Block Construction Timing and Wait Dynamics

Summary

Every participant in the MEV-Boost pipeline is incentivized to wait. Builders delay submitting their best bids to capture more CEX price information; validators delay signing to collect higher bids; relays buffer to find the maximum-value block. The result: a systematic pipeline of delays whose aggregate effect is measured in seconds, not milliseconds. A 2026 analysis identifies three key metrics — block value, capacity management, and time-to-inclusion — and finds that the spread between median and 95th-percentile bid timing exceeds 1 second. This is pipeline design, not physical necessity.

The Three Metrics of Block Construction Performance

1. Block Value (Who Captures It?)

Block value is highly concentrated among a small number of builders:

The top 2-3 builders consistently produce 80% of all blocks
This concentration is self-reinforcing: high-value builders attract exclusive order flow → exclusive order flow increases block value → high block value improves relay ranking
Consequence: 2-3 builders setting prices → less competitive bidding than a more distributed market would produce
The marginal block value from the 4th or 5th builder is significantly lower, but their presence is what forces the top builders to bid competitively

❓ Open question: is today’s builder concentration above or below the value-maximizing level for proposers?

2. Capacity Management (Lean Blocks, Late Submissions, Blob Compounding)

Builders manage block capacity strategically:

Lean blocks: builders often leave gas unused rather than filling to the limit. Reasons:

Late-arriving high-fee transactions may arrive after the block was finalized
Sparse blocks allow the relay or builder to append additional transactions post-construction (see relay merging)
Simulation cost: simulating edge cases at the gas limit is expensive; builders accept the risk of leaving value on the table

Late submission: the longer a builder waits, the more CEX price information they have. See Timing Games and Proof-of-Time for the full timing game analysis. The key capacity implication: late submission means less time to fill the block. Builders solve this by pre-building block templates and incrementally updating them as new transactions arrive and CEX prices move.

Blob compounding: post-Dencun, blocks include both execution gas and blob space. A builder that fills blob space optimally may sacrifice execution gas revenue and vice versa. Blob-heavy blocks can crowd out high-fee execution transactions. Under Fusaka’s increased blob limits, this tradeoff becomes more acute.

3. Time-to-Inclusion (The Hidden Tax)

Time-to-inclusion measures how long a transaction waits between submission and block inclusion. Unlike block propagation time (which is a network problem), time-to-inclusion depends on builder strategy:

Key empirical finding (Natale & Moser 2023, arxiv:2312.09654):

There is no public cutoff time beyond which builders stop accepting transactions
Validators gain approximately 5% extra block value per week by strategically delaying block signing
Solo stakers (who can’t optimize timing as precisely as institutional validators) are systematically disadvantaged: they face the same waiting as other validators but lack the infrastructure to extract timing alpha

1-second spread: the spread between median bid submission time and 95th-percentile bid submission time exceeds 1 second within the same slot. This means different proposers — depending on their infrastructure — observe wildly different bid landscapes at their actual decision moment.

Cascade effect: each participant waiting a little longer compounds:

Builder waits to last safe moment (3.9s into 12s slot)
Relay waits to verify merging viability
Validator waits for marginal bid improvement
Total actual pipeline: sometimes 4-5 seconds of delay before a block is actually committed

The Nash Equilibrium of Waiting

Why does everyone wait? Because unilateral early submission is dominated:

A builder submitting at second 2 when all others submit at second 4 loses 2 seconds of CEX price observation
A validator signing at second 4 when others sign at second 6 misses 2 seconds of higher bids
A relay merging at second 3 rather than second 4 merges a smaller pool of candidate transactions

The equilibrium is: everyone waits until the latest safe moment. “Safe” is defined differently by each participant based on their risk tolerance for missed slots.

Implication for protocol design: timing games cannot be fixed by asking participants to submit earlier. They must be fixed by changing the incentive structure (PoT, ePBS, 2-Prop) or by shortening the slot time (sub-slot execution).

Solo Staker vs. Institutional Validator Asymmetry

Solo stakers are disproportionately punished by timing dynamics:

Infrastructure gap: institutional validators run custom bid aggregators and monitoring infrastructure that alerts them to the optimal signing moment. Solo stakers use off-the-shelf MEV-Boost with default timeouts.
5% timing alpha: the empirical 5% extra weekly value from timing-optimal strategies is available to institutions but not solo stakers.
Attestation risk: signing late to capture higher bids risks missing the attestation deadline. Institutions know their exact network latency to the beacon chain; solo stakers have higher variance.
Relay coverage: institutional validators often maintain relationships with multiple relays and can fall back quickly. Solo stakers’ fallback on missed relay responses is slower.

Systematic 5%+ underperformance compounds over time, threatening solo staker economics and validator decentralization.

Blob Timing and Proof Generation Dependency

Blobs (post-Dencun) add a second timing dimension:

zkEVM proof generation depends on knowing the block contents. ZK rollups that post proofs cannot start proving until they receive the full block.
A block published late (because the builder waited) starts the ZK proving clock later.
Under Fusaka’s increased blob counts, later block arrival → later proof availability → tighter proof submission windows for rollup sequencers.

The head vote penalty for late block publishing affects not just the current slot’s rewards but downstream ZK rollup operations. Late validators penalize the L2 ecosystem, not just themselves.

Pipeline Design, Not Physics

The core argument of the 2026 analysis:

The observed delays are not imposed by physics (network propagation times are known and bounded)
They are imposed by incentive structures that reward waiting
The >1 second spread between median and 95th-percentile submission is evidence that the timing game is played, not that the network is slow

Counterfactual: if builders submitted their best blocks at second 1 of every slot, proposers would receive very similar values (the marginal value of extra CEX observation is highest near the deadline). The reason they don’t is not technical: it’s economic.

Implication: solutions to timing games don’t need to change the network’s physical properties. They need to change the economic calculation that makes late submission rational.

Interventions and Their Efficacy

Intervention	Mechanism	Effectiveness
Proof-of-Time (PoT)	Physics-certified submission deadlines	Raises V* threshold; reduces incentive to delay
ePBS	Proposer commits to payload hash early	Eliminates builder-side timing game post-commitment
2-Prop (two proposers)	Competition between proposers	NE is fast proposal (homogeneous case)
Sub-slot execution (TOOL)	Shorter slots reduce forecast window	Reduces value of waiting per sub-slot
Relay block merging	Post-auction non-contentious appends	Does not address timing directly; may add latency

No single intervention addresses all three metrics (block value concentration, capacity management, time-to-inclusion) simultaneously.

Bid value evolution over a slot: rises approximately log-linearly from second 0 to second 4, then flattens (see Natale & Moser 2023)
Builder market share: top 2-3 builders = ~80% of blocks; BuilderNet = ~26% (Coin Metrics, 2026)
Relay market: Titan, BuilderNet, Flashbots relay cover ~80%+ of blocks; relay diversity has declined since 2024