Author: Biteye core contributor Anci
Compared to other noisy narratives in the web3 world, the ZK track has long represented a long, boring but meaningful infrastructure. It's like a difficult practice, with people inside struggling to find answers and people outside looking in the dark. However, the good news is that in the past two years, the development speed of ZK has far exceeded expectations, and the two ZK Rollup giants ZkSync and Starknet have made great progress in performance and cost. After the Ethereum EIP-4844 upgrade, ZK Rollup has gradually become stronger in the competitive landscape with OP Rollup. What's even more exciting is that former competitor StarkWare has cooperated with Polygon Labs to upgrade the STARK protocol, which has always represented ZK's cutting-edge strength, and officially launched Circle STARK, which has brought another leap in ZK's proof capabilities. If you have read last year's article (L2 Summer is coming? Master the technical principles and ecology of StarkNet in one article), and want to learn more about the wonderful ZK proof process behind Starknet, but are hindered by various daunting mathematical formulas and complex technical schools, then you might as well follow this article to understand some key issues about ZK. We will try to avoid the annoying mathematical part and discuss the technical advantages of Starknet on this basis, especially the major breakthroughs released recently. 01 Let's start with ZK ZK is a label and an abbreviation for Zero Knowledge Proof Systems. As a popular subject at present, ZK proof is like a mysterious legend-proving a fact without revealing any additional information. How to achieve such an idealistic goal? Here we need to draw an analogy with a scenario that is familiar to all students. Usually, the simplest way for a student to prove his academic excellence is to show a transcript. On the basis of ensuring that the examination system is effective and fair, a transcript with a weighted grade of A can often endorse the student's academic level without showing any specific academic content. The ZK proof process is also very similar. In simple terms, its core members have two parts: the prover and the verifier. The prover is like a school examination system, with a fixed process to generate transcripts for students as proof of their academic ability, and present them to the parents/company that serves as the verifier. The parents/company use the transcripts to verify the students' ability level.Here we can see that the most difficult part of the entire proof process is the process of the prover generating the proof. In the specific ZK proof, it can be divided into two parts - arithmetic and polynomial commitment. 1.1 Arithmetic Arithmetic is to convert complex proof problems into algebraic problems. Specifically, it is to convert the witness we want to prove into a set of polynomial constraints. This is similar to how we convert a student's academic ability into a set of scores through exams. Witness: Witness is what we usually call the raw data of off-chain calculations, including transaction data, account status data, intermediate calculation results, etc. It is the private data we use to prove the validity of the transaction but do not want to make public. Polynomial Constraints: Polynomial constraints. What needs to be done in the ZK proof process is to convert complex problems into mathematical problems, and the most critical part of the mathematical proof method is to find a polynomial and finally prove that you have found it. Polynomial constraints refer to the conditions that this polynomial needs to meet. 1.2 Polynomial Commitment Polynomial commitment means proving in a specific mathematical proof that you have found a polynomial that satisfies all the constraints generated by the arithmetic in the previous step. If the polynomial proof is valid, then the mathematical proof is successful, which means that the problem we want to prove is established. This process is similar to the weighted average score or transcript we finally get, which can ensure that the student's grades are all A, and then prove that the student's academic ability is excellent. However, you may question that in real life, a transcript often cannot accurately express a person's academic ability, because there are still too many loopholes and uncontrollable things in our human examination system. In the world of ZK, with the help of clear-cut mathematics and open and transparent procedures, this wish is being realized (just as smart contracts and blockchains guarantee fairness and transparency). 02 SNARK vs STARK SNARK and STARK are currently the two most commonly used ZK proof protocols, and are also the underlying protocols used by ZkSync and Starknet respectively.Because similar names and tracks are often compared. However, before the comparison, we might as well introduce two characters to better understand the ZK proof system built by these two protocols from the perspective of development history. 