Just a sequence number for each rollup block will work. The nodes can copy from local chain and post it to the remote at any time since the sequencer already signed it locally.
Sequencer would have signed conflicting blocks or roots.
If the data or roots on any of the chains diverge, that’s a fault and a local implementation takes over the last valid root and everyone exits.
Sequencer X posts all the blocks on N chains. If they don’t include the expected txs, there is a local fault, and the local implementation takes over the root and users can exit. If roots diverge, that’s a fault, etc…
It would mirror on all N, each can post txs that must be included. If it ever diverges, there is an attributable fault everyone exits from the last valid root.
Why does it need 1 layer instead of N? When there is a fault, each layer only cares about its own tokens, all the remote are worthless.
It could be used to swap eth to sol, stuff like that.
Why not a really dead simple L2 that only runs an amm but it’s an L2 in multiple chains. If it’s sequencer fails to include a tx that’s locally posted, a local implementation of the amm takes over the root in exit mode and local coins can exit.
Issue a da challenge to fund out