Research profile Zhuo Cai

Research

Three research thrusts and representative papers.

My research spans three interconnected directions in cryptography, distributed systems, and blockchain economics. Prior and earlier work follows below.

My research builds the cryptographic and game-theoretic foundations that make decentralized protocols trustworthy at scale. I work across three directions: (1) proof systems — constructing efficient, updatable SNARKs and lookup arguments for verifiable computation over evolving data; (2) cryptographic coordination protocols — designing bias-resistant, incentive-aware randomness and leader-election protocols secure against strategic manipulation; and (3) blockchain mechanism design — analyzing and repairing transaction fee mechanisms for modern architectures including DAG consensus, sharded execution, and parallel transaction processing. My work appears at CCS, IJCAI, OOPSLA, ICBC, and IEEE Blockchain. I am a PhD student at HKUST, supported by the Hong Kong PhD Fellowship (HKPFS), and a former visiting scholar at the University of Oxford.

Work in Progress

Ongoing directions and papers under review.

These include manuscripts currently under submission and active research directions without a published paper yet.

IEEE DAPPS 2026 Under submission

LockFee: Refundable Transaction Admission via Time-Locked Collateral

J. Ballwegα T. Barakbayevaα Zhuo Caiα A. Goharshadyα α Alphabetical order

IEEE International Conference on Decentralized Applications and Infrastructures · 2026

Under review at IEEE DAPPS 2026.

Irreversible transaction fees create a persistent revenue flow from users to validators, fueling MEV extraction, specialized builder markets, and the economies of scale that progressively centralize block production. We study LockFee, a principal-preserving admission primitive in which each accepted transaction locks collateral for a fixed duration and recovers the principal in full afterward; users bear only the opportunity cost of locked capital. Under a simplified equal-size-transaction model, we prove that sustaining full admission at throughput q with lock amount L and lock duration D requires aggregate locked capital of order qLD. A baseline protocol pairs this admission rule with inflation-funded validator rewards, achieving validator participation and self-spam resistance without any irreversible fee flowing from users to validators, thereby reducing centralization pressure. As a downstream effect, LockFee screens users by heterogeneous opportunity cost of capital rather than by a uniform irreversible payment.

transaction fee mechanismblockchain economicsincentive designMEVdecentralization
Summary
ePrint 2025 Manuscript

Strategic Mining in Proof-of-Stake with Practical Random Election

Zhuo Cai

IACR Cryptology ePrint Archive · 2025

IACR ePrint 2025/1428.

This manuscript studies strategic mining in proof-of-stake systems when random beacon output is fixed across multiple future slots within an epoch. It formalizes the lookahead effect created by practical random election and gives an efficient optimal attack algorithm showing profitable deviations at much lower stake fractions than earlier per-slot-randomness models.

proof of stakestrategic miningdistributed randomnessblockchains
Local PDF
Preprint
Research Direction Active

Blockchain Mechanism Design: Transaction Fees and Strategic Equilibria

Most transaction fee mechanism (TFM) theory—including the foundations of EIP-1559— assumes a linear chain with one proposer per block in isolation. Real blockchain infrastructure has moved well beyond this model: strategic builders can delay transactions across multiple blocks, DAG-based consensus (Bullshark, Shoal++) runs concurrent proposers across overlapping transaction sets, sharded systems must price cross-shard atomicity, and parallel execution engines (Solana, Monad, Sui) remove ordering constraints while introducing new strategic manipulation via conflict graphs. This direction asks which classical TFM desiderata survive these architectural changes and what new mechanisms are needed to restore incentive compatibility, user welfare, and revenue adequacy. It also studies incentive design for parallel execution: whether conflict injection can be detected and penalized, how MEV manifests under parallelism, and what validator compensation models are viable when transaction ordering is no longer the primary strategic lever.

Related publications

transaction fee mechanismblockchain mechanism designincentive designblockchain economicsDAG blockchainshardingparallel executionMEV

Current

Proof Systems and Verifiable Computation

The core challenge is achieving succinct verification while supporting rich query structure and practical update patterns. My work uses updatable lookup arguments and related proof components to bridge theory-heavy ZK ideas with real systems. The Rogue paper (CCS 2026) advances updatable matrix lookup arguments and connects directly to verifiable database applications, pointing toward a broader framework for dynamic SNARKs over evolving state in rollups and verifiable infrastructure.

Context

Why this direction matters

This is the technical core of my research agenda. It points toward cryptographic interfaces that let users trust results from outsourced computation and data services without trusting the provider, a primitive that becomes more important as compute and storage move off-chain.