2.1 Groth and SNARK Jens Groth is a professor at the Department of Computer Science at UCL (now an honorary professor and currently the chief scientist of Nexus, zkVM direction). This person has been productive every year since 2009, publishing many zero-knowledge related articles. Groth09, Groth10, etc., which we often hear in the ZK field, are composed of his name and the year of publication. (Note: Usually, because the names of papers are long and difficult to pronounce, the academic community will use the name + year abbreviation to refer to some important papers. Common ones include BBHR18, which laid the foundation for STARK, and PGHR13 used by Zcash. They are composed of the initials of several authors and the year. It is rare for one person to have a full name like Groth, and to be able to record the years continuously.) The two most famous ones are: 【Groth10】Short Pairing-based Non-interactive Zero-Knowledge Arguments, which proposed a complete non-interactive proof scheme and is considered to be the theoretical pioneer of SNARK. 【Groth16】On the Size of Pairing-based Non-interactive Arguments, based on Groth10, simplifies the proof size and improves verification efficiency. It is still widely used today. It is based on Gorth's research that SNARK has been developed and improved. SNARK, the full name of which is Succinct Non-interactive Argument of Knowledge, is a concise zero-knowledge proof system. Its strong usability has enabled ZK to be quickly applied in the field of cryptocurrency. 2.2 Eli Ben-Sasson and STARK It is worth mentioning that the co-founder of Zerocash, the first protocol to apply SNARK to cryptocurrency, was none other than Eli Ben-Sasson, who later became the co-founder of StarkWare and one of the inventors of STARK. Not only that, in his early years, Eli Ben-Sasson actively promoted the implementation of the SNARK protocol, and published papers in 2013 and 2014, proposing and optimizing the structure of SNARK, improving its practicality and efficiency, and helping SNARK to truly gain widespread attention and application.However, perhaps because they are too familiar with the difficulties faced by SNARK, in 2018, Eli Ben-Sasson et al. published [BBHR18] Scalable, transparent, and post-quantum secure computational integrity. The STARKs proof system was formally proposed, and a more complete solution was proposed for ZK Rollup. STARK, Scalable Transparent Argument of Knowledge, has advantages in large-scale proofs, and the entire proof process is transparent, without relying on a trusted third party, and has security levels that are resistant to quantum attacks. (Note: It should be stated that although the stories of heroic epics are always popular, no achievement is achieved by one person. On the contrary, both SNARK and STARK are the result of the joint efforts of countless scientists. We pick out a single person to talk about it only to make this key development history of ZK more vivid from one aspect. Even for a great man like Groth, his paper can only be realized based on the research results of KZG and others (Aniket Kate, Gregory Zaverucha, Ian Goldberg), and the authors who proposed STARK are all skilled, and we can continue to dig deeper in the future.) 2.3 SNARK vs STARK So what was the reason that made Eli Ben-Sasso decide to start anew? What difficulties does SNARK face? 2.3.1 Transparency Before answering the above questions, we may have to answer another question first: What is the most expensive thing in the field of encryption? The answer given by Satoshi Nakamoto is - trust. SNARK just stepped on this minefield. SNARK uses the KZG method when making polynomial commitments, and the KZG commitment requires a trusted setup to generate a common reference string (CRS), which in turn generates the key for the proof and verification process. Let’s go back to our transcript example. The reason why parents or companies can judge that a student’s academic level is excellent when they see an average grade of A is because we all recognize that the ranking of academic ability from high to low is A, B, C, D. Only under this standard does the grade of A make sense. But what if the school’s grading system is hacked and the ranking of academic ability becomes C, A, B, D?Students who originally got C will be treated as top students and given priority. This leads to misjudgment. From this we can see that the security of this commonly recognized standard is crucial. In the crypto world that follows the Dark Forest Law, this Trusted Setup has become a huge hidden danger. Knowing this, why does SNARK insist on using the KZG method? This is because the size of the final proof obtained using KZG is too small. Remember what the "S" in SNARK stands for? Succient! The temptation of small size is too great, especially before the Ethereum Cancun upgrade, the streamlined proof size brings better practicality and efficiency to SNARK, and it has been accepted by more projects for a long time. So it is trade off everywhere. Back to STARK, in order to bite the hard bone of Non-Trusted, STARK adopted the FRI (Fast Reed-Solomon Interactive Oracle Proofs) method for polynomial commitment. Specifically, the FRI method encodes the polynomials with Reed Solomon, stores them in the form of Merkle trees, and completes multiple rounds of interactions between the validator and the prover through Oracle, thereby achieving verifiability and transparency (transparent, the "T" in STARK). (Note: It should be noted that the Oracle here is not the centralized or semi-centralized oracle we often see in the web3 world, but a decentralized virtual entity simulated locally by the validator and the prover according to the protocol rules, which is an interactive proof mechanism.) If we still want to use the example of the transcript, we can regard the polynomial commitment process of the STARK system as a scoring system built on the chain, and the fairness and transparency of the entire system are guaranteed by blockchain technology. In addition, in the STARK proof, the validator and the prover can also simulate the interaction process by using a public random beacon, and finally package it into a complete proof to achieve non-interactive proof and obtain better availability and asynchronicity. 