation (in the SoK paper terms) which works in practice but has its drawbacks.
The construction is rather complex, and the intermediate states held by the two parties are different.
This inherent asymmetry prevents easily extending the protocol to support multi-party channels.
Eltoo suggest another, symmetric state revocation mechanism, which is arguably better modulo one crucial limitation: it depends on a non-existent SIGHASH_NOINPUT signature flag.
The good news is, this change can be implemented relatively easily via a soft fork and doesn’t seem to be very contentions.
If that happens, it would be possible to replace state revocation mechanism in the live Lightning network to Eltoo while preserving all other aspects (channel synchronization vis HTLCs, routing algorithms, etc).