Rollups are not decentralized yet: sequencer status and shared sequencers
Thread616 words47 likes
Reviews sequencer and validator centralization at Optimism, Arbitrum, zkSync Era and Polygon zkEVM and failure modes. Presents shared sequencers as decentralization-as-a-service for rollups.
Key claims
Optimism fraud proofs were still under development with sequencer and validator run by the foundation
Arbitrum had 13 whitelisted validators
If Polygon zkEVM's sequencer fails user funds are frozen, but any user can submit a ZKP if the aggregator fails
0/ only few are aware that rollups aren't real (yet)
a 馃У on the centralization issues rollups face, the current state of rollup sequencers & the role of shared sequencers in taking the training wheels off rollups (0/19)
1/ as rollups scale and grow, one risk attracting significant attention is the potential failure of sequencers and validators. And rightly so
2/ @optimismFND, the 6th largest network by TVL, could see users' funds locked if a whitelisted validator goes offline and @zkSync Era currently lacks a mechanism to deal with offline sequencers and validators
3/ but the main problem is, that most rollup networks run a centralized sequencer. But let's have a look at the current state of the leading rollups 馃憞
4/ @optimismFND
in Optimism, both sequencer and validator are centralized and run by the foundation. If the sequencer goes offline, users can still force transactions to the #L2 network via L1
5/ what not many are aware of either, is that #Optimism fraud proof system is still under development. So users need to trust the block proposer to submit correct L1 state roots
6/ additionally, an offline validator in Optimism's case actually means that users can't withdraw from L2 to L1. However, #Optimism plans to decentralize the sequencer in the future using economic game theory and governance mechanisms
7/ @arbitrum
#Arbitrum operates a centralized sequencer, run by the Foundation, with 13 whitelisted validators. Similar to #Optimism, users can still force transactions on #L2 through L1 if the sequencer goes offline
8/ unlike #Optimism though, Arbitrum has a fraud-proof system in place, albeit only accessible to whitelisted validators. #Arbitrum also has plans for future sequencer decentralization (and permissionless fraud proving)
9/ @zkSync #Era
in #zkSync Era, a centralized operator serves as both the sequencer and validator. Given the project's early stage, it lacks a contingency plan for operator issues. Future plans include decentralizing the operators by establishing roles of Validators & Guardians
10/ @0xPolygon #zkEVM
#Polygon #zkEVM features a centralized sequencer operated by the Foundation. The Validator (or Aggregator) that submits ZKP is also centralized
11/ if the sequencer fails, user funds are frozen; however, if the aggregator fails, any user can still submit a ZKP to withdraw from #L2 to L1. Polygon #zkEVM plans to decentralize the sequencer and aggregator using a PoE consensus algorithm
12/ so centralized sequencers pose a risk to rollup users as they have the power to censor transactions or maliciously exploit #MEV and they pose a risk as they are a single-point-of-failure
13/ given the risks inherent to centralized sequencers and validators, various projects like @AstriaOrg, @EspressoSys, @radius_xyz, OP Stack, and Flashbots' #SUAVE are striving to decentralize rollup network sequencing
14/ they provide rollups with a sequencer network, determining the transaction order and passing them to the rollup network
15/ shared sequencers offer several benefits over the current design:
馃敼 Censorship resistance
馃敼 Atomic cross-rollup composability
馃敼 Liveness (low/no downtime) guarantees
馃敼 Plug-and-play solution for existing & new rollups
16/ so, instead of the rollup bootstrapping new hardware for an entirely new set of sequencers, shared sequencer networks basically provide "Decentralization as a Service"
17/ shared sequencer projects have found a cozy little niche by relying on teams to be lax about "low-priority" project milestones
18/ recently, Ethereum Foundation researcher Justin Drake proposed a different approach: Based Rollups
I'll drop a thread on this over the weekend, so stay tuned 馃憖
19/ sequencer decentalization will definitely be a key issue for rollups to tackle as they enter the next stage of maturity 馃殌
some modular & rollup chads to follow馃 馃憞
@splinter0n
@Only1temmy
@Slappjakke
@Louround_
@thesaint_
@0xSalazar
@CryptoShiro_
@jake_pahor
@0xTindorr
@poopmandefi
@ShivanshuMadan
@0xShivani
@hmalviya9
@wacy_time1
@hooeem
@poopmandefi
@0xDinoEggs
@jinglingcookies
@DOLAK1NG
2 images on X