2.3.2 Scalability STARK’s progress is also reflected in its versatility and flexibility in handling large-scale complex computing problems, and its average proof size will decrease as the scale of the proof increases, forming a network effect, which is what “S” stands for, Scalable. Unlike SNARK, which uses circuit calculation methods represented by R1CS for arithmetic, which requires redesigning circuits for different problems, STARK uses the AIR (Algebraic Intermediate Representation) method, which is a general machine calculation method that links different circuits through state transition equations. state, almost any computational problem can be abstracted into a set of polynomial constraints. At the same time, because the FRI method used by STARK uses a recursive structure to gradually reduce the degree of the polynomial when generating polynomial proofs, the growth of the proof size is much slower than the growth of the problem size (logarithmic level). Therefore, it has significant advantages when processing large-scale operations. Returning to the example of grades and exams, if our arithmetic process is compared to a test, then SNARK and STARK are like traditional test papers and computer-based tests respectively. In the short term or for a small school, traditional test papers are cheap and quick, while computer-based tests require preparation in software and hardware, which seems expensive and troublesome. But for large-scale global examination institutions, one computer can conduct examinations of different types and levels. The richness and randomness of the question bank no longer require teachers to prepare questions for each examination, saving a lot of manpower. In the long run, as the number of test takers accumulates, the investment costs of software and hardware will also be significantly diluted. 2.3.3 Resistance to quantum attacks In addition to the achievements of "S" and "T", STARK also uses quantum-resistant hash functions (such as Rescue hash, which is generally considered to be post-quantum secure, while the traditional SHA-256 hash function It is considered that it may be fragile in quantum computing), security algebraic problems (some complex algebraic problems that the prover needs to prove are currently considered difficult to solve on quantum computers), etc., to achieve resistance to quantum attacks. 03 Circle STARK is more than that. Having said that, it is not difficult to find that SNARK is an indispensable and fast and feasible solution in the short term. But as time goes by, as transaction volume increases, computational complexity explodes, and people become more and more aware that trust is actually the most expensive luxury in the encryption field, STARK’s superiority will become more and more obvious. The more prominent it is. This is also gradually revealed in the industry. Using the Boojum version launched by ZkSync, the head application of SNARKs, we have begun to explore the gradual transformation from SNARK to STARK. Polygon, which is known for its integrity, has also turned to STARK early. This year’s upgraded proof system Plonky3 is based on the latest Circle STARK jointly developed by Polygon Labs and StarkWare. Circle STARK is a new generation ZK proof protocol based on STARK update. Due to the clever introduction of Circle Curve, the small prime number field M31 has been successfully applied to the proof system, greatly improving the proof efficiency. In the ZK proof system, the prime number field plays a crucial role. It is by operating on the field of prime numbers that the proof is achieved. The choice of the prime field represents a balance between efficiency and security. The smaller the prime number field, the smaller the amount of operations required and the higher the efficiency. On the other hand, large prime number fields usually represent a higher level of security, which is why both STARK and SNARK used large prime number fields in the past. The innovation of Circle STARK is reflected in the use of a small prime number field such as M31 through a special combination of circular curves, which not only improves the proof efficiency, but also successfully ensures post-quantum security. StarkWare has currently launched and open sourced a new generation prover based on Circle STARK, Stwo. It is expected that the proof efficiency of Stwo will be 100 times that of the first generation prover Stone. Stwo will be fully compatible with Advanced Cario, the current Stone Prover-based Starknet Pover (SHARP Prover), which will also use Stwo in the future. At that time, developers and users of the Starknet ecosystem will directly benefit from the performance improvements brought by Stwo without having to perform any operations. In addition to improving the speed of proof, Polygon co-founder Brendan Farmer also mentioned that the application of Circle STARK will eventually significantly reduce costs and expand to more application proofs. Eli Ben-Sasson is even more optimistic that the launch of Circle STARK can be regarded as an important milestone. The most efficient proof system will be available in the near future, and more breakthroughs and improvements will continue. 