Summary
Timing games are a structural problem in MEV-Boost: builders delay bid submission to the last safe millisecond before the block deadline, capturing more MEV by observing CEX price movements longer. This directly trades missed-slot risk against higher block value. Proof-of-Time (PoT) is a proposed mechanism that enforces submission deadlines using physical timing properties rather than trust, potentially eliminating timing games without requiring protocol changes.
The Problem: Why Builders Delay
In MEV-Boost, the proposer must broadcast their chosen block with enough time for it to propagate before the next slot. The typical “safe” deadline is ~4 seconds into the 12-second slot (leaving 8 seconds for propagation).
But builders notice: the longer they wait before submitting their best block, the more CEX price information they have. In CEX-DEX arbitrage, the builder that submits at second 3.9 sees ~4× more price information than one that submits at second 1.
Nash equilibrium: if Builder A delays to second 3.9, Builder B must also delay or lose competitive advantage. All builders race to submit as late as possible → systematic delay.
Cost: missed slots. If the winning block doesn’t propagate in time, the slot is missed and no rewards are distributed. The equilibrium delay is the point where the MEV gain from later submission exactly equals the expected cost of a missed slot.
Quantitative Impact
From the APR/latency study:
- Reducing block propagation time by 50–150ms is worth 0.66–1.97% APR uplift for validators
- Faster builds = better MEV capture per unit time = 13–16% MEV bid uplift under optimal conditions
- The mump2p RLNC approach (Optimum) can reduce block propagation latency by ~120–150ms globally
Timing games are the primary mechanism by which geographic co-location (builder near validator) translates to revenue advantage.
Proof-of-Time (PoT)
PoT proposes replacing trust-based bid deadlines with a physical timing guarantee derived from the speed of light.
Trust-Based PoT (Current)
Relays enforce submission deadlines contractually. Builders that miss the deadline get blocked. But relays can be gamed: a builder co-located with the relay gets effectively earlier deadlines.
Physics-Based PoT
A PoT oracle network measures actual network propagation times and signs timing certificates. The relay uses these certificates to enforce deadlines based on observed message arrival, not claimed submission time.
Nash equilibrium analysis:
- Let V = block value (MEV), c = speed-of-light propagation cost
- The equilibrium condition: E[S] = V/c - 1
- Threshold V* ≈ $8.67–$87 USD per block (depending on model parameters)
- Below V*: builders submit immediately (no timing game)
- Above V*: builders delay to extract maximum information value
With physics-based PoT, the effective V* is raised significantly because the penalty for late submission is a hard physical limit, not a soft trust-based one.
AdaptiveSwitch
The AdaptiveSwitch proposal (from the PoT research) adaptively selects between two submission modes:
- Turbo mode: submit at ~50ms before deadline (maximum MEV extraction, some missed-slot risk)
- Full mode: submit at ~127ms before deadline (safer, lower MEV)
The switch uses real-time network conditions to choose the safer option when missed-slot risk is elevated. Key driver for adoption: AI agents are now a majority of MEV consumers, and they update strategy faster than human traders — meaning timing games become more intense over time.
Sub-Slot Execution as an Alternative
The TOOL network (NuConstruct) proposes sub-slot execution as a different approach:
- Slots are divided into ~200ms sub-slots
- After each sub-slot, partial state transitions are shared
- Subsequent builders build on updated state rather than 12-second-old state
The claim: shorter forecast windows (1s vs. 12s) reduce variance for traders, increasing volume. More trades → more total fees even if per-trade profit is flat. From a timing game perspective, sub-slots reduce the advantage of delaying because there are frequent checkpoints.
Proposer game theory: proposers don’t see sub-slotted block values directly (blinded design) to prevent just-in-time switching to PBS blocks if the sub-slot value is lower.
Connection to Geographic Concentration
Timing games create a structural advantage for builders co-located with relays and major validator clusters:
- A builder 1ms closer to the relay gets 1ms more price observation time per block
- Over thousands of blocks, this compounds into significant MEV extraction advantage
- US/EU relay concentration means Asian validators and builders are structurally disadvantaged
RLNC-based block propagation (Optimum) can reduce the variance of propagation time, making the timing game less sensitive to geographic position.
