Ethereum consensus model
A transaction waits an expected 6 sec for inclusion and an independent half-round wait of 6 sec before the block becomes input to the finality gadget. Together those waits take 12 sec on average. Then 2 voting rounds of 12 sec each. The first completed voting round marks justification. Expected finality arrives in 36 sec.
128,000 validator identities remain, with a modeled round of 1 slots, or 12 sec, at a total modeled gain of 4×.
Selected point: approximately 12 sec full-set round
DC contributes an optimistic 4× gain, bundled propagation 2×, and source batching 1–2× depending on validator count. Selected gains multiply in this simplified scenario. Slot length never changes the raw finality-vote throughput.
The validator count is a manually refreshed network snapshot used only as a chart marker. The model uses 1.1M as a rounded historical-high workload scenario, not an exact record. Proposal wording was checked against the local EIP and EthResearch corpora.
Proposal tracker
Local EIP snapshot 2026-07-23 · reviewed 2026-08-19Raises the maximum effective balance to 2,048 ETH while retaining the 32 ETH minimum activation balance, allowing operators to consolidate stake into fewer validator identities.
Enables movement toward lower validator counts on the x-axis; it does not change round time at a fixed validator count. Consolidation is opt-in, and validator identities are not a measure of independent operators or people.Creates a separate consolidation churn limit and proposes roughly doubling consolidation throughput.
Changes how quickly the network could arrive at a lower validator count, not the round-time curve at that count. The proposal trades faster validator-set reconfiguration against a shorter weak-subjectivity period and remains under review.Makes slot duration configurable and currently specifies an eight-second placeholder, down from twelve seconds.
Changes expected inclusion time and the slot-boundary quantization of finality. It does not increase full-set finality-gadget attestation throughput. Eight seconds is a draft placeholder. Shorter slots can require a voting cohort to pipeline across slot boundaries; that timing is modeled separately from throughput improvements.Lets validators scheduled in the same committee publish one pre-aggregated attestation instead of many individual messages.
Uses an optimistic 2× source-side gain at 500,000 validators and above, tapering linearly to no gain at 128,000 validators as consolidation removes overlapping work. The gain is a modeling assumption, not a measured network-wide result. Its taper avoids treating source batching and validator consolidation as independent, multiplicative savings.Uses gossipsub partial messages to deduplicate attestation data during subnet propagation, with a stated upper-bound bandwidth saving of up to 50%.
Models the stated upper-bound bandwidth saving as a fixed, optimistic 2× full-set attestation-throughput multiplier. Up to 50% less subnet bandwidth does not establish 2× round throughput; the curve is explicitly optimistic and should be validated with measurements.Separates a fast available-chain committee from a trailing full-set stabilization and finality pipeline so each cadence can be optimized independently.
Models an optimistic 4× full-set voting gain by pipelining finality-gadget attestations across the slot and aggregating in the following slot. The 4× case assumes X = 0: validators vote immediately after fast confirmation, maximizing the propagation window. A later X trades latency or propagation budget for more available-chain stabilization.Uses a finality gadget that can finalize in one full-set voting round under its stated optimistic or adversarial assumptions.
Changes the finality formula from two completed voting rounds to one; it does not make an individual round faster. One-round constructions have distinct thresholds, recovery behavior, and safety assumptions. The toggle is a timing scenario, not a protocol commitment.From today to the north star
Decoupled consensus gives fast confirmation and finality separate lanes.
One visual step = one slot · finality flips at 100% · north star is illustrative.
Decoupled voting pattern
Each slot activates one cohort and samples a fresh AC committee from the full set.