Ethereum

not sure i'm looking at this correctly, but

would it make sense architecturally to have a hybrid between state channels and zk-rollups where only a subset of the txs are sent to main (i.e., for the state channel-like activity only send the opening/closing states, but store the rest on the L2)?

I feel like this gets close to fractal rollup designs, where you can have a "main" L2 and some validiums on top of it as L3s. Starkware introduced iirc it https://starkware.co/resource/fractal-scaling-from-l2-to-l3/, zkSync also has hyperchains

I feel like this gets close to fractal rollup designs, where you can have a "main" L2 and some validiums on top of it as L3s. Starkware introduced iirc it https://starkware.co/resource/fractal-scaling-from-l2-to-l3/, zkSync also has hyperchains https://blog.matter-labs.io/introduction-to-hyperchains-fdb33414ead7

Robust Incentives Group @ Ethereum Foundation

https://rig.ethereum.org

A validium isn't exactly a state channel, but maybe close to that in spirit

1/
thank you, these are all amazing leads! re: hyperchains in this context, i'll have to spend a bit more time to fully grasp them

indeed, the fractal scaling is similar to what i'm thinking of (although not for scaling but rather for compartmentalization purposes) with what StarkNet suggests as an L3 ...

3/3
... and secured on L1 with less verification load and bloat (arguable if the trade-off is meaningful)

2/ 
... to be integrated in the L2 (in the form of an onchain "state channel"), and produce validity proofs on L1 for a subset of the txs. this way the "state channel" activity is verifiable onchain (which seems to be an issue with validiums iiuc) ...