Sub-Slot vs. PoT vs. ePBS
| Approach | Mechanism | Timing game impact |
|---|---|---|
| PoT (physics-based) | Certify submission timing via oracle network | Reduces advantage of late submission |
| Sub-slot execution | Divide slot into checkpoints | Reduces value of late submission |
| ePBS | Commit-reveal at consensus level | Builder commits to payload hash at ~4s; reveal at ~8s; no last-second games |
| Relay block merging | Non-contentious appends post-auction | Doesn’t directly address timing; may add latency |
ePBS most directly addresses timing games: once a builder commits to a payload hash (at the consensus-level bid), there’s no incentive to wait for late information, since they already committed.
Timing Games on FCFS Chains: Probabilistic Backrunning
On FCFS (First-Come-First-Served) chains (most L2s), a different form of timing game emerges: probabilistic backrunning (Mazorra, Schlegel, Mamageishvili, Flashbots/Offchain Labs, Feb 2026).
Instead of a proposer delaying block reveal, searchers compete to place transactions immediately after an opportunity that arrives at an unpredictable time during an inter-block window.
Equilibrium structure:
- n players; opportunity arrives uniformly at random; observed with delay
- High cost (c ≥ 1/e): each player submits at most one transaction in equilibrium (mixed strategy timing)
- Low cost (c < 1/e): multiple transactions per player → spam is the equilibrium
- Key insight: timing in FCFS chains is a spatial location game, not a speed race — players must randomize
This is the theoretical foundation for spam MEV on L2s: low gas costs push searchers into the multi-transaction equilibrium.
2-Prop: A Mechanism Design Solution
Rasheed, Desai, Gujar (IIIT Hyderabad, Feb 2026) propose 2-Prop: select two proposers per slot, each proposes a block. A reward-sharing policy incentivizes fast proposal.
Latency Game equilibrium:
- Homogeneous proposers: NE is to propose without delay (competition eliminates timing games)
- Heterogeneous proposers: faster proposer only delays if the slower proposer’s expected propagation time is near the attestation deadline (i.e., effectively no competition)
This is the first game-theoretic mechanism design solution to timing games, complementary to the empirical/engineering approaches (AdaptiveSwitch, PoT).
Open Questions
- What is the empirically correct V* threshold for timing game emergence on mainnet?
- Can PoT be deployed as an off-protocol relay enhancement, or does it require protocol enshrining?
- How do AI agents change the equilibrium dynamics of timing games over time?
- Does sub-slot execution with blinded proposers actually prevent timing games, or just shift them to sub-slot level?
- Can priority fee ordering on L2s reduce the FCFS spam equilibrium?
Related Pages
- PBS and MEV-Boost — PBS architecture where timing games arise
- Relay Block Merging — Post-auction merging that may interact with timing
- ePBS: Enshrined Proposer-Builder Separation (EIP-7732) — Protocol-level fix for timing games
- Latency and Validator Revenue — Latency’s impact on validator rewards and builder advantage
- Spam MEV: Probabilistic Block Space Extraction on High-Throughput Chains — Probabilistic backrunning leading to spam MEV on L2s
- Paper: Timing Games — Probabilistic Backrunning and Spam (Flashbots/Offchain Labs) — Mazorra et al. paper digest
- Paper: Let Leaders Play Games — Improving Timing via 2-Prop — 2-Prop mechanism paper digest
Key Sources
- Proof of Time at EthCC: Trust-Based to Physics-Based (2026) — Nash equilibrium; V* threshold; AdaptiveSwitch
- What Emerged from the Blockspace Forum Workshop in Cannes (Apr 2026) — sub-slot execution; TOOL/NuConstruct; proposer blinding
- Optimizing a $100B Market (2026) — APR uplift from latency reduction; mump2p RLNC
- Timing Games: Probabilistic backrunning and spam (Mazorra, Schlegel, Mamageishvili, Flashbots/Offchain Labs, Feb 2026) — FCFS timing games; spam equilibrium
- Let Leaders Play Games: Improving Timing (Rasheed et al., IIIT Hyderabad, Feb 2026) — 2-Prop mechanism; Latency Game Nash equilibrium