Starknet 10,000-Character Research Report: Poised for Takeoff, Long-Term Visionaries Making Steady Progress
TechFlow Selected TechFlow Selected
Starknet 10,000-Character Research Report: Poised for Takeoff, Long-Term Visionaries Making Steady Progress
Explore Starknet's technical advantages in an accessible way and simplify the explanation of ZK proof processes.
Navigating Web3 tides with focused insightsAuthor: Anci, Core Contributor at Biteye
Editor: Crush, Core Contributor at Biteye
Compared to other noisy narratives in the web3 world, the ZK sector has long represented slow, tedious yet profoundly meaningful foundational construction.
Like a difficult spiritual practice, those inside struggle to find answers while outsiders remain confused and distant.
The good news is that ZK's development speed in recent years has far exceeded expectations. The two leading ZK Rollups—ZkSync and Starknet—have made tremendous progress in performance and cost efficiency.
After Ethereum’s EIP-4844 upgrade, ZK Rollups have gradually gained strength in their competition with OP Rollups.
Even more exciting is that former rivals StarkWare and Polygon Labs have collaborated to upgrade the STARK protocol—the longtime frontier of ZK technology—officially launching Circle STARK, marking another leap forward in ZK proof capabilities.
If you've read last year's article (Is L2 Summer Coming? A Complete Guide to StarkNet’s Technical Principles and Ecosystem), and want to better understand the fascinating ZK proof mechanisms behind Starknet but feel intimidated by daunting mathematical formulas and complex technical schools of thought, then this article will help you grasp some key questions about ZK.
We’ll try to avoid complicated math and instead focus on discussing Starknet’s technical advantages, especially its recent major breakthroughs.
01 Starting with ZK
ZK is both a label and an abbreviation for Zero-Knowledge Proof Systems.
As today’s hot topic, ZK proofs seem almost mythical—proving a fact without revealing any additional information.
How can such an ideal goal be achieved? Let’s use a familiar student scenario as an analogy.
Typically, a student wanting to prove academic excellence can simply show their transcript. Assuming the exam system is fair and reliable, a weighted GPA of A usually suffices to validate academic ability, without needing to disclose specific coursework.
The ZK proof process works similarly. In simple terms, it consists of two core components: the Prover and the Verifier.
The Prover acts like a school examination system, following a fixed procedure to generate a transcript representing academic capability, which is then presented to the Verifier—such as parents or employers—who evaluate the student’s competence based solely on the transcript.
From this, we see that the most challenging part of the entire proof process lies in how the Prover generates the proof. Specifically, ZK proofs consist of two stages: arithmetization and polynomial commitment.
1.1 Arithmetization
-
Arithmetization refers to converting complex proof problems into algebraic ones—in particular, transforming what we call a witness (Witness) into a set of polynomial constraints (Polynomial Constraints). This is similar to how a student's academic ability is converted through exams into a set of numerical scores.
-
Witness: A "witness" refers to off-chain computation data such as transaction records, account state data, and intermediate computational results—data used to prove transaction validity that one wishes to keep private.
-
Polynomial Constraints: These are conditions that polynomials must satisfy during the ZK proof process. The essence of turning complex problems into mathematical ones lies in finding a polynomial and proving that it indeed meets these constraints.
1.2 Polynomial Commitment
Polynomial commitment involves proving, within the concrete mathematical framework, that one has found a polynomial satisfying all the constraints generated during the prior arithmetization step.
If the polynomial proof is valid, the mathematical proof succeeds, confirming the truth of the original statement. This resembles calculating a final weighted average score or transcript that certifies all grades are As, thereby validating academic excellence.
You might object: in real life, a transcript often fails to accurately reflect true academic ability because human examination systems still contain too many loopholes and uncontrollable factors.
But in the world of ZK, leveraging precise mathematics and transparent programs, this ideal becomes achievable—just as smart contracts and blockchains ensure fairness and transparency.
02 SNARK vs STARK
SNARK and STARK are currently the two most widely used ZK proof protocols, forming the underlying foundations for ZkSync and Starknet respectively.
Due to their similar names and shared domain, they are frequently compared. But before diving into comparisons, let’s introduce two key figures to better understand the historical evolution of these ZK proof systems.
2.1 Groth and SNARK
Jens Groth is a professor in the Computer Science department at UCL (now emeritus), currently serving as Chief Scientist at Nexus focusing on zkVM.
Starting from 2009, he published prolifically each year on zero-knowledge topics. Many papers commonly referred to in the ZK field—like Groth09, Groth10—are named using his surname and publication year.
