On April 11, the "Over the Moon—Chasing Waves Web3" summit was held in Hong Kong. This event is hosted by MetaStone and Odaily, co-organized by MixMarvel and AWS, and supported by top partners such as CREGIS, Jasmy, HOPE, Souff l3, LD Capital US, ArkStream Capital, SevenX, Jubi, zkMe, etc.
The third roundtable theme of this conference is Future of ZK Convo. 6 guests discussed ZK technical difficulties and future ecological development. The following is a review of the text.
Moderator Wei Zhang: I am an industry account manager of AWS Web3. You should have heard that Amazon Cloud has supported a lot of users in the Web3 industry. You are welcome to continue to use Amazon Cloud to provide stronger and more reliable cloud support for Web3. Let's first ask a few guests to briefly introduce themselves, or your institution, and the recent situation of the project.
Yuxing: Hello everyone, I am Yuxing, and I am investing in sevenX now. I am more concerned about Layer 1, Layer 2, ZK, data and wallet fields. SevenX is the leading encryption fund in Asia. We encourage innovative ideas and adhere to the principle of co-construction Ideas to invest in applications that can change the industry, and entrepreneurs with ideas are welcome to contact us.
Warren Fang：Hello everyone, I am Warren Fang from Arkstream. Arkstream is a research-driven fund, mainly focusing on breakthrough innovations at the application layer and technology layer. ZK has been a direction we have been focusing on for a long time, and we pay more attention to the ZK-based ecology Or the scenario of zero-knowledge proof for large-scale user entry.
Tina Liu：Thank you for the organizer. I am Tian, and I am from Mina Foundation. I am now in charge of Japan, South Korea, India, Vietnam, Turkey, Russia and other places. We are actually a relatively long-standing public chain. The mainnet will be launched in 2021, and this year will be the mainnet In the third year of the line, we are recruiting developers from around the world in a large-scale development ecology this year, and successfully selected 14 projects, and we have also issued grants to them.
Shuyao Kong：Hello everyone, I am Shuyao from ConsenSys.
Leal Cheung：Hello everyone, my name is Leal. I used to do engineering research, and I am also a serial entrepreneur. I have mainly been doing technology-related research and development in the Crypto industry for the past two years.
Let me introduce our project. zkMe is a Web3 Credential Network based on zero-knowledge proof and MPC. We build protocols around user identities. We have three core cases, the first one is one face, one DID, which is different from the way Worldcoin uses hardware devices to collect irises to create DIDs, we use the face recognition function of mobile devices and fully homomorphic encryption to help users Create unique DIDs. Another case we have is zkKYC. We believe that some centralized solutions used by some DeFi decentralized applications are very inconsistent with the spirit of decentralization. zkKYC is very in line with the business needs of the de-centralized world. While maintaining privacy, it also ensures business compliance. We just launched the zkKYC solution recently, so if you are interested in zkKYC about DeFi or other decentralized applications, you can also contact us.
Everyone knows the importance of oracles to the blockchain. Chainlink pushes data such as prices to the chain through open APIs. Similarly, we also push data related to user identities to the chain. Data related to user identities include 1) Demographic Data, that is, the data on the ID card. 2) Identity Data, including Education Background, Working Experience, Credit Score and other data, and other types of data include Web2 behavior data, etc. We also hope to realize it through ZKP-based oracles. I also hope to develop more Web3 applications based on infrastructure such as identity oracles, thank you everyone.
Seabook：Hello everyone, my name is Seabook, I am from Starkware and I am a Starkware APAC Dev relationship. How many people know about Starkware? It is a company valued at US$8 billion last year. It has two major products, Starkex and Starknet. We have our own language, and everyone here can join our ecosystem. Thank you everyone.
Moderator Wei Zhang: In the introduction process, the previous guests mentioned that ZK is relatively popular recently. The premise of ZK’s development is driven by the Layer 2 ecology. May I ask all the guests present that ZK is still relatively slow in the development process Yes, where do you think its technical sticking points are? After analyzing these technical sticking points, what do you think it is in the Web3 industry or not in the Web3 industry?
Seabook：The first question is where is the difficulty? The first is the problem of the circuit. Writing a circuit requires the use of high-level functions. Generally, it can only be written by hand. This is a very slow and high-level process. For example, in the case of Scroll or ZKSync, they are all handwritten circuits, which is very cumbersome.
There are less than 1,000 top people in the world who know about ZK. Perhaps with the strong promotion of ZK Capital, this number will increase. What I want to express is that ZK itself is very difficult. Although the theory of ZK can be traced back to the 1980s, the real breakthrough is the real breakthrough in 2018 and 2017 with the continuous development of computers and hardware. Starkware also has a few very talented scientists. They invented a relatively high-level language called Cairo, which can be automatically translated into circuits.
The biggest application and integration point of ZK at present is the combination with blockchain and Crypto, but ZK can be applied to various aspects in the future.
Leal Cheung：Thank you for sharing. I don’t think ZK technology is stuck here that leads to the slow development of ZK, but because the industry demand has not been tapped, or it is precisely because of the continuous emergence of demand at this stage of development that new technologies will continue to develop and break through. solve these problems.
Everyone knows that the privacy issue of the blockchain has been accompanied since the first day of its birth. In the previous years of development, everyone was actually improving some efficiency and security issues, but at this stage everyone It is found that ZK can solve the problems of expansion and privacy, and can effectively promote the development of ZK technology. As the previous guests said, the problem of ZK development has prevented many developers from entering this field. Of course, some development teams have recently done some related visual circuit programming, which can be effectively converted into programming languages to achieve what we want. logic.
Execution efficiency is also a very big issue. There are many solutions for efficiency, such as ZKP compression or batch processing, or a combination of methods to achieve ZK execution efficiency.
There are also some about the development of ZK’s ecology. Now everyone pays more attention to ZK as a Rollup solution, but ZK can solve many problems in many scenarios. We are doing it ourselves and hope that ZK can solve the problems related to Identity. For privacy issues, the space for identity applications is very wide. As an application, it needs to be associated with identity, which can expand more possibilities.
In addition to these, I believe that ZK still has many opportunities for development. The second question is where is its future application space? In the traditional field, I believe that if I buy a medical insurance this year, for example, I don’t want to tell the insurance company my medical history, so ZK has a good solution. In fact, financial companies can also use ZK to do privacy calculations, such as finding When I open an account in a bank, I want to open a VIP customer, but I don’t want to disclose my assets in another bank to him. ZK is a very important privacy solution for information sharing.
Therefore, I believe that in the future, there will be many opportunities in both the application of private transfers and the development of projects related to Identity.
To give a simple example, everyone knows that Web3 GameFi hopes to acquire high-quality users with game-related experience. At this point, we can push the user's game data on the Steam platform to the chain through ZK, so that the game can be developed based on these data. For another example, we can push the credit score of the user in real life to the chain to realize Undercollateralized Loan. This opens up the next stage of development for Web3 applications.
Shuyao Kong：I will not repeat what the previous two guests said. ZK has two uses. You can think about the application scenarios yourself. One is capacity expansion and the other is privacy. But most of the challenges don't require ZK, because it is a very expensive way.
Tina Liu：Thank you for the host’s question. I very much agree with the views of the previous three guests. At present, ZK’s technical stuck point is mainly that the proof generated by ZK requires a lot of evidence. The original data and transaction process are uploaded to the chain through encryption, which leads to its The amount of calculation is very large. After you generate a zero-knowledge proof, you have to verify the proof, and after verifying the proof, you need to establish the proof. In fact, the three steps will generate a very large amount of calculation. Such a calculation amount It also leads to the low efficiency of its operation, which is the current stuck point encountered in the development of Mina.
