← All articles

Geography Is Not Neutral: How Block Building Shapes Where Validators Live

@0x3hu79
@0x3hu79
April 7, 2026 · Protocol Design
#PBS#ePBS#decentralization#validators#latency#block-building

Where Ethereum validators physically operate matters more than most protocol discussions acknowledge. If validators cluster in the same regions, they become jointly vulnerable to regional outages, jurisdictional actions, and correlated infrastructure failures. The 2021 Bitcoin mining ban in China demonstrated what geographic concentration risk looks like at scale.

Yet major blockchain protocols don’t encode geography in their rules. Validator locations emerge from economic incentives, infrastructure availability, and deployment choices. The question is: do Ethereum’s block-building mechanisms systematically push validators toward geographic concentration?

A paper by Sen Yang (Yale/IC3), Burak Oz (Flashbots), Fei Wu (King’s College London), and Fan Zhang (Yale/IC3), published in ACM SIGMETRICS, provides the first formal answer. Using analytical modeling and agent-based simulations calibrated to real-world latency data, they show that neither local nor external (PBS) block building is geographically neutral - and that the two paradigms centralize geography through fundamentally different mechanisms.

Two Paradigms, Two Migration Pressures

The paper models two block-building approaches that Ethereum validators can adopt:

Local block building: the validator collects signals from distributed sources (centralized exchanges, wallet RPCs, mempool data) and constructs the block locally. The validator’s payoff depends on latency to both information sources and attesters.

External block building (PBS): the validator outsources construction to a builder/supplier and simply proposes the resulting block. The validator’s payoff depends primarily on latency to the chosen supplier.

Both paradigms create location-dependent payoffs - validators in certain regions earn systematically more than validators in others, creating incentives to migrate. But the mechanisms differ in ways that matter for protocol design.

Local building: pulled toward information hubs

Under local building, a validator’s block value grows with the freshness of data from each information source. Lower latency to a signal source means fresher prices, newer transactions, better MEV. Since these gains are additive across sources, the migration advantage scales with the number of sources concentrated in a region.

The result: validators are pulled toward well-connected hubs where many information sources cluster - the Atlantic corridor (US-East, Western Europe) that already dominates Ethereum’s validator distribution.

Importantly, the validator must also remain close enough to attesters for the block to become canonical. Under local building, the proposer broadcasts the block directly, so the proposer’s location determines propagation to attesters. This creates a balancing force: migrate toward information, but not so far that you can’t reach attesters in time.

External building: pulled toward suppliers

Under PBS, the calculus changes. The validator doesn’t need to be close to information sources - the builder handles that. What matters is the latency to the selected supplier (builder/relay), because this determines how late the validator can commit while still getting the block propagated to attesters in time.

The migration advantage under external building is bounded by the single proposer-to-supplier hop. Moving closer to one more supplier doesn’t help if you’re already close to the best one. The advantage doesn’t scale with the number of suppliers - it’s bottlenecked by the relationship with the single best supplier.

The result: validators migrate toward builder and relay infrastructure, not toward information sources. Under PBS, supplier geography becomes a central determinant of validator positioning.

The Counterintuitive Finding: Source Placement Sensitivity

Perhaps the most surprising result is how the two paradigms respond differently to where information sources are placed geographically.

Under local building, concentrating signal sources in low-latency regions (US-East, Europe) amplifies centralization most strongly. Validators migrate toward the hub where sources AND attesters are already concentrated - a positive feedback loop.

Under external building, the opposite holds: concentrating information sources in high-latency regions (Africa, South America, Southeast Asia) creates stronger centralization pressure. Why? Because validators migrate to co-locate with suppliers who serve those remote sources. A major builder operating from a remote region creates a larger marginal payoff gain from co-location than a builder already in a well-connected hub.

This means that under PBS, the geographic distribution of builders and relays has outsized importance for validator concentration. The protocol doesn’t directly control where builders set up shop, but the resulting validator migration pattern is a direct consequence of builder geography.

Consensus Parameters as Hidden Geographic Levers

The paper identifies two protocol parameters that modulate geographic centralization pressure in ways that protocol designers may not anticipate:

Attestation threshold (γ)

The attestation threshold determines what fraction of attesters must vote positively for a block to become canonical. The paper finds:

This is a striking result: the same parameter change has opposite geographic effects depending on which block-building paradigm dominates. Since external building (via MEV-Boost) currently accounts for over 90% of Ethereum blocks, the external-building dynamic is the one that matters in practice.