(Note: Since academic paper titles tend to be long and unwieldy, researchers often refer to landmark works using authors’ initials plus the year—for example, BBHR18, which laid the foundation for STARK, or PGHR13 used by Zcash. It’s rare for a single researcher like Groth to dominate naming conventions across multiple years.)
Among his most notable contributions:
-
【Groth10】Short Pairing-based Non-interactive Zero-Knowledge Arguments – introduced a complete non-interactive proof scheme, considered a theoretical precursor to SNARK.
-
【Groth16】On the Size of Pairing-based Non-interactive Arguments – streamlined proof size and improved verification efficiency over Groth10, remaining widely used today.
It was on Groth’s research that SNARK matured and evolved.
SNARK stands for Succinct Non-interactive Argument of Knowledge—a compact zero-knowledge proof system whose high usability enabled rapid adoption in cryptocurrency applications.
2.2 Eli Ben-Sasson and STARK
Notably, Eli Ben-Sasson—one of the co-founders of Zerocash, the first protocol applying SNARK to cryptocurrencies—later became a co-founder of StarkWare and one of the inventors of STARK.
Moreover, in earlier years, Eli Ben-Sasson actively promoted SNARK implementation, publishing papers in 2013 and 2014 proposing and refining SNARK constructions, enhancing practicality and efficiency, helping SNARK gain widespread attention and application.
However, perhaps due to deep understanding of SNARK’s limitations, in 2018 Eli Ben-Sasson et al. published 【BBHR18】Scalable, transparent, and post-quantum secure computational integrity, formally introducing the STARK proof system and offering a more robust solution for ZK Rollups.
STARK—Scalable Transparent Argument of Knowledge—excels when handling large-scale proofs, offers full transparency, requires no trusted third party, and provides quantum-resistant security.
(Note: While heroic narratives make compelling stories, no achievement is built by one person alone. Both SNARK and STARK are the result of collective scientific effort. Highlighting individuals merely adds vividness to this critical chapter of ZK history. Even someone as brilliant as Groth builds upon prior work such as KZG (by Aniket Kate, Gregory Zaverucha, Ian Goldberg). The authors behind STARK are equally accomplished, and we may explore them further in future articles.)
2.3 SNARK vs STARK
So why did Eli Ben-Sasson decide to abandon SNARK and build something new? What challenges does SNARK face?
2.3.1 Transparency
To answer this, we should first ask: what is most valuable in cryptography?
Satoshi Nakamoto answered: trust.
SNARK steps right into this minefield. During polynomial commitment, SNARK uses the KZG method, which requires a trusted setup (Trusted Setup) to generate a common reference string (CRS), used later to produce keys for proof generation and verification.
Returning to our transcript analogy: parents or employers accept an A average as proof of excellence only because everyone agrees that A > B > C > D. Only under this standard does an “A” carry meaning.
But what if the grading system were hacked, changing the order to C, A, B, D? Students who earned Cs would suddenly appear as top performers, causing misjudgment.
Thus, the security of this universally accepted standard is crucial. In the lawless “dark forest” of crypto, this Trusted Setup becomes a significant vulnerability.
Knowing this, why does SNARK persist with KZG? Because KZG produces extremely small proofs. Remember what the “S” in SNARK stands for? Succinct!
The allure of small proof sizes is immense. Before Ethereum’s Cancun upgrade, compact proofs gave SNARK superior practicality and efficiency, making it the preferred choice for many projects. Trade-offs exist everywhere.
In contrast, to tackle the challenge of non-trustworthiness, STARK adopts FRI (Fast Reed-Solomon Interactive Oracle Proofs) for polynomial commitment.
Specifically, FRI encodes the polynomial via Reed-Solomon coding, stores it in a Merkle tree, and enables multi-round interaction between Prover and Verifier through an Oracle, achieving verifiability and transparency (“T” in STARK).
(Note: This Oracle isn’t the centralized or semi-centralized oracle commonly seen in web3, but rather a decentralized virtual entity simulated locally by Prover and Verifier according to protocol rules—an interactive proof mechanism.)
Using our transcript analogy again, STARK’s polynomial commitment can be seen as a blockchain-based grading system, secured by blockchain technology to ensure fairness and transparency.
Additionally, STARK allows the Verifier and Prover to simulate interactions using a public random beacon, packaging everything into a single non-interactive proof, improving usability and asynchronicity.
2.3.2 Scalability
STARK also excels in handling large-scale, complex computations with generality and flexibility. Its average proof size decreases as scale increases, creating network effects—this is the “S” in STARK: Scalable.
Unlike SNARK, which uses R1CS-based circuit computation requiring custom circuit design per problem, STARK employs AIR (Algebraic Intermediate Representation)—a universal computing method linking states via transition equations, capable of abstracting nearly any computation into polynomial constraints.
