"The epoch-and-slot architecture is clearly the right solution, but the specifics of its structure and slot implementation still require exploration."
— Vitalik Buterin
Introduction
One critical aspect of blockchain user experience is fast transaction confirmation times. While Ethereum has significantly improved over the past five years—thanks to EIP-1559 and transitioning to Proof-of-Stake (The Merge)—L1 transactions typically confirm within 5–20 seconds, comparable to credit card payments. However, achieving sub-second latency for certain applications remains a priority. This article explores practical improvements to Ethereum’s transaction confirmation times.
Current Approaches and Technologies
Single-Slot Finality (SSF)
Ethereum’s Gasper consensus uses a slot-and-epoch architecture:
- Slot: Every 12 seconds, a subset of validators votes on the chain’s head.
- Epoch: 32 slots (6.4 minutes) allow all validators to vote once.
These votes function like PBFT consensus messages, achieving finality after two epochs (12.8 minutes) with strong economic guarantees.
Challenges:
- Complexity: Interaction between slot-to-slot voting and epoch-to-epoch finality mechanisms introduces bugs.
- Long Wait Times: 12.8 minutes is impractical for users.
SSF Solution:
- Replaces Gasper with a Tendermint-like mechanism where block N is finalized before block N+1 is proposed.
- Retains the inactivity leak to handle offline validators (allowing chain recovery if >⅓ validators drop out).
Limitation:
- Validators must publish two messages every 12 seconds, creating high chain load (partially addressed by proposals like Orbit SSF).
- User Impact: SSF speeds up finality but doesn’t reduce the 5–20s wait for initial confirmations.
Rollup Pre-Confirmations
Ethereum’s rollup-centric roadmap delegates scalability to L2s (e.g., rollups, validiums), while L1 focuses on core security. However, L2s need faster confirmations than L1’s 5–20s.
Current L2 Options:
Decentralized Sequencer Networks:
- Validators sign blocks every few hundred milliseconds, with fraud proofs to penalize conflicts.
- Slow adoption due to complexity (akin to building a new L1).
Based Pre-Confirmations:
- Proposal: Use Ethereum proposers (MEV-aware entities) to offer pre-confirmation services.
- Users pay extra fees for instant guarantees of inclusion in the next block.
- Proposers are slashed if commitments are violated.
- Extends to "Based" rollups (where L2 blocks are L1 transactions).
The Epoch-and-Slot Architecture
Why Epoch-and-Slot?
Achieving rough consensus is faster than economic finality:
- Node Efficiency: Approximate consensus requires fewer nodes; finality needs majority participation.
- Optimization: Specialized subsets (e.g., high-quality nodes) can accelerate consensus (~2s possible).
Design Considerations:
- Slot Time: Could reduce from 12s to 8s (or even 1s with Orbit SSF).
- Separation of Concerns: Stronger division between fast consensus (slots) and finality (epochs).
Practical Implications for L2s
L2s have three strategic paths:
| Strategy | Description | Example |
|---|---|---|
| 1. "Based" Rollups | Align with Ethereum’s values (decentralization, anti-censorship). | Branded shards with VM experiments. |
| 2. Server-like Systems | Use STARK proofs for rule enforcement, combining blockchain benefits with server speed. | Centralized sequencers with fallbacks. |
| 3. Hybrid Models | Fast chains (~100 nodes) with Ethereum-backed interoperability. | Many current L2 roadmaps. |
Key Question: Can Ethereum-native epoch-and-slot achieve ~1s slots? If yes, hybrid models (Strategy 3) become less relevant.
FAQs
Q1: How does SSF improve user experience?
A1: It speeds up finality (from 12.8 minutes to 12 seconds) but doesn’t eliminate the initial 5–20s confirmation delay.
Q2: What’s the role of pre-confirmations?
A2: They provide instant guarantees (e.g., 500ms) by leveraging proposer commitments, usable by L1 and L2.
Q3: Why keep the epoch-and-slot model?
A3: It balances speed (slots) and security (epochs), optimizing for both approximate consensus and finality.
Q4: Will L2s always need custom solutions?
A4: Yes—especially for off-chain data systems (e.g., validiums), but Ethereum-native improvements can reduce reliance on hybrids.
👉 Explore Ethereum’s roadmap for deeper insights into L2 scalability.
Adapted from Vitalik’s original post.