04 Starknet continues to make efforts to improve performance. Through the above analysis, we can clearly see that the STARK proof system and its latest upgraded version Circle STARK are well-deserved cutting-edge forces and stars of tomorrow. As StarkWare’s son, Starknet is in ZK Rollup has a bright future. But perhaps it’s because all good things come hard, Starknet has been controversial for a long time. Reasons include experience and cost. Fortunately, through StarkWare’s continued efforts, these problems are becoming history one by one. Below we will review several recent important upgrades of Starknet, as well as further actions planned according to the roadmap. 4.1 V0.12 Starknet Alpha v0.12.0, codenamed Quantum Leap, will be launched on the mainnet in July 2023. The focus of this optimization is to improve network performance and optimize user experience. Throughput and latency are generally regarded as measures of network performance. By optimizing the sorter to Rust and upgrading the Cario language, Starknet's block execution time has been significantly reduced, and the throughput has soared from 30,000 CSPS (Cairo steps per second, Cario steps per second) in the v0.11.0 version to 220,000 CSPS. Performance has been greatly improved. The interactive experience problem that has been widely criticized has also been solved. The pending status that once lasted an average of 20 minutes in order to wait for the main network confirmation will become a thing of the past. After the client confirms in Layer 2, the transaction is considered successful, so the transaction time is shortened to about 10 seconds, and the experience is greatly improved. This upgrade, regarded as a milestone, helped Starknet’s TVL successfully exceed $100 million, with a single-week increase of more than 43%. 4.2 V0.13 The v0.13.0 version launched in January 2024 has expanded the block size, significantly reduced computing costs by 50%, and reduced data availability costs by 25%. V0.13.1 deployed support for Ethereum EIP-4844 in advance, so Starknet enabled blob functionality within hours of the Cancun upgrade and became the first L2 to significantly reduce user fees. In the rest of this year, according to the roadmap plan, V0.13.2 will support transaction parallelization, which can process more transactions at the same time, improve network throughput and reduce latency. V0.13.3 will integrate Cairo Native into the Starknet sorter to further improve the performance of the sorter. At that time, the network speed will be further accelerated. 4.3 V0.14 and subsequent upgrade plans According to the roadmap, the highly anticipated Volition is expected to be launched in the V.0.14.0 upgrade. Currently, Data Avaliablity (DA) on Ethereum consumes the vast majority of gas fees on the Starknet network, so reducing DA storage on Ethereum is the key to reducing fees.Volition allows developers to choose to store part of the data on Starknet L2, and finally submit the state root of this part of the data to Ethereum L1. In this way, the DA storage cost of L1 is greatly reduced, further achieving the purpose of reducing fees. V.0.14.0 also plans to use Applicative Recursion to batch process the L1 footprint (Layer 1 footprint, which is the data and computing tasks that need to be stored and processed on Ethereum to support Starknet operation) of multiple blocks at a time to reduce the cost. "https://community.starknet.io/t/upcoming-feature-starknet-applicative-recursion/113868" Currently, each block of starknet has a proprietary proof, and each block needs to spend a fixed operating cost on Ethereum. Therefore, the network often needs to accumulate enough transactions to share the block cost before packaging a block. This leads to uncertainty in block time and inefficient utilization of block costs. With the application of recursion, the validator can package several blocks together for proof, which shortens the block time and shares the cost. In addition, Starknet will also explore more DA compression solutions in technology to achieve the effect of reducing costs. 