Splitting a blockchain into separate execution, settlement, consensus, and data-availability layers. Optimising each independently. Composing them on demand. The modular thesis says the monolithic chain has always been an artefact of constraint, not design.
Modularity is the architectural counter-thesis to monolithic chains. Mustafa Al-Bassam's LazyLedger paper in 2019 — which became Celestia — argued that a blockchain's job is consensus and data availability, not execution. Celestia mainnet went live October 2023 with TIA airdropped to ETH stakers, IBC users, and Cosmos contributors. The pitch is clean: rollups post their data to Celestia, get cheap DA, and bring their own execution environment. Data availability sampling (DAS) lets light nodes verify availability with logarithmic resources. The design is elegant; whether the market wants it is a different question.
EigenDA — built by EigenLabs on top of EigenLayer's restaked ETH — is the institutional alternative. Same problem (DA), different security model (restaked ETH instead of TIA-staked). Avail spun out of Polygon and runs its own DA layer with similar primitives. NearDA piggybacks on Near's existing chain. Polygon CDK, Arbitrum Orbit, OP Stack — every L2 platform now lets builders pick their DA layer at launch time. The DA market is small in dollars and structurally important to the rollup-centric Ethereum thesis.
The honest debate is whether modular wins. Solana's Anatoly argues that the integrated stack is faster and cheaper at the application layer, that latency between modules is the dealbreaker, and that monolithic execution will out-iterate modular composition. The Celestia and Ethereum camps argue that sovereign rollups with cheap DA scale beyond what any monolith can support, that specialization beats integration at scale, and that the next billion users will run on chains we haven't named yet. Both sides are betting real treasuries on the answer. The next three years tell us who's right.
We propose LazyLedger, a design for distributed ledgers where the consensus layer's only role is to order and guarantee the availability of transaction data. Application logic — the rules for what makes a transaction valid — is enforced by clients themselves rather than by the consensus protocol. This separation of concerns lets the consensus layer scale by doing less.LazyLedger paper · arXiv →
The modular thesis: a blockchain does four jobs — execution, settlement, consensus, data availability — and there's no reason a single chain has to do all four. Mustafa Al-Bassam's 2019 LazyLedger paper articulated the split. Celestia made it a network. EigenDA and Avail made it a market. Sovereign rollups proved settlement is optional. The bet is that specialization wins, and that DA bandwidth is the scarce resource the next decade fights over.
Celestia is the first production data availability layer. It does one job: order data and guarantee it was published. Mustafa Al-Bassam, Ismail Khoffi, and John Adler shipped mainnet October 2023. Light nodes use Data Availability Sampling — randomly downloading small chunks to probabilistically verify the full dataset is available — which scales DA bandwidth with the number of light nodes rather than against it. Rollups post blob data to Celestia and inherit DA security without paying Ethereum's premium. The clearest 'do one thing well' bet in crypto infrastructure.
EigenDA is a data availability service built on EigenLayer — restaked ETH securing an AVS that signs over data blobs. Operators run DA nodes; restakers delegate ETH to them; if operators sign for unavailable data, they get slashed. Throughput targets are aggressive (10MB/s+), and the trust model is restaked Ethereum stake rather than a sovereign token. Mantle and several Arbitrum Orbit chains use it. The pitch: DA-as-a-service that inherits Ethereum's economic security through restaking rather than competing with it.
Avail spun out of Polygon in 2023 and shipped mainnet in 2024 as a standalone DA layer. KZG commitments plus erasure coding plus light-client DAS — same fundamental architecture as Celestia, different validator set and different go-to-market. Anurag Arjun (ex-Polygon co-founder) leads. Avail Nexus pitches a unified rollup hub layered on top. The DA market is now a three-way race for blob throughput: Ethereum blobs (4844), Celestia, EigenDA, Avail. Pricing pressure across all of them is real.
A sovereign rollup posts data to a DA layer but settles within its own social consensus — no smart contract on a base chain enshrines the canonical state. Users running full nodes (or watching light clients) decide which fork is real. Pioneered conceptually by Adler and Al-Bassam, made practical by the Rollkit framework on Celestia. The trade-off: no enforced bridge to a settlement chain, but full sovereignty over upgrades and forks. Closer to how Bitcoin and Ethereum actually work than to traditional smart-contract rollups.
DAS is the cryptographic primitive that makes modular DA scale. A block is encoded with a 2D Reed-Solomon erasure code; light nodes randomly sample a few chunks; if the sampling succeeds across enough nodes, the data is provably available with overwhelming probability. Al-Bassam, Sonnino, and Buterin's 2018-2019 work formalized it. The genius: throughput scales with the number of light nodes rather than asking each one to download everything. Ethereum's danksharding roadmap, Celestia, and Avail all converge on this primitive.
Conviction is stated as conviction; you decide what to do with it. Tiers below — Core, Conviction, Watch, Speculative — reflect how much of FRQNCY's attention each project currently earns, not a recommendation to buy.
EI
Conviction is theatre without practice. Five steps that turn the thesis above into something the body actually does, not just something the mind agrees with.
Sync the chain, sample data availability. Feel what DAS actually verifies.
Use Rollkit, Sovereign SDK, or Dymension RollApps. Post data to Celestia. Build the chain in a weekend.
Post 1MB to Celestia, EigenDA, Avail, and Ethereum blobs. The fee differences are the modular thesis in numbers.
Mustafa's data-availability sampling argument is the most important modular-thesis primitive.
Use IBC or a shared sequencer. Notice what cross-rollup composability requires.
The Crypto hub is the index of all sectors and the freedom-technology frame they share. The Fund is what happens when the same conviction gets put to work on behalf of the network.
Specialization beats integration at scale.
Consensus, data availability, execution — separable, composable, replaceable.
The next billion users run on chains we haven't named yet.
Modularity is a values bet.
FRQNCY · Crypto