As for the current applications of ZK, everyone generally knows about private computing, transactions, and multi-party computing. There is an interesting project, using ZK technology to achieve cross-chain interoperability. This is proposed by our team. They want to build a light settlement system on Mina. They are a quantitative team with high-frequency trading requirements. Very high, they read a lot of public chains that can achieve cross-chain fast to meet the needs of high-frequency transactions, and finally chose Mina, because Mina shared lightweight, which is a very interesting direction for the development of ZK, which is to achieve cross-chain For interoperability, we are also actively expanding to see if there are new applications.
Warren Fang：In fact, zero-knowledge has been developed for decades. Recently, the demand for capacity expansion has been brought up again. There are three main application scenarios of ZK. Shuyao talked about two, capacity expansion and privacy protection, and one is computing outsourcing. Capacity expansion is what everyone is most familiar with. Whether it is zkVM (StarkNet) or zkEVM (ConsenSys’ Linea, Polygon’s zkEVM, zkSync’s Era, Taiko, Scroll, and ZK’s lightweight public chain (Mina) are all expansion solutions. More calculations and transactions can be placed on the chain. In addition, using zkp to achieve privacy protection, such as zkMe, has also been explored in the Web3 industry.
We found that the difficulty of ZK-based projects is that there are not enough ecological developers, and the barriers for developers to enter are very high. I think there are three difficulties. First, the technical requirements for developers are relatively high. They need to have a certain background in mathematics and cryptography, and there are very few technical talents who are proficient in academic papers and engineering code practice. Second, academic papers are developing rapidly. However, the inconsistency of technology stacks in engineering and the silos of circuit languages led to the fragmentation and confusion of the ZK system in the early stages of technological development. The time overhead and computing cost are relatively high, and the resulting performance effects are not enough for large-scale application.
The combination of these difficulties makes it more difficult for developers to enter and build an ecosystem. We also know that there are many application scenarios and market demands in this area. We also look forward to seeing more good projects come out, thank you.
Yuxing：The previous guests have already spoken very comprehensively, so I will make a small summary and supplement. In my opinion, there are many reasons for the stuck points encountered by ZK, which can be summarized into three points:
First, it is difficult to do. Now ZK is a bit like artificial intelligence at the beginning. I have an algorithm background. When I first started learning artificial intelligence, I needed to manually write the underlying algorithm formula with C code. Later, Python has various functions and scaffolding to build it. I will write the algorithm and build it. The model is much faster, just call the function and apply the template directly, and then modify it. Now most ZK developers also want to build from the bottom, without good scaffolding. With the construction of scaffolding, ZK will become better and better, and this situation will change. This scaffolding includes language, function report, template and so on.
Second, it is difficult to use. One is that its generation speed is slow, and its verification speed is relatively fast. The slow generation speed includes algorithm problems and hardware problems. The algorithm is that many calculations will encounter the computing power of the underlying hardware, such as NTT and MSM. Removing these slow things can make it faster. The hardware is to optimize and solve those computing power card points, so as to better support the algorithm, just like Bitcoin's mining algorithm, so as to make it faster.
Third, it is very interesting that it is difficult to find a market fit. In fact, privacy transactions like zkKYC and ZK are very easy to achieve. There is a company in the United States that uses ZK to protect the privacy of users when they prove their identity. When the government needs to intervene, everyone can still restore the person’s information by using ZK. This thing is not pure ZK, and it is preserved to a certain extent Possibility of certain recovery information. When an application is looking for the most suitable solution for this market, it is not something that technical idealism can do. It will encounter many difficulties, which will lead to slower adoption and implementation, and affect the application to find a perfect market fit.
You mentioned the last new application scenario. In fact, several guests have just talked about it very comprehensively, so I won’t go into details.
Moderator Wei Zhang: Thank you for the sharing of the guests. We have seen that ZK is developing relatively slowly. There are still some bright things in combination with Crypto. The last question is to hear whether the guests are interested in bright eyes. In addition, do you think there are recent projects that you want to vote for on the ZK track, or what do you think are the new highlights of the ZK project?
Yuxing：I have indeed seen some new things recently. A guest just mentioned the functions of ZK, including privacy, expansion and verifiable calculation. Verifiable computation is about outsourcing computation off-chain, such as ZKML. Artificial intelligence algorithms consume a lot of computing power, and it is currently difficult to move the calculations to the chain. At present, even Starkware is difficult to support artificial intelligence running on the chain. Then when we use the centralized running algorithm, we can prove that the operation of the artificial intelligence algorithm has not been tampered with by adding ZK.
In addition, it can also prevent artificial intelligence fraud such as Deep Fraud. For example, now that AIGC is so powerful, it is difficult to doubt whether the words of a certain leader are falsified or true when they see a video broadcast on the news broadcast one day. It may be generated by deep learning, and we can give ZK It proves that the video played now is real, not from AIGC. In addition to videos, there are also photos. Some time ago, Professor Sun Yi's team also produced corresponding research. The photo of news photography can be embedded with ZK's proof at the moment it is taken, proving that the photo is a real record, not an AIGC scene.
Warren Fang：Because several guests on stage are very influential builders in ZK, I will not play tricks on their track classes. Some new innovations worthy of attention can be mentioned, such as the field of full-chain games. There was a Dark Forest game on Starknet before, which attracted the attention of many developers and the industry. They used ZK to realize the fog of war in the game. We also saw that many developers of full-chain games will use ZK to make some attempts.
I just talked about StarkNet, and we also saw a ZK machine learning project above. They made a product and tried to do AI things through ZK. There are also some new narratives, such as ZK's programmability, interoperability and scalability, plus the bridge and DID components implemented by ZK to realize the full stack of ZK. For privacy, ZK is used for data indexing and calculation verification to achieve a minimum level of trust, including zkMe, which is also based on ZK to do some DID or privacy protection attempts. We think these new possibilities to explore privacy protection and large-scale application scenarios, and to use ZK to implement functional applications that cannot be realized on the chain, we find it very interesting, and hope to discuss and share more with you. These are not related to interests and are just for discussion.
Tina Liu：I have already shared it just now. We are now an ecologically more interesting project. Let me mention it again. Our Mina is indeed developing a ZK language. Developers can come to program ZK. We welcome everyone to come to Mina for development.
Leal Cheung：Thank you for the sharing of the guests just now. Our own team has thought about how to achieve decentralization for a long time when we are working on the zkKYC solution. Under the circumstances, the Unlock of Raw Files can also be realized in a decentralized way.
We are also doing the ZK machine learning mentioned by the last guest. There is a reason for us to do this. Doing zkKYC is restricted by the policies of many countries, and the centralized database of their residents is not open. We have to adopt the traditional KYC method, that is, face recognition and OCR, which must ensure the privacy and security of users. ZK-based machine learning can be done off-chain, and user data can also achieve privacy and efficiency through ZK, which is a very important balance.
I usually do technology-related research myself, and I also look at innovations based on ZK. Just now, the guest mentioned that there are some social applications. Users do not need to know what is behind the technology, but they must know that users have been authenticated. , we have some users who use fake identities, and if they pass ZK verification, it is also a very important social way.
Of course, there are many innovations of all kinds. In fact, as long as you open up your imagination and continue to explore and discover, there are still many application opportunities, thank you.
Seabook：Finally, let me briefly talk about it. I personally think that all the bosses here should vigorously build ZK education. This education is far from enough. We are always chasing how to make applications, what is the underlying ZK, and how advanced the logic behind it is. I don't have much impression of Mathematics, everyone is that ZK is very hot, and a lot of capital is chasing it. I hope that everyone can settle down, or there is a Chinese version and an English version of the MOOC in China, so that ZK education can be really implemented, so that people in various industries can know that ZK can do these things, and combine their own background and practical application. Realize these applications and scenarios.
Moderator Wei Zhang:Thank you for the sharing of the guests. The points shared by the guest just now are also very good. In the whole development process of ZK, more Builders are really needed to participate, and inject some new ideas into the track of ZK to promote the development of the track. , and some popular knowledge points should be released to let more people get in touch with this track.