Furthermore, because FRI uses recursive structures to progressively reduce polynomial degrees, proof size grows much slower than problem size (logarithmic growth), giving STARK clear advantages in large-scale computation.
Back to exams: if arithmetization is like taking tests, SNARK and STARK resemble traditional paper exams versus computer-based testing.
For short-term or small-scale use, paper exams are cheap and fast, whereas computer testing requires upfront software/hardware investment—seeming expensive and cumbersome.
But for large-scale global exam institutions, one computer can administer various types of tests. Question banks become richer and randomized automatically, saving massive manpower previously needed to write new exams.
Long-term, as test volume grows, hardware/software costs are amortized significantly.
2.3.3 Quantum Resistance
Beyond “S” and “T”, STARK achieves quantum resistance by using quantum-safe hash functions (e.g., Rescue hash, believed to be post-quantum secure, unlike SHA-256 which may be vulnerable to quantum attacks) and hard algebraic problems (complex algebraic challenges that even quantum computers cannot easily solve).
03 Circle STARK: More Than Just Incremental Progress
By now, it’s clear that SNARK serves as a necessary short-term, practical solution.
But looking ahead, as transaction volumes grow, computational complexity explodes, and people increasingly recognize that trust is actually the most expensive luxury in crypto, STARK’s superiority will become ever more apparent.
This trend is already visible in industry. ZkSync, a leading SNARK-based application, launched Boojum, beginning its gradual shift from SNARK to STARK.
Meanwhile, Polygon—known for innovation—has long pivoted to STARK. Their latest proof system Plonky3, released this year, is built on Circle STARK, jointly developed by Polygon Labs and StarkWare.
Circle STARK is a next-generation ZK proof protocol based on STARK. By ingeniously incorporating circle curves, it successfully applies the small prime field M31 to the proof system, dramatically boosting proof efficiency.
In ZK proof systems, prime fields play a vital role—proofs rely fundamentally on operations over prime fields.
Choosing a prime field reflects a balance between efficiency and security. Smaller fields require less computation and offer higher efficiency.
Larger fields typically imply stronger security, which is why both STARK and SNARK historically used large prime fields.
The innovation of Circle STARK lies in combining circular curves with the small prime field M31, achieving both enhanced proof efficiency and post-quantum security.
StarkWare has already open-sourced Stwo, the next-generation prover based on Circle STARK, expected to deliver 100x faster proof generation than the original Stone prover.
Stwo will maintain full compatibility with advanced Cairo. The current Starknet SHARP Prover (based on Stone) will eventually migrate to Stwo. Developers and users in the Starknet ecosystem will directly benefit from Stwo’s performance gains without any action required.
Beyond speeding up proofs, Polygon co-founder Brendan Farmer noted that Circle STARK will ultimately significantly reduce fees and expand applicability to more use cases.
Eli Ben-Sasson expressed optimism, calling Circle STARK a milestone—predicting the most efficient proof system will emerge soon, with continuous improvements and breakthroughs ahead.
04 Starknet Presses Forward: Performance Upgrades
From the above analysis, it’s evident that the STARK proof system—and its latest iteration, Circle STARK—is undoubtedly a cutting-edge force and rising star. As StarkWare’s flagship project, Starknet holds immense promise in the ZK Rollup space.
Yet, despite its potential, Starknet has faced prolonged criticism—notably around user experience and cost.
Fortunately, thanks to StarkWare’s ongoing efforts, these issues are becoming history. Below, we review Starknet’s recent major upgrades and upcoming roadmap milestones.
4.1 V0.12
Starknet Alpha v0.12.0, codenamed “Quantum Leap,” went live on mainnet in July 2023. This update focused on boosting network performance and improving user experience.
Through Rust-based sequencer optimization and Cairo language upgrades, Starknet significantly reduced block execution time. Throughput surged from 30,000 CSPS (Cairo Steps Per Second) in v0.11.0 to 220,000 CSPS—a massive performance improvement.
The long-standing UX pain point—users waiting up to 20 minutes for mainnet confirmation—is now obsolete.
Now, once confirmed on Layer 2, transactions are considered successful—reducing wait times to ~10 seconds and greatly enhancing user experience.
This milestone upgrade helped Starknet’s TVL surpass $100 million, with weekly growth exceeding 43%.
4.2 V0.13
Version v0.13.0, launched in January 2024, increased block size, cutting computational costs by 50% and data availability costs by 25%.
v0.13.1 deployed early support for Ethereum’s EIP-4844, enabling blob usage within hours of the Cancun upgrade—making Starknet the first L2 to significantly reduce user fees.