05 Ecosystem Construction 5.1 Current Situation With the steady improvement of performance and the continuous reduction of costs, the ecological pattern on Starknet has now become perfect. In terms of infrastructure, the wallet projects Agent X and Braavos, as self-hosted smart wallets, not only ensure security, but also adapt to the native account abstraction of starknet. As the entrance to the web3 world, it brings users a good interactive experience. In terms of cross-chain bridges, there are both native StarkGate, and Orbiter Finance, MiniBridge and rhino.fi, which focus on cross-chain bridges, have joined. DID's head project Starknet.id assumes the role of ENS on Ethereum, supporting users to cast NFT as identity and pass on the Starknet chain. In terms of the traditional red ocean Defi, leading projects such as Nostra, Ekubo, zkLend, ZKX, and Carmine Options have also grown on Starknet. While rapidly occupying the key business sectors of Dex, pledge, lending, and contracts, each Defi project is also working hard to innovate its products.For example, ZKX uses gamified interaction and DAO governance to create a unique autonomous perpetual contract exchange; Ekubo launched a singleton design that uses one contract to manage all fund pools, helping users reduce transaction friction costs; mySwap's one-click rebalancing function can effectively reduce impermanent losses when market fluctuations are too large, etc., all of which have injected more vitality into the ecosystem. Gamefi is a sector that Starknet officials have high hopes for. In addition to the strategic full-chain games represented by Realms in the head project Loot ecosystem, there are also street-style Dope Wars, strategic games based on interstellar exploration, Influence, and Topology based on physical knowledge created by the Starknet native team. They can be called the four kings of full-chain games on Starknet. In addition, xfam.tech appeared on Socialfi, similar to the previously popular friend.tech, filling the gap in the social field. Since the $STRK airdrop at the beginning of this year, the activity on Starknet has increased significantly. Not only that, projects within the ecosystem such as zkLend, Ekubo and ZKX have successively released native tokens $ZEND, $EKUBO and $ZKX; Ecological leader Nostra Finance has also released the first native USD stablecoin $UNO on Starknet and $NSTSTRK obtained by staking $STRK. The multi-level token issuance is undoubtedly a shot in the arm for the Starknet ecosystem. By the first half of this year, the overall data of the Starknet ecosystem was good. However, in today's Layer1 and Layer2 melee, in order to ensure continued vitality, dual innovations in products and technologies are also needed to create truly explosive applications, so that the vitality of the chain can steadily move to a higher level. Behind this, in addition to the official team, what is needed is the continued efforts of the developer community. This is also one of the reasons why the Starknet team has always been extremely developer-friendly and even gave unprecedented developer rewards in airdrops. 5.2 Challenges and choices As we said before, STARK was born for secure large-scale complex proofs from the beginning, and the same is true for Starknet, which is a continuation of the same lineage. In order to achieve this grand and pure goal, many efforts are inevitable, and Cairo language is one of them.(Note: Cairo is a programming language designed by StarkWare specifically for the STARK proof system. It can efficiently generate proofs, optimize off-chain calculations, and effectively make up for the limitations of Solidity in executing proofs.) Unlike other Layer 2s that use Solidity to develop smart contracts, developers must use the native Cairo language for development on Starknet, which directly increases the learning cost and entry threshold for developers. On the other hand, since Cairo VM is not compatible with EVM, many mature projects on Ethereum cannot be directly migrated to Starknet, which also makes it difficult for Starknet to enjoy the dividends of the big ecosystem even though it is an Ethereum Layer2. Currently, more than 90% of the dApps on the Starknet chain are chain-native, and the cost of project development is not small. In this dilemma, we can find the answer to Starknet's choice from Eli Ben-Sasson's article "Stick or Persist" at the beginning of the year. "https://paragraph.xyz/@think-integrity/stubborn%2C-or-steadfast" The article quoted the allusion of "riding a tiger and it is difficult to get off", saying that if you sacrifice security to pursue beautiful performance in the short term, it is like sitting on a galloping tiger with endless troubles. As a true believer in sticking to technology, we will never sell second-rate technology in first-rate packaging. What Starknet wants to do is to provide real proof that can withstand the monstrous data and the dark forest. And sticking to the proof is sticking to security. In order to stick to it, Starknet has a lot of incentives for developers. In addition to community activities such as hackathons, it has recently launched a seed funding program (Seed Grand Program) with real money. The selected teams will receive non-dilution funding rewards of up to 25,000 USDC to support their ecological construction on Starknet. "https://mp.weixin.qq.com/s/S-nVc60Sfyk2C2SP9WkNuQ" Specifically for the game sector, the Foundation’s current game support pilot program (The Propulsion Pilot Program) will select up to 20 games and provide funding rewards based on their Gas consumption on the Starknet mainnet. Each game can receive a maximum reward of 1 million US dollars."https://mp.weixin.qq.com/s/ZFSTfMyLm60M8FMR8KOG8w" In