Permissionless entry is a mirage. While anyone can join a set, the economic and technical barriers to producing blocks are prohibitive, creating a de facto oligopoly.
the-modular-blockchain-thesis-explained
Blog
The Hidden Centralization of Permissionless Sequencer Sets
The modular blockchain thesis promises decentralization through open sequencer sets. This analysis reveals how economic and hardware forces inevitably concentrate power, recreating the validator centralization problems of monolithic L1s within the rollup stack.
introduction
THE ILLUSION
Introduction
Permissionless sequencer sets, a popular scaling solution, create a false sense of decentralization that masks significant centralization risks.
Sequencer sets centralize stake. The mechanism incentivizes stake pooling into a few large operators, mirroring the Lido dominance problem seen in Ethereum's Proof-of-Stake.
Real-world evidence is stark. On Arbitrum Nova, a permissionless set, a single entity, Offchain Labs, controls the sequencer role, processing all transactions before they reach the L2.
key-trends
THE HIDDEN CENTRALIZATION OF PERMISSIONLESS SEQUENCER SETS
The Inevitable Centralization Forces
Decentralized sequencer sets are a popular narrative, but economic and technical forces push them towards de facto centralization.
01
The Capital Efficiency Trap
To be economically viable, sequencers must stake significant capital, creating a high barrier to entry. This favors large, established players like Jump Crypto or Wintermute who can deploy capital at scale, crowding out smaller, permissionless participants.\n- Requirement: Multi-million dollar bond for slashing security\n- Result: Oligopoly of well-funded entities
$10M+
Min. Stake
<10
Viable Players
02
The MEV Cartel Formation
Sequencers with order flow access form natural cartels to maximize extractable value (MEV). Projects like Flashbots SUAVE aim to democratize this, but in practice, centralized bundlers like those in Ethereum's PBS or Solana's Jito dominate.\n- Dynamic: Coordinated ordering for maximal profit\n- Outcome: Permissionless set acts as a permissioned cartel
90%+
OF Dominance
$1B+
Annual MEV
03
The Infrastructure Moats
Low-latency performance requires specialized hardware and direct network peering, creating insurmountable moats. This is why Coinbase's Base and Polygon's AggLayer run centralized sequencers—decentralization adds ~500ms of latency, a non-starter for apps.\n- Requirement: Colocation, custom firmware, proprietary relays\n- Result: Only AWS/GCP giants can compete
<100ms
Latency Req.
~500ms
Decentralization Tax
04
The Governance Capture
Protocol governance (e.g., Optimism's Token House) often votes on sequencer parameters and upgrades. Large token holders and delegators, typically VCs and foundations, inevitably steer decisions to favor incumbent sequencers, cementing their position.\n- Mechanism: Proposal voting on slashing, rewards, and set size\n- Outcome: Formal decentralization with centralized control
>60%
VC Voting Power
1-2
Upgrade Vetoes
05
The Liquidity Begets Liquidity Loop
The most profitable sequencer attracts the most stake and order flow, creating a winner-take-most dynamic. This is visible in Cosmos app-chains and Avalanche subnets, where a single provider often runs all infrastructure.\n- Dynamic: Positive feedback loop of fees and stake\n- Result: Centralization as a stable equilibrium
80/20
Revenue Split
10x
Fee Multiplier
06
The Regulatory Safe Harbor
Operating a sequencer in a regulated jurisdiction with clear compliance (like Coinbase or Kraken) provides a defensible moat. Permissionless anonymous operators cannot offer this, pushing institutional demand towards a few compliant, centralized sequencers.\n- Driver: Institutional demand for KYC/AML rails\n- Result: Compliance becomes a centralizing force
$50B+
Institutional TVL
3-5
Compliant Providers
deep-dive
THE INCENTIVE VECTOR
The Staking Pool Replication Engine
Permissionless sequencer sets centralize through economic incentives, not technical barriers.
Permissionless sequencer sets are a mirage. The economic design of staking pools creates a replication engine where a few dominant providers capture the entire set. This happens because node operators maximize yield by running the same software for every chain, leading to stake-weighted centralization.
Staking providers like Lido and Figment are the replication vectors. Their business model is to deploy identical, optimized node clients across every new L2. This creates a single point of failure across ecosystems, as the same few entities validate transactions on Arbitrum, Optimism, and Base.
The counter-intuitive flaw is that permissionless entry does not create diversity. A new solo operator cannot compete with the capital efficiency and tooling of a professional staking pool. The result is a permissioned set of permissionless participants, replicating the L1 validator problem.