Core ideas

Recurring technical ingredients

  • Updatable and dynamic lookup arguments for verifiable databases
  • Bridging ZK proof theory with concrete systems applications
  • Forward vision: dynamic SNARKs for rollup state transitions and verifiable infrastructure
CCS 2026 Accepted

Rogue: Updatable Matrix Lookup Arguments and Applications to Verifiable Databases

C. Pappas Zhuo Cai D. Papadopoulos

ACM SIGSAC Conference on Computer and Communications Security · 2026

To appear.

Rogue develops updatable matrix lookup arguments and studies how they can be used to support verifiable database applications, with an emphasis on proof-system expressiveness and efficient update handling.

zero-knowledge proofslookup argumentsverifiable databases
Summary

2022–present

Cryptographic Coordination Protocols

A recurring question in decentralized systems is how to achieve coordination guarantees when every participant is rational and no central authority can be trusted. My work addresses this at the cryptographic and game-theoretic level: from verifiable random functions and secret random number generation to smart-contract protocols that implement richer solution concepts without a trusted mediator. The unifying concern is robustness — bias resistance, incentive compatibility, and fairness — across adversarial and strategic settings. This is also where I study the strategic behavior of validators and miners in proof-of-stake systems, asking when rational participants deviate and how protocol design can close those gaps.

Context

Why this direction matters

This direction connects cryptography, distributed systems, and economics. The goal is not just secure randomness sampling, but coordination mechanisms that remain robust when participants are strategic and incentives diverge from protocol intent.

Core ideas

Recurring technical ingredients

  • Bias-resistant and gas-efficient decentralized random beacons
  • Leader election without trusted setup or centralized DRNG
  • Strategic mining analysis in proof-of-stake and its protocol implications
ePrint 2025 Manuscript

Strategic Mining in Proof-of-Stake with Practical Random Election

Zhuo Cai

IACR Cryptology ePrint Archive · 2025

IACR ePrint 2025/1428.

This manuscript studies strategic mining in proof-of-stake systems when random beacon output is fixed across multiple future slots within an epoch. It formalizes the lookahead effect created by practical random election and gives an efficient optimal attack algorithm showing profitable deviations at much lower stake fractions than earlier per-slot-randomness models.

proof of stakestrategic miningdistributed randomnessblockchains
Local PDF
Preprint
ICBC 2024 Published

SRNG: An Efficient Decentralized Approach for Secret Random Number Generation

T. Barakbayevaα Zhuo Caiα A. Goharshadyα α Alphabetical order

IEEE International Conference on Blockchain and Cryptocurrency · 2024

Alphabetical author order with Amir Goharshady.

SRNG presents an efficient decentralized approach to secret random number generation, aiming to provide tamper-resistant randomness while preserving the privacy properties needed by higher-level blockchain protocols.

distributed randomnessrandom beaconsblockchains
Local PDF
Preprint
ICBC 2024 Published

Gas-Efficient Decentralized Random Beacons

V. Abidhaα T. Barakbayevaα Zhuo Caiα A. Goharshadyα α Alphabetical order

IEEE International Conference on Blockchain and Cryptocurrency · 2024

Alphabetical author order with Amir Goharshady.

This paper develops decentralized random beacon protocols with explicit attention to gas efficiency, aiming to keep randomness services practical for on-chain use without weakening decentralization goals.

random beaconssmart contractsblockchains
Local PDF
Preprint
ICBC 2023 Published

Trustless and Bias-Resistant Game-Theoretic Distributed Randomness

Zhuo Caiα A. Goharshadyα α Alphabetical order

IEEE International Conference on Blockchain and Cryptocurrency · 2023

Alphabetical author order with Amir Goharshady.

This paper studies distributed randomness generation under strategic behavior and proposes a trustless, bias-resistant approach that aligns game-theoretic incentives with randomness quality.

distributed randomnessgame theoryblockchains
Local PDF
Paper Preprint
MARBLE 2023 Published

Game-Theoretic Randomness for Proof-of-Stake

Zhuo Caiα A. Goharshadyα α Alphabetical order

Workshop on Mathematical Research for Blockchain Economy · 2023

Alphabetical author order with Amir Goharshady.

This work studies randomness generation for proof-of-stake settings through a game-theoretic lens, focusing on protocol incentives and how strategic behavior shapes the quality of the randomness mechanism.

proof of stakedistributed randomnessgame theory
Local PDF
Paper Preprint
IEEE Blockchain 2023 Published

PureLottery: Fair Leader Election Without Decentralized Random Number Generation

J. Ballwegα Zhuo Caiα A. Goharshadyα α Alphabetical order

IEEE International Conference on Blockchain · 2023

Alphabetical author order with Amir Goharshady.

