>To some extent, the Ethereum project has just given up on scaling the main blockchain. “For Ethereum to scale and keep up with demand, it has required rollups” — do the work somewhere else and send back the result. The blockchain is only usable if you work around actually using it.
Zk-rollups are fundamentally superior to direct scaling because they move execution and proof generation cost to only one node (entity, could be a server farm) that only has to do it once, while ensuring correctness is verifiable by anyone cheaply forever.
No more scalable solution exists from a purely theoretical level. This in theory allows even running complicated multiplayer games on a blockchain (turn based directly, realtime by dropping state snapshots).
In principle, this should be implemented directly - but it's so early that a solution that can last for decades simply doesn't exist. In addition, decentralizing zkrollup implementations to third party teams allows for much faster innovation.
Zk-rollups are fundamentally superior to direct scaling because they move execution and proof generation cost to only one node (entity, could be a server farm) that only has to do it once, while ensuring correctness is verifiable by anyone cheaply forever. No more scalable solution exists from a purely theoretical level. This in theory allows even running complicated multiplayer games on a blockchain (turn based directly, realtime by dropping state snapshots).
In principle, this should be implemented directly - but it's so early that a solution that can last for decades simply doesn't exist. In addition, decentralizing zkrollup implementations to third party teams allows for much faster innovation.
Optimistic rollups are more questionable.