Evidence: Over 60% of the active validator set on a major L2 like Arbitrum Nitro is operated by three entities running the same Geth fork. This client monoculture is the hidden centralization that MEV extraction exploits.
THE HIDDEN CENTRALIZATION OF PERMISSIONLESS SEQUENCER SETS
Centralization Pressure Points: Monolithic L1 vs. Modular Sequencer Set
Compares the systemic centralization risks inherent in a single-chain execution environment versus a modular rollup ecosystem with a shared sequencer set.
| Pressure Point | Monolithic L1 (e.g., Solana) | Modular Sequencer Set (e.g., Espresso, Astria, Shared Sequencer) |
|---|---|---|
Validator/Sequencer Count | ~2000 Validators | 3-10 Active Sequencers (Initial Phase) |
Client Diversity |
| 1-2 Client Implementations (e.g., Espresso HotShot) |
Economic Capture Surface | Stake-weighted voting (Liquid Staking Derivatives > 30%) | MEV Auction Revenue + Base Fee Capture |
Censorship Resistance Failure Mode |
|
|
Time to Finality Under Attack | Network Halt (requires social consensus fork) | L1 Force Inclusion via Smart Contract (1-2 L1 blocks) |
Upgrade Governance | On-chain voting by token holders | Off-chain multi-sig (e.g., 5-of-9) for sequencer software |
MEV Extraction Point | Diffuse across validators | Centralized at Sequencer Set → Enables MEV Auctions (e.g., via SUAVE) |
Cost to Attack (Est.) | ~$10B+ (33% of staked SOL) | < $100M (Acquire sequencer keys/bribe set) |
counter-argument
THE HIDDEN CENTRALIZATION
The Shared Sequencer Fallifier
Permissionless sequencer sets create a false sense of decentralization while consolidating power with the infrastructure provider.
Permissionless does not mean decentralized. A set allowing anyone to join is useless if economic or technical barriers prevent meaningful participation. The infrastructure provider controls the set, dictating the client software, messaging layer, and economic model, creating a single point of failure and censorship.
Shared sequencers centralize MEV. Projects like Espresso Systems and Astria aggregate ordering rights, creating a new MEV cartel. This transfers extractable value from individual rollups to the sequencer network operator, contradicting the core promise of credibly neutral block space.
The liveness guarantee is a myth. A sequencer network claiming liveness via proof-of-stake or Tendermint relies on its own validator set. This substitutes Layer 1 liveness with a smaller, less battle-tested consensus layer, increasing systemic risk for all connected rollups like Arbitrum or Optimism.
Evidence: The leading shared sequencer testnets show <10 active block producers, with the top 3 controlling >60% of stake. This is more centralized than Ethereum's validator set and mirrors the early Binance Smart Chain model.
risk-analysis
THE HIDDEN CENTRALIZATION OF PERMISSIONLESS SEQUENCER SETS
The MEV and Censorship Endgame
Decentralized sequencer sets promise neutrality but often mask concentrated power, creating systemic risks for MEV extraction and transaction censorship.
01
The Permissionless Illusion
Open membership for sequencers doesn't guarantee decentralization. Economic and technical barriers create a de facto oligopoly.
- Staking requirements of $10M+ price out all but whales and institutions.
- Hardware/bandwidth needs for sub-second slots favor centralized cloud providers like AWS.
- The top 3-5 sequencers often control >60% of block production, a vector for covert collusion.
>60%
Top 5 Control
$10M+
Entry Cost
02
MEV Cartels by Another Name
A concentrated sequencer set is a proto-cartel. Without robust PBS, they can internalize the most profitable MEV.
- They can run front-running bots and arbitrage strategies with privileged order flow access.
- Cross-domain MEV (e.g., between Arbitrum and Optimism) is captured by the same few entities operating on both.
- This undermines the credible neutrality promised by L2s, reverting to a broker-dealer model.
~90%
MEV Captured
0s
Latency Advantage
03
Soft-Censorship is Inevitable
Regulatory pressure will target the choke point: the sequencer. A small set is easy to coerce into filtering transactions.
- OFAC compliance can be enforced by threatening the few corporate entities running sequencers.
- Techniques like delayed inclusion or exorbitant fees for 'non-compliant' tx create plausible deniability.
- This breaks the atomic composability guarantee for DeFi, fragmenting liquidity and state.
<10
Entities to Pressure
100%
Tx Filter Risk
04
The Builder-Proposer Separation (BPS) Mandate
The only viable endgame is enforcing Ethereum's PBS model at the L2 layer. Sequencers must be pure proposers.
- A decentralized, permissionless builder market (like Flashbots SUAVE) competes on block construction.
- Sequencers' role reduces to ordering and attesting to the highest-paying block, removing their MEV edge.
- This aligns incentives with Ethereum's roadmap, creating a unified, resistant cross-rollup ecosystem.
1-of-N
Trust Model
-99%
Sequencer MEV
05
Intent-Based Routing as a Bypass
Users can sidestep centralized sequencers entirely by expressing outcomes, not transactions. Protocols like UniswapX and CowSwap demonstrate this.
- Solvers compete to fulfill the intent, breaking the sequencer's monopoly on order flow.
- Cross-chain intents (via Across, Socket) use fraud proofs, not sequencer liveness.
- This shifts power from chain operators to a solver network, reducing systemic censorship risk.
1000+
Solver Network
0
Sequencer Reliance
06
The Shared Sequencer Trap
Shared sequencers (like Espresso, Astria) solve fragmentation but consolidate power into a new, network-level single point of failure.
- They create a meta-sequencer cartel controlling dozens of rollups and $10B+ TVL.
- While using Tendermint or EigenLayer for consensus, the validator set is often the same concentrated capital.
- This trades L2 centralization for L1.5 centralization, a more catastrophic failure mode.
1
Meta-Point of Failure
$10B+
Consolidated TVL
future-outlook
THE SEQUENCER TRAP
Beyond the Permissionless Illusion
The permissionless sequencer set is a marketing term that obscures extreme centralization risks in transaction ordering.
Sequencer sets are permissioned cartels. A network with 100 permissionless sequencers controlled by 5 entities is a 5-of-100 multisig. The Sybil resistance is zero; the economic security is the combined stake of the controlling cartel, not the full set.
Decentralization theater creates systemic risk. The Lido/Coinbase problem replicates in sequencing. If 60% of 'independent' sequencers run on AWS in us-east-1, the network has a single point of failure. This is worse than a known, audited centralized sequencer.
Intent-based architectures bypass the problem. Protocols like UniswapX and CowSwap abstract sequencing away from L2s. They route user intents via a solver network, making the underlying sequencer set irrelevant for final execution quality.
Evidence: As of Q1 2024, over 95% of transactions on major 'permissionless' rollups are ordered by a single sequencer operated by the founding team. The promised decentralization is a roadmap item, not a production feature.
takeaways
PERMISSIONLESS SEQUENCER TRAPS
TL;DR for Protocol Architects
The promise of decentralized sequencing is undermined by economic and technical centralization vectors that emerge post-launch.
01
The Liveness vs. Censorship Dilemma
A permissionless set of sequencers must be large for liveness, but small for effective governance against censorship. This creates a fundamental trade-off.\n- Key Problem: A large, anonymous set cannot be effectively slashed for malicious ordering.\n- Key Insight: Systems like Espresso or Astria rely on a staked, reputation-based subset for practical safety, reintroducing centralization pressure.
~10-20
Effective Set Size
>100
Theoretical Set
02
MEV Cartel Formation is Inevitable
Without enforced commit-reveal schemes or encrypted mempools, the economic incentive to form a dominant block-building cartel is overwhelming.\n- Key Problem: Top sequencers by stake will collude off-chain to maximize extractable value, marginalizing smaller players.\n- Key Insight: This mirrors the miner extractable value (MEV) centralization in Ethereum post-merge, but with faster feedback loops. Protocols need SUAVE-like designs to be viable.
>80%
Cartel TVL Share
$B+
Protected Revenue
03
The Infrastructure Moat
High-performance sequencing requires low-latency, globally distributed infrastructure and proprietary optimization software, creating a massive barrier to entry.\n- Key Problem: Permissionless entry is a myth; in practice, only well-funded entities (e.g., Figment, Chorus One) can compete.\n- Key Insight: This leads to a de facto whitelist, where the 'set' is just the few nodes that can afford the ~$50k/month in infra costs and R&D.
<100ms
P99 Latency Req'd
10x
Cost Advantage
04
Solution: Enshrined Proposer-Builder Separation (PBS)
The only viable long-term model is to separate block building (competitive, permissionless) from block proposal (randomized, staked).\n- Key Benefit: Breaks the link between stake and MEV capture, preventing cartel formation.\n- Key Benefit: Allows for a truly permissionless builder market, similar to Ethereum's roadmap, implemented at the rollup level via protocols like Astria's shared sequencer.
1/N
Proposer Chance
0
Builder Stake
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall