I just had a great conversation about the future of decentralised rollups and it was worth it to give a first hand perspective.

Below is the Substrate high level architecture chart. This is also a good mental framework of all components necessary to make a rollup

Thread 🧵👇  https://i.imgur.com/Nyd6s5a.png

Head of Product/Market Strategy @StarkWare

Another dimension is the consensus module. This is where we talk about the proof system being used: optimistic vs Validity, the protocol: Round fraud or S{T,N}ARK.
Most people start losing track here.
Because this discussion is missing out on the need for consensus separation between Tip of the chain and L1 settlement.

In the next few days, I will address each modules and explain how StarkNet is looking at it, giving a better glance of our next phases.

Next chapter
- A deep dive in how consensus will work on StarkNet as well as the separation sequencer/prover network

When people compared rollups today, they mostly focus on the execution engine (For Substrate, WASM runtime). In the current landscape, we have Nitro, OVM, StarkNetOS, ZKEVM, and so...
Those refer how the txs are being run and generally, gives the name to the said project. But it is not all one must consider.