This is indeed very interesting @sodazone. Building a shared sequencer on top of Polkadot has it’s own merit of being backed by billions of dollars in economic security. However, I feel this would also force the sequencer to follow Polkadot block times and other restrictions (if any) of the relay chain? For instance, Polkadot being an L1 would need to focus on liveliness whereas shared sequencers can opt for optimistic responsiveness as in the worst case users can withdraw from L1.
EDIT: Moreover, the shared sequencers also involve interactions with the L1 for reading and posting commitments. Is this possible by embedding the shared sequencer logic within the relay chain itself?