Modularity centralizes curation. Decoupling execution from consensus and data availability creates a marketplace of specialized providers, but the user's path through this maze is dictated by a few dominant sequencers, bridges, and indexers like Celestia, EigenDA, and The Graph.
public-goods-funding-and-quadratic-voting
Blog
The Cost of Centralized Curation in Modular Stacks
A critique of how modular blockchain architectures, while decentralizing execution, often re-introduce centralization at the critical application and funding layers through curated allow-lists and committees, creating a permissioned ecosystem that stifles innovation.
introduction
THE COST OF ABSTRACTION
Introduction: The Modular Mirage
Modular blockchain design outsources core functions, but the resulting fragmentation creates a new, opaque layer of centralized curation that dictates user experience and cost.
The 'best' chain is a mirage. Users don't choose an L2 like Arbitrum or Optimism based on its virtual machine; they choose the application and the liquidity path pre-selected by its developers, who are constrained by the integration deals of their chosen stack (e.g., OP Stack, Arbitrum Orbit).
Costs are obfuscated, not eliminated. While rollups reduce L1 gas fees, they add new cost layers: sequencer profit margins, cross-chain messaging fees via LayerZero or Wormhole, and premium data availability pricing. The final user price is a black box.
Evidence: Over 90% of rollup transaction ordering is controlled by a single, centralized sequencer, creating a de facto tollgate on what was promised as a permissionless execution layer.
thesis-statement
THE COST
Core Thesis: Curation is the New Choke Point
Centralized curation in modular stacks creates systemic risk and rent extraction, shifting the bottleneck from execution to data availability and interoperability.
The bottleneck has shifted. The scaling trilemma is solved by modular architectures like Celestia and EigenDA, but this creates a new problem: the coordination and curation of specialized components. The cost is no longer raw compute, but the overhead of managing a fragmented stack.
Centralized curation is a systemic risk. Teams default to trusted, centralized sequencers and bridges like Stargate and Axelar for speed, creating single points of failure. This reintroduces the censorship and liveness risks that decentralization was built to solve.
Curation extracts value. The sequencer cartel on major L2s captures MEV and transaction ordering rents. In bridging, centralized relayers and watchtowers act as tollbooths. The value accrues to the curators, not the users or the base layers.
Evidence: The top five L2s by TVL all use a single, centralized sequencer. The Celestia DA cost for a 1MB blob is ~$0.001, but the curation layer (the rollup's sequencer) charges users 100-1000x that in gas fees for the privilege of access.
key-trends
THE COST OF CENTRALIZED CURATION
The Centralization Playbook: Three Emerging Patterns
Modularity's promise of permissionless innovation is being undermined by centralized chokepoints that control access, data, and execution.
01
The Sequencer Monopoly
Rollups like Arbitrum and Optimism outsource block production to a single, profit-maximizing sequencer. This creates a single point of failure and censorship, while MEV is captured by the core team instead of the validator set.
- Problem: Centralized sequencer controls transaction ordering and latency.
- Solution: Shared sequencer networks like Espresso or Astria aim to decentralize this critical layer.
~12s
Forced Delay
100%
MEV Capture
02
The Data Availability Cartel
While Ethereum provides credible neutrality, its high cost pushes rollups to alternative Data Availability (DA) layers like Celestia or EigenDA. This creates a new curation game where a small committee of operators becomes the gatekeeper for L2 state validity.
- Problem: DA committees can theoretically withhold data, halting chains.
- Solution: Proof-of-stake sampling and light-client verification, as pioneered by Celestia, aim to minimize trust.
>100x
Cost Savings
<50
Key Operators
03
The Bridge Oligopoly
Interoperability is dominated by a handful of canonical bridges (Polygon PoS, Arbitrum Bridge) and third-party services like LayerZero and Wormhole. Their validator sets and oracle networks are centralized, creating systemic risk for ~$30B+ in bridged value.
- Problem: A 4/9 multisig failure can freeze billions.
- Solution: Light-client bridges and fraud-proof systems, as seen in IBC and Succinct, move towards trust-minimization.
$30B+
TVL at Risk
4/9
Multisig Threshold
MODULAR STACK COST ANALYSIS
The Curation Spectrum: From Permissionless to Permissioned
Comparing the explicit and implicit costs of data availability and sequencing curation models in modular blockchain architectures.
| Curation Model & Cost Vector | Permissionless (e.g., EigenDA, Celestia) | Permissioned Consortium (e.g., Espresso, Astria) | Sovereign Rollup (e.g., Fuel, dYdX v4) |
|---|---|---|---|
Data Availability Cost per MB | $0.10 - $0.50 | $1.00 - $5.00 | $0.00 (Self-published) |
Sequencer MEV Capture | High (Public mempool) | Controlled (Private mempool) | Sovereign (Rollup operator) |
Censorship Resistance Guarantee | Economic (Stake slashing) | Social (Consortium governance) | Technical (Force inclusion via L1) |
Time-to-Finality for L1 Settlement | ~20 minutes (Ethereum) | ~2 minutes (Fast L1 bridge) | ~0 minutes (Self-finalizing) |
Protocol Upgrade Latency | Weeks (Governance + fork) | Days (Consortium vote) | Hours (Sovereign upgrade) |
Interop Bridge Trust Assumption | Light client + fraud proof | Multi-sig / MPC committee | Native validation (IBC-like) |
L1 Execution Gas Overhead | ~8,000 gas (blob calldata) | ~50,000 gas (bridge contract) | ~0 gas (no verification) |
deep-dive
THE ARCHITECTURAL TRAP
The Real Cost: Stifled Innovation & Captured Public Goods
Centralized curation in modular stacks creates systemic risk by capturing public infrastructure and imposing innovation taxes.