addition, Nethermind, the Ethereum client development team that has reached a deep strategic cooperation with Starknet, also released the Starknet Grand Program with a total of 1 million US dollars. A single project can receive up to 250,000 US dollars in funding and technical support from the Nethermind team. "https://x.com/Starknet_ZH/status/1785159198868943151" Starknet is also working hard to break through the barriers with Ethereum from two aspects. On the one hand, the Warp project developed by Nethermind is committed to converting Solidity code into Cairo code to achieve compatibility at the high-level language level. On the other hand, the Kakarot zkEVM solution developed by members of the StarkWare team uses Cairo to simulate the EVM environment and try to create a provable EVM. The project is still under development. 5.3 Future With the efforts of StarkWare, the developer community of Cairo is growing. With the prosperity of the developer community, the ecosystem will also incubate more excellent products and tools, further attracting more excellent developers to join the Cairo community and forming a positive cycle. In addition to expecting the birth of excellent products in traditional fields such as DeFi, combined with the performance improvement and computational integrity advantages gradually realized by Starknet, some trend sectors that may represent the future have gradually emerged on Starknet. 5.3.1 Full-chain games The concept of Fully On-Chain Games (FOCG, also known as Infinite Games) was popular as early as the rise of blockchain and was highly anticipated by gamers. It stores the rules and data of the game completely on the chain, executes all operations and interactions based on smart contracts, allows players to truly own the game assets, ensures the transparency and verifiability of the rules, and provides an open economic system to create a freer and fairer gaming experience for players. However, due to the limitations of early on-chain throughput, fees, and interaction modes, full-chain games have long been just a beautiful vision and have not achieved mass adoption. However, with the continuous optimization of Starknet’s performance in all aspects, we can see from the following aspects that Starknet has great potential to become a fertile ground for nurturing full-chain games.5.3.1.1 Native Account Abstraction Account Abstraction (AA) is a key step to improve the interactive experience and help web2 users enter web3. In simple terms, it is to transfer the EOA account on Ethereum that we used to use as individuals to a smart contract account (CA). Due to the programmability of the contract account, it can simplify the current complex operation steps and optimize the user experience based on pre-settings while ensuring security. Starknet's design has made native account abstraction a must from the beginning. Each account is a smart contract account, which reduces the complexity of implementing AA from the perspective of system design. Native wallets dApp Agent X and Braavos both support account abstraction, which can bring users the same experience as Web2. In the game context, the role of account abstraction will be more obvious. For example, the session key function initiated by several leading blockchain game teams on the Starknet chain, Briq, Loot Realms and Topology, is now used in the online Loot Survivor, which uses wallet-free login (Session Wallet). Users no longer need to sign for each "attack" operation, which greatly optimizes the experience. 5.3.1.2 Cairo Ecosystem Gradually Improved With the continuous optimization and performance improvement of Starknet, a large number of game communities have settled in to build the Cairo ecosystem together. With the launch of a series of infrastructure, the current game framework of the ecosystem has basically taken shape. Dojo is a Cairo-based on-chain game engine, created in February 2023, and is currently jointly operated and maintained by the community. The game engine is the basis for building games. It provides game developers with a development framework consisting of contracts, tools and code libraries, so that they do not have to build the basic system of the game from scratch, making it easier to build on-chain games. (Note: The two most popular game engines for Web2 games are Unity and Unreal Engine. The familiar Hearthstone, Street Fighter, etc. are built on the basis of these two game engines. At present, the development of Web3 game engines is still in its early stages. The two existing encrypted game engines are MUD built on EVM and Dojo based on Cairo.Dojo is the first provable cryptographic game engine. ) Cartridge is a game launcher that provides developers with tools and code libraries based on Dojo for creating and deploying games, including on-chain game logic, scalable architecture, seamless user access, modular development, and monetization solutions. On the other hand, it simplifies the interaction process for players, making it easier for them to discover and enter their favorite games. 5.3.1.3 Layer 3 Starknet, as a general-purpose Layer2 using Zk Rollup, has outstanding security performance, high throughput, and greatly reduced costs. In order to meet specific needs, it has also launched customized development of Layer3 Appchain. Game developers can customize and optimize the on-chain execution environment and consensus mechanism according to their own needs to create a high-performance, low-latency, and low-cost game-exclusive chain. This also brings more possibilities for full-chain games. Realms and Cartridge formed a partnership at the beginning of the year to jointly develop "Realms World L3", which is expected to be officially launched in Q3 this year. The entire Realms ecosystem will be run on the basis of Starknet, which will have faster speeds and lower costs, further optimizing the user experience. In addition, Dope Wars also stated that it will cooperate with Cartridge to release a Layer3 and use $PAPER as the Gas token. 5.3.1.4 Summary Although in the eyes of the public, Starknet's performance in Gamefi may not be outstanding, if you focus on full-chain games, Starknet's eco-friendliness, clear full-chain mission and its own technical advantages are very attractive to developers and players. We can see that game developers on Starknet are currently very enthusiastic about the construction of the Cairo ecosystem, and the top projects have worked together to facilitate the completion of key infrastructure such as Dojo. If one day full-chain games can be fully realized, more and more game enthusiasts will enter the game world of web3 and truly obtain ownership of game assets. We have reason to expect that all this will happen on Starknet. 5.3.2 ZKML With the outbreak of both AI and blockchain technology, AI+Blockchain is increasingly widely regarded as the future development direction, and ZKML (Zero Knowledge Machine Learning) is a solution.The traditional ML model is like a black box. Since it is owned by a centralized organization, ordinary users cannot verify what model they are using and whether it is trained with reliable data except for trusting the endorsement of the large organization. This is also one of the main reasons why centralized large models are criticized. If the model is simply put on the chain and trained and run in a decentralized way, not only will the gas cost be too high, but due to the non-trusted environment on the chain, the reliability of the model and training data cannot be guaranteed. In this case, ZKML is very necessary. Think off-chain, act on-chain. By training and running the model off-chain and using ZK technology to generate proofs and submit them to the chain, the two dilemmas mentioned above - cost and reliability - are perfectly solved. At the same time, due to the privacy characteristics of ZK itself, ZKML also has broad prospects in sensitive fields such as finance and medical care. There is a natural advantage in building ZKML based on Cairo on Starknet. As a development language born for proof, Cairo has excellent computational integrity and is highly abstract in proof. Developers can directly call proof components without having to deal with proof problems themselves, which greatly simplifies the development process. At the same time, thanks to the scalability advantage of STARK, it has a significant network effect when processing large-scale computing data, and can efficiently and low-costly carry the massive data required for machine learning. Therefore, ZKML based on Giza Tech on Starknet is developing rapidly. Giza is a ZKML middleware platform on Starknet, providing the Orion development framework, enabling developers to use familiar frameworks (such as PyTorch, TensorFlow) for model training, and easily deploy them on Starknet. At the same time, Giza also launched an agent framework Agents that combines ZKML with multi-chain behavior. Developers can create on-chain AI agents based on ZKML, interact with smart contracts, and make decisions according to predefined rules. At present, Giza has been applied in multiple projects, such as social networking, cooperating with Circles Network to conduct social graph analysis and detect fake users. In terms of DeFi, Yearn Finance cooperates to provide it with intelligent investment strategies and risk management solutions based on ZKML.ML Village, which was recently selected for the Starknet seed funding program, introduced ZKML into blockchain games for decision-making through Giza, showing broad application prospects. 06 Summary According to the recent official release of Starknet, it is expected that the gas fee can be reduced to far below $0.01 in 2024, and hundreds of TPS will be achieved, becoming the Layer2 with the highest TPS. This goal is not ambitious for Starknet. In fact, StarkWare's positioning of Starknet is far more than just a Layer 2. In terms of key product and market choices, Starknet is more similar to Solana, abandoning the dividends and constraints of EVM and building from scratch. However, unlike Solana's flaws in decentralization, Starknet cleverly inherits Ethereum's extreme decentralization through Ethereum Layer2+ZK, while ensuring scalability and security, making the impossible triangle possible. This seemingly happy result, as we said at the beginning, is a practice of adhering to long-termism and countless times of going far and carrying heavy loads. We look forward to Starknet releasing more vitality in the future.