Lingting 2020 | Ping An Yi Lu Tong Yi Yi Fan: Achieve privacy protection at the expense of blockchain value, meaningless and meaningless

On December 27th, the world's first blockchain year-end speech "Ling Listening 2020" was held in Hangzhou Canal Culture and Art Center. "Ling Ting 2020" takes "Sense of Certainty" as the theme, and is led by Tang Xia Ling, vice president / editor of Babbitt and the founder of Ling Ting Blockchain. Summarize 2019, look forward to 2020, look at the trend from the cycle, and correct the mentality in chaos, and find a way to the future.

Lu Yifan, general manager of Ping An Finance's Blockchain business, shared it from the perspective of industry practitioners. His topic was "Blockchain Privacy: Full Encrypted Digital Network". Lu Yifan believes:

1.Trusted computing, hash on-chain, encryption on-chain and other privacy protection solutions all achieve privacy protection by sacrificing the value of the blockchain, which is neither valuable nor meaningful;

2. The value of the blockchain is to break the data silos, make the data traceable, penetrable, and verifiable. The privacy protection that is solved by "chain splitting" and "channels" is essentially equivalent to re-creating data silos, causing low efficiency. And cannot expand

3. It should be changed. The data and platform are already decentralized, so there is no need to pursue decentralization. Instead, the decentralized data and platform have a unified presentation. Data can be shared without sharing. Shared effect. 31821577523492_.pic_hd

The following is the full text of the speech:

The style of my speech today will be different, one is a bit brain-burning, and the other is to talk about real cases, hoping to vividly show the value of the blockchain, and by the way counterfeiting.

Before I was at IBM, I proposed an imagination to the boss and a colleague one day. Later, this imagination became a secret project. The secret project was first called "BlueChain", later changed to "OpenChain", "OpenBlockChain", and finally changed Everyone knows the name now, called "Hyperledge fabric".

Later I came to Ping'an and felt the "KPI culture", just like a great entrepreneur once said, "Anything that doesn't make money is a hooligan." Any technology must generate business value. Therefore, when we do one thing, we must first think about where the business value is. The same is true for blockchain. It cannot be just a new computer language or development framework for developers. The long-term "thunder and small rain" of the blockchain means that it has many problems to be solved, and we need to find it and solve it.


I start with value. Everyone knows the database, and the data in the database is used unilaterally. The blockchain uses a so-called consensus mechanism to synchronize the database to all nodes on the chain and ensure that the data contained in each node is the same. Why sync? Because it will weaken the various fraud and efficiency problems caused by data islands. When all participants add, delete, modify, and check are in a logically common data store, the data mismatch caused by data islands will disappear. In this way, financial service providers can obtain more real and valuable data.

For example, banks were unwilling or afraid to lend to small and medium-sized enterprises in the past because of too much fraud. For example, a company might get a credit after millions of credit loans from a bank, and then go to another bank to repeat the credit and continue tomorrow. The main reason that can cause this problem is that the data between the data platforms is not connected, and the data that is not connected is the data island. The commercial value of the blockchain is to break the silos of data and make the data penetrable, verifiable and traceable. In the financial field, its real value is to control the risk control of the financial market by logically integrating cross-platform and cross-subject data.

Since the blockchain can reduce the financing risk between enterprises and banks, why is there a thunderstorm and a little rain? There must be something wrong here. The first is data. In the past, my data was stored in my own database and I needed to worry about the risk of hacking and other data leaks. Now through the blockchain network, I want to synchronize the data to methyl ethyl propylene, how can my data be protected? How is my customer's privacy protected? In business society, data sharing is absolutely impossible. Even if data can be contributed, who can guarantee that their data will be as secure when stored in other commercial organizations as they are here.

Let's talk about smart contracts. Many people say that smart contracts can improve efficiency through automated execution, but what value does it have compared to traditional computer technology? For example, for example, Alice makes a transfer to Bob. In the traditional database, Alice can directly transfer to Bob. However, in the blockchain network, we need to make the data the same at each node after execution, so before and after the transaction, so The logic of the transfer transaction must be executed synchronously on all nodes through smart contracts. At this time, the value of smart contracts has also become its problem. Because the smart contract is ultimately a string of code, and the code needs data to execute. I have just said that data sharing is the ideal impossible state. Now I want to let others run the execution logic of private data with you. Is it possible? impossible.

