Briefing
The foundational challenge in State-Machine-Replication (SMR) protocols is achieving low-latency consensus across geographically distributed systems without compromising the core guarantees of safety and liveness. This research introduces Minimmit, a Byzantine-Fault-Tolerant (BFT) SMR protocol that fundamentally separates the quorum requirements for two critical operations → view progression (leader rotation) and transaction finality (committing a block). By allowing the view progression mechanism, which dictates throughput, to operate on a distinct, lower quorum threshold, Minimmit significantly reduces the communication overhead associated with leader changes. This architectural separation yields a measurable reduction in transaction latency, establishing a new state-of-the-art for fast finality and enabling a more responsive, globally scalable blockchain architecture.
Context
Prior to this work, the prevailing theoretical model for achieving 2-round finality in BFT protocols, such as those operating under the $5f+1 leq n$ fault assumption, mandated a single, high quorum threshold for all critical operations. This design choice, while guaranteeing consistency, imposed a significant and unavoidable communication cost. In a globally distributed network, this monolithic quorum requirement translates directly into high transaction latency and slow view changes, as the protocol must wait for a supermajority of nodes across vast distances to communicate and agree on every step. This established limitation created a bottleneck, forcing developers to trade off global decentralization for speed, or conversely, accept high latency for robust security.
Analysis
Minimmit’s core mechanism is the introduction of dual quorum thresholds, which decouples the protocol’s liveness mechanism from its safety mechanism. In previous BFT systems, the quorum needed to finalize a transaction was identical to the quorum needed to progress to a new leader (view change). Minimmit maintains the high, robust quorum for the safety-critical transaction finality step, ensuring the integrity of the ledger. Simultaneously, it introduces a separate, lower quorum for the view progression mechanism.
This lower quorum is sufficient to enable the network to quickly agree on a new leader and proceed with block production, which drastically reduces the time the network spends waiting for cross-continental communication during leader failure or network instability. The logic is that faster view changes improve liveness and throughput, and the system’s safety is preserved because the transaction finality still relies on the original, higher quorum.
Parameters
- Transaction Latency Reduction → 10.7% reduction in transaction latency compared to the state-of-the-art.
- View Latency Reduction → 23.1% reduction in view latency, directly impacting network throughput.
- Fault Tolerance Assumption → $5f+1 leq n$, which is the number of processors $n$ required to tolerate $f$ Byzantine faults.
Outlook
This theoretical breakthrough opens a new line of inquiry into the optimization of BFT communication complexity, moving beyond the traditional assumption of unified quorum structures. In the next 3-5 years, this dual-quorum design is expected to be integrated into high-performance Layer-1 and Layer-2 consensus protocols, enabling a new generation of decentralized applications that require instantaneous confirmation times. The research provides a blueprint for constructing globally distributed, high-throughput networks where the speed of block production is no longer strictly bound by the most conservative safety threshold, thereby resolving a critical architectural tension in current blockchain design.
Verdict
The Minimmit protocol introduces a fundamental optimization to Byzantine consensus theory, proving that liveness and safety can be governed by distinct quorum thresholds to achieve superior latency.