Centralized sequencers and shared DA layers become single points of failure and control. This architecture replicates the client diversity problem of early Ethereum, where Geth dominance created systemic risk. A single sequencer like those on Arbitrum or Optimism controls transaction ordering and MEV extraction, creating a de facto rent-extracting gatekeeper.
The innovation tax is a hidden cost. Builders must seek permission from and integrate with a curated set of providers (e.g., Celestia for DA, EigenLayer for restaking). This centralized roadmap dictates protocol evolution, stifling competition from alternatives like Avail, Near DA, or Babylon. The ecosystem fragments into walled gardens of interoperability.
Public goods become captured infrastructure. Shared security layers like EigenLayer and cross-chain messaging layers like LayerZero or Axelar function as private toll roads on public rails. Their economic models and upgrade governance are not credibly neutral, creating long-term rent extraction from the applications built atop them.
Evidence: The 2024 EigenLayer operator set is permissioned and curated. A failure or malicious action within this set compromises every AVS (Actively Validated Service) built on it, demonstrating the systemic risk of centralized curation at the base layer.
counter-argument
THE TRADEOFF
Steelman: "But We Need Quality Control!"
Centralized curation in modular stacks introduces a single point of failure and rent extraction, undermining the core value proposition of decentralized infrastructure.
Centralized curation is a single point of failure. A core team or foundation acting as a gatekeeper for rollup sequencers, data availability layers, or shared sequencer networks creates a critical vulnerability. This reintroduces the trusted intermediary that modular architecture was designed to eliminate.
Curation creates rent-seeking bottlenecks. Entities like Celestia's core team or an EigenLayer AVS operator can extract value by controlling access to critical resources. This centralizes economic power and stifles permissionless innovation at the infrastructure layer.
Market mechanisms outperform top-down control. Permissionless competition between data availability providers like Celestia, Avail, and EigenDA, or between shared sequencer networks like Espresso and Astria, provides superior quality control. Poor performance is punished by the market, not a committee.
Evidence: The rapid commoditization of data availability proves this. After Celestia's launch, competitors like Avail and EigenDA emerged, driving down costs and improving service through permissionless competition, not centralized curation.
takeaways
THE COST OF CENTRALIZED CURATION
Key Takeaways for Builders and Investors
Centralized sequencers and data availability committees create systemic risk and hidden costs in modular stacks. Here's where the leverage is.
01
The Problem: Sequencer Extractable Value (SEV)
Centralized sequencers are the new miners, capturing MEV and transaction reordering rents. This creates misaligned incentives and degrades user experience.
- Censorship Risk: Single entity controls transaction inclusion.
- Revenue Leakage: Value that should accrue to the protocol or users is captured by the sequencer.
- Fragmented Liquidity: Each rollup's sequencer creates its own liquidity silo.
>20%
of tx value
~500ms
Latency Tax
02
The Solution: Shared Sequencing Layers
Networks like Astria, Espresso, and Radius decouple execution from sequencing. This enables atomic cross-rollup composability and democratizes block building.
- Atomic Composability: Enables seamless interactions between sovereign rollups.
- MEV Redistribution: Fair ordering protocols can redistribute value.
- Credible Neutrality: No single entity controls the transaction pipeline.
0 to 1
New Primitive
$1B+
Market Cap
03
The Problem: Data Availability (DA) Cartels
Relying on a small Data Availability Committee (DAC) or a single chain like Ethereum for data creates a centralization bottleneck and cost floor.
- Cost Inefficiency: Paying for ~80 KB/s of global throughput on a monolithic chain.
- Single Point of Failure: A DAC's liveness failure breaks all derived rollups.
- Vendor Lock-in: Builders are tied to the economic and governance model of one DA layer.
~80 KB/s
Global Throughput
4-8 Members
Typical DAC Size
04
The Solution: Modular DA & Proof Markets
Layers like Celestia, EigenDA, and Avail provide scalable, dedicated data availability. Proof markets like Near DA and zkPorter further separate security from execution.
- Cost Scaling: DA costs decrease with dedicated, scalable layers.
- Security Choice: Builders can select DA based on security/cost trade-offs.
- Interoperable State: Enables light clients to verify data across chains.
100x
Cheaper DA
~2s
Finality
05
The Hidden Tax: Interoperability Fragmentation
Each curated stack (e.g., OP Stack, Arbitrum Orbit, zkSync Hyperchain) creates its own walled garden. Bridging between them reintroduces the very trust assumptions modularity aimed to solve.
- Trusted Bridges: Moving assets between stacks often requires a new trusted bridge.
- Liquidity Silos: Capital is trapped within each ecosystem.
- Complexity Explosion: Developers must integrate N bridges for N ecosystems.
N Bridges
Integration Cost
$10B+ TVL
At Risk
06
The Endgame: Intents & Shared Security
The architecture shifts from transaction-based to intent-based, abstracting away curation. Networks like Succinct, Hyperlane, and Polymer provide modular interoperability and security.
- User Abstraction: Users express what they want, not how to do it (see UniswapX, CowSwap).
- Universal Interop: Light clients and ZK proofs enable trust-minimized communication.
- Restaking Security: Projects like EigenLayer and Babylon allow reuse of Ethereum's economic security for new modules.
90%+
UX Improvement
1 SDK
To Integrate All
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