We often hear a lot of blockchain use cases, and it ends up in two categories: one is depository, legal deposit, copyright deposit, bill deposit, and so on. But as a person who has been in the crypto industry for a long time, I can tell you that if your goal is to immutable data, digital signatures have already solved this problem 40 years ago. Why use a troublesome and costly What about blockchain? If you are worried about losing your digital signature, you can store it in the public clouds of several vendors. The other is that the blockchain can improve efficiency. For example, the paper process that used to take ten days can be solved in ten seconds after using the blockchain. Then I tell you, if you use a traditional database or point-to-point connection, it only takes ten milliseconds to solve it. The efficiency improvement of the blockchain is far worse than traditional technology.


Privacy security issues have actually been discovered long ago, so we now see many so-called privacy security solutions. But why is the blockchain still thunderous and raindrops small? Because most of the solutions we have heard so far are basically achieved at the expense of the blockchain to break the value of data silos.

Trusted computing, hash on-chain, encryption on-chain and other privacy protection solutions all achieve privacy protection by sacrificing the value of the blockchain. As anyone familiar with trusted computing knows, it is actually a more centralized solution than the cloud. Put the data in the cloud, you can also move it out by yourself, but once you rely on trusted computing, your data will always be locked by the key of the device provided by the vendor, don't try to take it out. If it is theoretically possible for banks to replace mainframes now, once you use trusted computing, don't even think about it because they have the keys provided by the vendors.

Once the data is hashed or asymmetrically encrypted, the data is written to the blockchain. What does this mean? No one but you can verify unless you tell him the data. Therefore, the hash on the chain basically degenerates into a high-cost blockchain-based digital signature solution.


Currently, the most widely used privacy protection scheme is called "chain division", and it is called "channel" in the hyperledger. What's the problem with it? Some people say that it can increase scalability. I don't dispute this, but it is definitely not a solution to privacy. why? For example, Alice, Bob, Carol, and David transfer money. If Alice and Bob use a smart contract to perform transfer transactions, you may find that Carol and David's books are not the same as them. Carol used to have 100, but now it's 50, because others don't know, I can duplicate my assets.

"Chaining" is to divide a large block into multiple small blocks, and then divide a unified ledger into multiple sub-ledgers. The value of the blockchain was originally to eliminate data islands. Now, data islands are re-created with sub-chains, because only you and your opponent can see the data. After the data island exists, the enterprise can fraud through different channels or sub-chains because the data is not connected. Why doesn't the data get through? To achieve the so-called privacy protection.


After discovering the various problems of the blockchain, we felt that we needed to make some changes. How did it change? We have adopted an extreme new framework called the "full encryption framework". In all blockchain networks of Ping An, all data is completely encrypted by the uploader before uploading. After the data in the database is encrypted, you can no longer see it. The data is indeed synchronized. Everyone sees a copy of the data when you add, delete, change, and check, but you cannot see the other party. Through the role-based decryption authorization, the data owner can direct different levels of decryption authorization for different parties, or establish authorization rules to complete automatic authorization.

We have put into production many applications, including the establishment of a trade finance network for more than a dozen multinational banks in Hong Kong, and a blockchain network for Tianjin Customs. For example, if an enterprise wants to sell 10,000 pairs of shoes for 100 yuan each, this is an order of 1 million yuan. First of all, you need to get financing from the bank to buy raw materials. He hopes to report more. In order to get more financing, he must report taxes to the customs when transporting. In fact, he wants to report less and pay less taxes.

Ping's full encryption framework, all data is encrypted using fields. For example, an order usually has 100-200 fields. Each field is encrypted with a separate key. We use cryptography to control participant B (logistics company) to see only logistics data, participant C (customs) to see logistics, amount and declaration signature, and participant D (bank) to see amount, unit price, and text. The amount of data that banks and customs can see is the same. This company must make a choice if it wants to raise capital, because what Dajia sees is a copy of the data. We can decide who can see what data, who can change what data, and who can manipulate what data through the control of cryptography. Several previous speakers said the importance of data, which is the most important thing for Ping An One Account. We must fundamentally solve the problem of privacy of all data in order to reflect the value of data.

I pause here for a moment and throw a question for everyone. Everyone is doing blockchain, and it is always said to be decentralized, but have you ever thought that today the data and platforms of different institutions are already decentralized, and the data is scattered in the databases of different platforms. If we use a different idea when using the blockchain, instead of decentralization, but let the data that has been decentralized have a logical and unified presentation, is it better and more ideal? Ping An One Account is doing this.


