Introduction
The fundamental performance of Proof-of-Work (PoW) blockchains, such as Bitcoin, is constrained by a rigid trade-off between security, throughput, and latency. To maintain a 50% adversarial resilience, Bitcoin's mining rate is intentionally throttled to minimize "forking"—a state where nodes mine blocks simultaneously, potentially allowing an adversary to overtake the honest chain. This restriction forces a low transaction throughput and high confirmation latency. Prism is a novel protocol that breaks this trilemma by "deconstructing" the traditional block into separate functional components, allowing the system to approach the physical limits of the network.
The Prism Architecture
Prism's core innovation lies in decoupling the various roles a single block typically performs: proposing data, voting on history, and carrying transactions. This is achieved through three distinct block types, assigned to miners via a process called cryptographic sortition.
- Transaction Blocks: These blocks carry the actual ledger data and are mined at high frequencies to saturate the network's capacity ($C$).
- Proposer Blocks: These form a tree structure that acts as the ledger's skeleton. They do not contain transactions but instead point to transaction blocks.
- Voter Blocks: Prism utilizes thousands of parallel "voter chains." These chains vote on which proposer block should be recognized as the "leader" at each level of the proposer tree.
By using many parallel chains, Prism applies the Law of Large Numbers across space rather than time. Instead of waiting for a single chain to grow "6 blocks deep," Prism waits for thousands of parallel votes, achieving high certainty in the time it takes for a single block to propagate.
Performance and Security
Prism is proven to be secure against an adversary controlling up to 50% of the total hashing power. Unlike previous attempts to scale PoW, Prism does not sacrifice security for speed. Its throughput is limited only by the actual bandwidth of the internet, achieving an optimal rate of $(1-\beta)C$, where $eta$ is the adversary's fraction of hash power.
As illustrated in Figure 1, Prism's confirmation latency scales with the network propagation delay ($D$) rather than the desired error probability ($\epsilon$). In Bitcoin, latency increases linearly with the security requirement ($\log 1/\epsilon$), whereas Prism maintains a near-constant latency $O(D)$ for a wide range of security parameters. This makes Prism "communication-limited" rather than "security-limited."
Conclusion
Prism demonstrates that by deconstructing the blockchain, it is possible to reach the physical limits of a network—capacity and propagation delay—while maintaining the gold standard of 50% adversarial resilience. While the current model assumes a synchronous network, future work will focus on extending this architecture to asynchronous environments, refining game-theoretic incentive structures, and exploring Proof-of-Stake adaptations.
Summary updated