While most rollup teams are targeting decentralized sequencers, there's a real use case to centralizing sequencing for maximum efficiency. (Nuances aside w/ provers etc.) Sequencers are powerless: they can only reorder transactions or withhold them. Stealing not possible.
But there's no incentive - if the rollup operator misbehaves they are just pushing people to competing solutions. That's why in the 18 months we have had rollups, there's not been a single case of misbehaviour AFAIK - even with billions at stake now. Be wary of new forks, though.
All competitive rollups will have an escape mechanism to the settlement layer, so you can't actually censor users from quitting your product and using a competitor. Just make it harder. So, there's no real incentive for an established rollup operator to misbehave.
The realistic attack vector is if a nation state subpoenas and shuts things down. But we'll see many regulated blockchain products where this is not an issue. In other cases, a federated sequencing model may suffice - i.e. appoint entities from all over the world to sequence.
The latter could also be semi-decentralized - perhaps a governance vote elects 20 sequencer operators or so.
To be clear: sequencers should absolutely be decentralized for credibly neutral rollups - just saying that there's also merits to not doing so in specific cases.
To be clear: sequencers should absolutely be decentralized for credibly neutral rollups - just saying that there's also merits to not doing so in specific cases.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
