Multi-L1 support

Supporting heads targeting different L1s

Separation of custody and execution has another important property. Typically, if these things are conflated, the means of validating / executing transactions become intertangled with the details of how you prove that execution on the L1 in order to disburse funds. This makes it difficult to port that solution to another chain. If we were to try to use Arbitrum’s optimistic rollup on Cardano, for example, nearly every aspect of its design would need to be rethought, to the point of being an entirely new protocol (similar to Midgard taking heavy inspiration from Fuel). Every fraud proof, the data that gets published, the data availability layer, every bridge, deposits and withdrawals, and more.

With our separation, and the unifying abstraction of the “evacuation map”, porting Gummiworm to support multiple L1s becomes much easier. Each L1 simply needs the following primitives accounted for, and the existing infrastructure plugs in cleanly:

• The ability to hold funds segregated from user control (a multisig or a vault)
• The ability to validate some signature from the head and coil peers that ensures eventual deterministic execution of a chain of transactions which disburse these funds
• The ability to accept or reject deposits in a timely manner without risk that a rollback or user can revoke that ability
• The ability to fall back to a separate regime where funds are managed via a system of voting on newer snapshots
• The ability to publish some commitment to the evacuation map, and validate claims against that commitment to disburse funds

All of these are easily achievable on any chain with a minimal set of programmability. For evidence that this is a reasonable assumption, see the Blockchain specifics section, which details concrete implementations for Cardano and a feasibility sketch for Midnight.

There’s even some hope that with the right legal setup, a gummiworm head could interface with traditional banking infrastructure by issuing ACH transfers for its evacuation map.

Supporting heads targeting multiple L1s

In fact, we can go a step further. By leveraging the notion of compartments, Gummiworm heads can maintain multiple evacuation maps. The consensus rounds would produce each of the required signatures for the L1s it interfaced with. So long as the ledger has knowledge of each of the L1s that is being used, it can intermingle those funds, such as supporting a native ADA/ETH market. As each order executes, ADA moves from one owner to another in the ADA evacuation map at the same time that ETH moves in the opposite direction. As long as the coil is able to atomically produce signatures across both evacuation maps, the details of how to evacuate that custody can be handled by the L1 specific implementations of the vault on each chain.

This enables an incredibly rich design space and pulls the weft tight on an interconnected blockchain fabric.