Tell me why I'm wrong: 

Ethereum L1 or L2 gas fees make micropayments too expensive for everyday consumer apps. 

Less frequent payments? Sure. But 10x a day? No.

i agree, would do it with an L3 plasma with alt-DA

@base contributor #001; onchain cities w/ OAK & city3

and fwiw i think that's totally ok and good.

blockspace is the next generation of compute. scaled apps will have their own compute (L3), smaller apps will use shared compute (L2), shared compute (L2) will serve as the hubs that connect apps, all of this will be secured by a maximally secure base layer (L1).

That’s an interesting take!

I feel these benefits of L3 you mentioned (cheap compute) are not a property of a “rollup-on-a-rollup”, but a property of application-specific rollups in general.

Like, couldn’t you get this cheap compute on an application-specific L2?

I like programming languages, rollups, and great storytelling. Building a permissionless fraud-proof system.

I see! Like this composability layer between other app-specific rollups.

We were playing with similar setups: L2 (app-specific) rollup as composability layer, but everything still settling to L1. It relied on cascades, a way to do instant one-way composability between rollups.

It feels the possibilities are endless.

you could, but new capital and users would need to bridge through the L1, which would be prohibitively expensive for most people. IMO this makes L2s the natural anchor point for app specific rollups.