PureLottery studies fair leader election without relying on a decentralized random number generation service, showing how protocol design can still achieve fairness guarantees in blockchain settings.

leader electionfairnessblockchains
Local PDF
Preprint Code

2024–present

Blockchain Mechanism Design

Most fee mechanism theory — including the foundations of EIP-1559 — assumes a linear chain with one proposer per block in isolation. Real infrastructure has moved beyond this: builders can delay transactions across multiple blocks, DAG-based consensus runs concurrent proposers across overlapping transaction sets, and parallel execution engines remove ordering constraints while introducing new manipulation surfaces via conflict graphs. I study what classical TFM desiderata survive these architectural changes and what new mechanisms restore incentive compatibility and user welfare. My work on trustless sampling of correlated equilibria extends this to a richer question: how can smart contracts act as trustless mediators implementing solution concepts beyond Nash equilibria?

Context

Why this direction matters

This is the broadest and most forward-looking of my three directions. It addresses strategic design questions that become unavoidable as blockchain infrastructure scales toward DAG, sharded, and parallel execution environments.

Core ideas

Recurring technical ingredients

  • Transaction fee mechanisms under multi-block, DAG, and parallel-execution architectures
  • Smart contracts as trustless mediators for correlated equilibria
  • MEV and incentive design for next-generation blockchain infrastructure
IEEE DAPPS 2026 Under submission

LockFee: Refundable Transaction Admission via Time-Locked Collateral

J. Ballwegα T. Barakbayevaα Zhuo Caiα A. Goharshadyα α Alphabetical order

IEEE International Conference on Decentralized Applications and Infrastructures · 2026

Under review at IEEE DAPPS 2026.

Irreversible transaction fees create a persistent revenue flow from users to validators, fueling MEV extraction, specialized builder markets, and the economies of scale that progressively centralize block production. We study LockFee, a principal-preserving admission primitive in which each accepted transaction locks collateral for a fixed duration and recovers the principal in full afterward; users bear only the opportunity cost of locked capital. Under a simplified equal-size-transaction model, we prove that sustaining full admission at throughput q with lock amount L and lock duration D requires aggregate locked capital of order qLD. A baseline protocol pairs this admission rule with inflation-funded validator rewards, achieving validator participation and self-spam resistance without any irreversible fee flowing from users to validators, thereby reducing centralization pressure. As a downstream effect, LockFee screens users by heterogeneous opportunity cost of capital rather than by a uniform irreversible payment.

transaction fee mechanismblockchain economicsincentive designMEVdecentralization
Summary
IJCAI 2025 Published

Smart Contracts for Trustless Sampling of Correlated Equilibria

T. Barakbayevaα Zhuo Caiα A. Goharshadyα K. Keypoorα α Alphabetical order

International Joint Conference on Artificial Intelligence · 2025

Alphabetical author order with Amir Goharshady.

This paper gives two trustless smart-contract protocols for sampling correlated equilibria on blockchain, replacing the traditional trusted mediator with cryptographic techniques based on oblivious transfer and zkSNARKs.

game theorysmart contractsequilibrium computationblockchains
Local PDF
Paper

Prior and Earlier Work

Work predating the current three-thrust agenda.

These directions contributed to my research background and inform current work, but are not active primary thrusts.

2023

Smart Contract Resource Analysis

Prior work on automated static analysis for reasoning about gas consumption in smart contracts before deployment.

OOPSLA 2023 Published

Asparagus: Automated Synthesis of Parametric Gas Upper-Bounds for Smart Contracts

Zhuo Caiα S. Farokhniaα A. Goharshadyα S. Hitarthα α Alphabetical order

Proceedings of the ACM on Programming Languages · 2023

Alphabetical author order with Amir Goharshady.

This paper presents Asparagus, a method for automatically synthesizing parametric gas upper bounds for smart contracts so that developers can reason about cost behavior before deployment.

smart contractsstatic analysisformal methods
Abstract
Code
Earlier work

Multimodal Learning for Scientific Imaging

Earlier work on vision-language modeling for visual question answering on pathology images.

ACL 2021 Published

Towards Visual Question Answering on Pathology Images

X. He* Zhuo Cai* W. Wei Y. Zhang L. Mou E. Xing P. Xie * Equal contribution

Annual Meeting of the Association for Computational Linguistics · 2021

* Equal contribution.

This paper studies visual question answering over pathology images and explores how multimodal models can answer medically grounded questions in a scientific imaging domain rather than a natural-image setting.

machine learningmultimodal learningmedical imaging
Local PDF
Paper