From my non dev understanding, the ZK rollup (vitaliks favorite kind) takes the data, breaks it down to the "answer," and gives a proof which is very tiny, 16MB or something like that, because it just gives the answer without the question itself needed. That's why it's called zero-knowledge proof because you're trusting that they are giving you the right answer. Vitalik alls this "magic," so if it's magic to him, it might as well be some alien to me.
However, you're right; these do not promote decentralization directly. One could say that adding less data to the chain makes it eth more decentralized in terms of being able to run a full node easier etc. But the kicker is to do a zk proof is more comp intensive than just making the smart contract outright. So you'll have onramps with their own native tokens (sigh, more premines), and you'll need to pay them in said token to do that proof for you because no one is giving out free resources. It will still have a glass fee to post to the chain.
- This also includes trusting that this person is giving you correct proof, so there will need to be some form of degov involved in selecting the trusted validators. So you almost get a dpos scheme right there as you're ordinary eth goer isnt going to be doing zk rollups themselves.
Thank you, great explanation!
Exactly, that's my point, you still need another layer of consensus, so ZK rollups are not really that magical solution that has no compromise. The consensus is not as secure or decentralized as ETH itself, it has the same disadvantages as other Layer 2s