Later this year, per the roadmap, v0.13.2 will enable transaction parallelization, processing more transactions simultaneously to boost throughput and reduce latency.
v0.13.3 will integrate Cairo Native into the Starknet sequencer, further boosting sequencer performance and accelerating network speed.
4.3 V0.14 and Future Upgrades
According to the roadmap, the highly anticipated Volition feature is expected to launch in v0.14.0.
Currently, data availability (DA) storage on Ethereum consumes most of Starknet’s gas fees. Reducing DA storage on Ethereum is key to lowering costs.
Volition allows developers to store some data on Starknet L2 and submit only the state root to Ethereum L1. This drastically reduces L1 DA storage costs, further lowering fees.
v0.14.0 also plans to adopt Applicative Recursion—batch-processing multiple blocks’ L1 footprints (data and computation tasks required on Ethereum to support Starknet)—to further cut costs.
https://community.starknet.io/t/upcoming-feature-starknet-applicative-recursion/113868
Each Starknet block currently has a dedicated proof, requiring a fixed operational cost on Ethereum. Thus, the network often waits to accumulate sufficient transactions before bundling a block—leading to unpredictable block times and inefficient cost utilization.
With Applicative Recursion, provers can batch-prove several blocks together—shortening block intervals and sharing costs.
Additionally, Starknet will explore further DA compression techniques to reduce costs.
05 Ecosystem Development
5.1 Current State
With steady performance improvements and continuously decreasing fees, Starknet’s ecosystem has matured significantly.
On infrastructure, self-custodial smart wallets Agent X and Braavos not only ensure security but also support native account abstraction on Starknet, providing excellent UX as entry points to the web3 world.
For cross-chain bridges, native StarkGate is complemented by specialized projects like Orbiter Finance, MiniBridge, and rhino.fi.
Top DID project Starknet.id plays a role similar to ENS on Ethereum, allowing users to mint NFTs as identities and passports on Starknet.
In the competitive DeFi space, leading projects like Nostra, Ekubo, zkLend, ZKX, and Carmine Options have emerged rapidly capturing market share in DEX, staking, lending, and derivatives sectors, while innovating product features.
For instance, ZKX combines gamified UX with DAO governance to create a unique autonomous perpetual exchange;
Ekubo introduces a singleton design managing all liquidity pools via one contract, reducing friction;
mySwap’s one-click rebalancing function effectively mitigates impermanent loss during high volatility—injecting vitality into the ecosystem.
GameFi is a sector strongly supported by Starknet. Beyond Loot’s Realms—a strategic cross-chain game—there are street-style Dope Wars, space-exploration title Influence, and physics-based Topology built natively on Starknet—collectively known as the “Four Heavenly Kings” of Starknet GameFi.
SocialFi has seen xfam.tech emerge, akin to friend.tech, filling a gap in social applications.
Since the $STRK token airdrop early this year, activity on Starknet has visibly increased. Moreover, ecosystem projects like zkLend, Ekubo, and ZKX have launched native tokens $ZEND, $EKUBO, and $ZKX respectively.
Ecosystem leader Nostra Finance launched Starknet’s first native USD stablecoin $UNO and $NSTSTRK from staking $STRK.
This multi-layered token distribution has energized Starknet’s ecosystem, resulting in strong overall metrics in the first half of the year.
However, amid fierce competition among Layer1s and Layer2s, sustained vitality requires dual innovation in products and technology to create true breakout applications and elevate on-chain activity to the next level.
Beyond official teams, this demands continued developer community engagement—which explains why Starknet maintains extreme developer-friendliness and offered unprecedented developer rewards in its airdrop.
5.2 Challenges and Choices
As mentioned earlier, STARK was designed from the start for secure, large-scale, complex proofs—and so is Starknet.
To achieve this grand and pure vision, certain trade-offs were inevitable—one being Cairo language.
(Note: Cairo is a programming language specifically designed by StarkWare for the STARK proof system, enabling efficient proof generation and optimized off-chain computation, overcoming Solidity’s limitations in proof execution.)
Unlike other Layer2s using Solidity for smart contracts, developers on Starknet must use native Cairo, increasing learning curve and entry barriers.
Moreover, since Cairo VM isn’t EVM-compatible, many mature Ethereum projects cannot migrate directly to Starknet, depriving it of spillover benefits despite being an Ethereum L2.
Over 90% of dApps on Starknet are native, implying substantial development costs.
Faced with this dilemma, Starknet’s stance is clarified in Eli Ben-Sasson’s article from earlier this year: “Stubborn, or Steadfast.”
https://paragraph.xyz/@think-integrity/stubborn%2C-or-steadfast
Quoting the idiom “riding a tiger,” sacrificing security for short-term performance gains leads to endless trouble down the road.
True believers in technology never package second-rate tech as first-class.
What Starknet aims to deliver is genuine proof—proof resilient against massive data and the dark forest. Upholding proof means upholding security.
To uphold this principle, Starknet offers rich developer incentives—including hackathons and recently, a Seed Grant Program offering up to $25,000 USDC in non-dilutive funding to selected teams building on Starknet.
https://mp.weixin.qq.com/s/S-nVc60Sfyk2C2SP9WkNuQ
For gaming, the Foundation’s Propulsion Pilot Program will select up to 20 games, rewarding them based on gas consumption on Starknet mainnet, with up to $1 million per game.
https://mp.weixin.qq.com/s/ZFSTfMyLm60M8FMR8KOG8w
Additionally, Nethermind—the Ethereum client team in deep strategic partnership with Starknet—launched a $1 million Starknet Grant Program, offering up to $250,000 per project and technical support.
https://x.com/Starknet_ZH/status/1785159198868943151
Starknet is also working on bridging the gap with Ethereum. On one side, Nethermind’s Warp project translates Solidity code into Cairo, achieving high-level language compatibility.
On the other, Kakarot zkEVM—developed by StarkWare members—uses Cairo to simulate an EVM environment, aiming to create a provable EVM. The project remains in development.
5.3 The Future
Thanks to StarkWare’s efforts, the Cairo developer community continues to grow. As the community thrives, more excellent tools and products will emerge, attracting even more developers to join—creating a positive feedback loop.
Beyond traditional DeFi, we anticipate emerging trend areas on Starknet that could represent the future—driven by its steadily realized performance gains and computational integrity advantages.
5.3.1 Fully On-Chain Games
The concept of Fully On-Chain Games (FOCG), also known as Infinite Games, gained popularity early in blockchain’s rise, captivating gamers’ imaginations.
These games store all rules and data on-chain, execute gameplay via smart contracts, grant players true ownership of assets, ensure transparent and verifiable rules, and provide open economies—delivering freer, fairer gaming experiences.
But due to early limitations in throughput, cost, and interaction models, fully on-chain games remained a dream without mass adoption.
Now, with Starknet’s continuous optimization, we see strong potential for it to become fertile ground for FOCG.
5.3.1.1 Native Account Abstraction
Account Abstraction (AA) is key to improving UX and easing the web2-to-web3 transition.
Simply put, it shifts personal Ethereum accounts (EOAs) to programmable smart contract accounts (CA), simplifying complex steps while maintaining security.
Starknet embeds native AA by design—every account is a smart contract account—reducing implementation complexity.
Native wallet dApps like Agent X and Braavos support AA, delivering web2-like UX.
In gaming, AA shines even brighter. For example, session keys—developed by top Starknet game teams Briq, Loot Realms, and Topology—are already used in Loot Survivor, enabling wallet-free login (Session Wallet). Players no longer need to sign every “attack” action—greatly improving UX.
5.3.1.2 Maturing Cairo Ecosystem
With ongoing Starknet optimizations, numerous game communities have joined to build the Cairo ecosystem. With a series of infrastructure launches, the gaming framework is now largely established.
Dojo is a Cairo-based on-chain game engine, founded in February 2023 and now maintained by the community.
A game engine provides developers with a framework of contracts, tools, and libraries—eliminating the need to build basic systems from scratch, making on-chain game development easier.
(Note: The most popular Web2 engines are Unity and Unreal Engine. Well-known games like Hearthstone and Street Fighter are built on them. Web3 game engines are still nascent, with existing options including MUD (EVM-based) and Dojo (Cairo-based). Dojo is the first provable crypto game engine.)
Cartridge is a game launcher offering developers tools and libraries via Dojo for creating and deploying games—including on-chain logic, scalable architecture, seamless user access, modular development, and monetization.
For players, Cartridge simplifies discovery and access to favorite games.
5.3.1.3 Layer 3
As a general-purpose ZK Rollup L2, Starknet offers exceptional security, high throughput, and low fees. To meet specific needs, it also supports customizable Layer3 Appchain development.
Game developers can customize execution environments and consensus mechanisms to build high-performance, low-latency, low-cost dedicated chains—unlocking new possibilities for FOCG.
Realms and Cartridge announced cooperation early this year to jointly develop “Realms World L3,” expected to launch in Q3 2024, running the entire Realms ecosystem atop Starknet—offering faster speeds and lower fees, further improving UX.
Dope Wars also plans to collaborate with Cartridge on a Layer3
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@BiteyeCN
