Nice model @Ohad-StarkWare ! Some questions about it:
- How the proving threshold would be defined and who would do it on the model? Because even if not completely clear to me, it seems that if its update would follow the same L2 consensus mechanism, maybe there could be some attack vector. My line of thought is that this threshold could potentially be “subjective”, so not even easy to determine if the entity suggesting a change on it is acting honestly.
- On the proposed second-time window, when the chosen sequencer didn’t publish the proof, would the “racing” sequencers be rewarded even if in the end they are not the ones submitting the proof? Thinking about their profitability model, if it would to a model similar to PoW.
- As you mention on StarkNet decentralization : Kicking off the discussion, one of the challenges is
The problem is that the rate at which transactions are collected is faster than proving those transactions. If I understand correctly this model, even if the time slots from sequencing until proof-submitted are short, both sequencing and proving need to be completed sequentially; so it is not that sequencers are ordering txs on one side, while provers are on another side. Does this have implications on the finality of the L2 txs? (which I think is quite important). - How is managed election of the next sequencer on L1? As I assume this process would involve submitting a regular transaction to L1, so could it be a problem to have too short time and in cases of network congestion (L1) affecting the election dynamics? (and with them the whole consensus of L2).