Slot time (Δ)

With active discussion around shorter slots (EIP-7782 proposes 6 seconds), the paper tests halving the slot time from 12s to 6s.

The finding: shorter slots don’t materially change aggregate geographic concentration - the Gini coefficient, HHI, and liveness coefficient remain roughly the same. But shorter slots increase reward disparity between regions. The same absolute latency difference (say, 50ms between US-East and Southeast Asia) represents a larger fraction of the available timing window, amplifying the payoff advantage of well-positioned validators.

In practical terms: shorter slots don’t make more validators cluster together, but they make the reward gap between well-positioned and poorly-positioned validators larger. Over time, this reward disparity could strengthen migration incentives that the aggregate metrics don’t yet capture.

Even Starting Uniform, Validators Converge

One of the paper’s most important demonstrations: starting from a completely uniform geographic distribution (validators spread evenly across all regions), both paradigms produce emergent geographic concentration. The Gini coefficient rises from near-zero to substantial values over 10,000 simulated slots.

This means geographic centralization isn’t just an artifact of Ethereum’s historical development or the fact that early adopters were in the US and Europe. It’s a structural outcome of the protocol’s incentive design. Even a hypothetical fresh start with perfect geographic balance would converge toward concentration under current rules.

When initialized with Ethereum’s actual validator distribution (heavily US and Europe), both paradigms converge even faster - the initial concentration creates a self-reinforcing dynamic where migration benefits saturate quickly.

What This Means for the Broader Conversation

Supplier geography is a first-order concern under PBS

The blockspace market paper identified “portability without proximity lock-in” as a guiding principle. This paper formalizes exactly why that principle matters: under PBS/ePBS, builders and relays are the geographic anchors that pull validators. The more concentrated builder infrastructure is, the stronger the centralization pressure on the entire validator set.

This has direct implications for relay block merging, multi-relay coordination, and BuilderNet - all mechanisms discussed at the Blockspace Forum Workshop. If these coordination mechanisms consolidate relay infrastructure into fewer geographic locations, they could inadvertently amplify validator concentration. Conversely, if they enable effective relay operation across more regions (using technologies like Optimum’s RLNC for cross-region communication), they could weaken the migration pressure.

Protocol parameter changes need geographic impact assessment

The paper makes a compelling case that attestation threshold changes and slot-time reductions - typically discussed in terms of security, liveness, and throughput - should also be evaluated for their geographic side effects. This is a new dimension for EIP review that the community hasn’t systematically considered.

Geographic decentralization should be a design objective, not an afterthought

The paper’s strongest recommendation: treat geography as a first-order protocol design concern. The current approach - assume decentralization emerges naturally - doesn’t hold under the paper’s analysis. Both building paradigms create systematic incentives that undermine geographic diversity, and these incentives are amplified by asymmetric infrastructure placement and existing validator concentration.

MEV-burn has geographic benefits

The paper notes that mechanisms like MEV-burn, which reduce the economic reward from timing games, would weaken the latency-based migration incentive. This is typically discussed as a fairness intervention - the geographic benefit is a secondary effect worth recognizing.

For Block Builders and Relay Operators

Where builders and relays operate isn’t just an operational choice - it’s a protocol-level influence on validator geography. Under external building, validators migrate toward supplier infrastructure. If the builder/relay ecosystem concentrates in a few locations, the validator set follows. Builders who operate from diverse geographic locations actively contribute to geographic decentralization - even if that isn’t their primary motivation.

Shorter slot times increase the value of geographic positioning. If Ethereum moves to 6-second slots (or sub-slot execution becomes prevalent), the reward gap between well-positioned and poorly-positioned infrastructure widens. Builders and relays with low-latency positions benefit more; those in higher-latency regions face a proportionally steeper disadvantage.

Multi-region relay coordination (like the workshop discussions around constraint sharing and Optimum’s RLNC) has geographic decentralization benefits beyond the immediate operational ones. If relays can effectively coordinate across regions without concentrating in one place, the migration pressure on validators weakens.

Sources

Original paper:

Authors’ summary post:

Simulation dashboard:

Related protocol proposals:

Background: