Nuh on Nostr: Let me say this more clearly; 1. L1 needs to stay very small, mostly to act as a ...

Let me say this more clearly;
1. L1 needs to stay very small, mostly to act as a clock, and ossify hopefully after at least OP_CAT, or alternatively we are stuck with BitVM awfulness forever.
2. Sidechains are not for creating another large global state, instead they are for having expressivity to try every scaling approach we need.
3. Sidechains are ephemeral chains that need to be as long as the last pegout, because you can't reorg earlier than the last pegout.
4. Sidechains can get very big blocks without adding too much storage because #3
5. Sidechains can have very big blocks without bandwidth problems too much, because they can either split bandwidth to sub sidechains, or use Data Availability Sampling so every node or light client needs to download log(size) of each block.
6. Sidechains can have stateless clients where they don't maintain the state but it must be sent to them with blocks so they can validate it, and because Rootstock commit to the previous state in previous blocks that is more ergonomic than in Bitcoin.
7. Sidechains can have very small state and very low computational requirements, if they are used mostly for proof of publication, but 3rd party nodes hand the computation and the state for their own scope, something similar to what Ordinals are doing with Bitcoin where Bitcoiners who aren't interested in Ordinals don't care about its state or state transition... A more appropriate comparison is Client Side Validation like RGB or better ShieldedCSV where you don't even use UTXOs as nullifiers you just post opaque data that everyone can ignore except the ShieldCSV users, and each user maintains only their state... Or the state of an application they care about like Satoshi Dice for example, but only for as long as they care about playing Satoshi Dice then they can ignore it again.

I hope this makes sense to you and excites you about Sidechains.