Some degree of data control can be achieved through cryptography, but it is far from enough. Because in many cases, especially in cross-border trade scenarios, data field-by-field encryption is not enough, because your data cannot be shown to the opponent at all. Therefore, we have introduced a 3D zero-knowledge algorithm system. In simple terms, it can do two things: one is the relationship of all encrypted data, that is, ">" "<" "=", which can be quickly verified; the second is that it can Do zero-knowledge homomorphism verification, and four arithmetic operations "+", "-", "*", "/". Without decryption, complete the whole process of calculation and verification in 3 milliseconds. At this point, we may have about 10,000 times higher performance than the algorithm that is closest to us internationally.

Performance is not a major issue yet, the key is what to do with it? For example, the relationship between the invoice and the order, the invoice is given by the seller, and the order is given by the buyer. In our full encryption framework, all data is encrypted, except that the invoice is encrypted by the seller and the order is encrypted by the buyer. Orders and invoices may be a one-to-one relationship or a many-to-many relationship. But no matter who is encrypted, the final invoice and order amount should be able to match. 3D zero-knowledge algorithms allow any third party to verify their matching characteristics without being able to decrypt them.

What to do when logistics companies come in? Also using the previous example, if all 10,000 pairs of shoes are shipped, one million invoices should be generated here. If I only ship 5,000 pairs of shoes, I should only be able to produce a maximum of 500,000 invoices. Why is this important? Because my invoice will eventually become an account receivable, and the account receivable will become an ABS product. The ABS product may be the underlying asset behind the wealth management product that everyone invests in. How can I make the securities dealers and ABS wealth management product buyers know and trace back, without letting the buyers and sellers publish the data because of the retrospective? We want to make the data not shared, but also achieve the effect of sharing, what should we do? This involves the verification of zero-knowledge isomorphism.

In simple terms, when the logistics information changes, it is a calculation process of the cipher text "+" "-" "*" "/". The current ledger increases the corresponding shipments. For example, increase the shipment of x pairs of shoes. Is the encrypted logistics quantity multiplied by the encrypted unit price equal to the amount on the invoice? If it matches, it means the invoice is real; if it doesn't match, there is a problem with the invoice, and later ABS and financial management have problems. This is true traceability and traceability. You can't use traceability and traceability as a reason, and let manufacturers that have nothing to do with it provide you with its data, because this is very important and very confidential business information.

For example, Hong Kong Trade Finance Network began applying a zero-knowledge cross-validation system in 2018. Suppose I have an order of 1 million, and pledged 500,000 to bank A for financing and another 700,000 to bank B for financing. 500,000 + 700,000 greater than the total amount of 1 million orders. But Bank A will not tell Bank B how much it has pledged the order, and it will not even tell Bank B that it has pledged the order. What should Bank B do at this time? It passes the zero-knowledge test to find out if the total amount of pledged by Bank B and all other banks exceeds the total order share. In this way, cross-matching and cross-validation of data can be achieved without knowing the data at all.


Let's talk about a case of import and export trade finance. The importer creates an order at Bank A, and the exporter verifies this order at Bank B. This is a cross-application order that is common in Hong Kong and other trade and financial networks. According to the process I just talked about, the enterprise financing to two banks. Through the blockchain network, we can allow the invoice created by the exporter on the platform of Bank B to be cross-matched with the invoice created by the importer on Bank A. Cross-platform authentication Data comparison. If the company goes to bank A for financing and then goes to bank B for financing, it will conduct early warning and detection of repeated financing.


Let me give another example. For the import and export business of the multinational company Alfa Semiconductor, the relationship between the buyer and the seller is the parent company, the subsidiary company, or it may be collusion between the two companies. Then the buyer and seller's verification of the order and invoice will not work. If companies use orders or invoices to finance, banks have no confidence because buyers and sellers can collude easily and at low cost.


Therefore, we verified the poor cross-chain data of the trade financing network and customs logistics network. What did we do with cross-chain? You don't need to worry about what each process does, you just need to pay attention to the fact that when cross-border trade occurs, data is formed simultaneously in the trade financing network and logistics data network. From purchasing, shipping, customs declaration, warehouse receipt generation, and release, a series of information is formed simultaneously in the two blockchain networks, of course, it is encrypted. What should a trade finance bank do at this time to finance the order?

Through Ping An's independent innovative 3D zero-knowledge cross-chain technology, we first took out the encrypted information of the Trade Finance Network and the encrypted information of the logistics network to match the encrypted information of commercial flow information and logistics. If the encrypted information can be matched, it means that the company's financing success rate is very high, and the bank will trust the company. This is not to say that it is impossible to counterfeit, but that after we integrate the data, the cost of construction becomes very high, because not only must we collude with buyers and sellers, but also we must collude with logistics and warehousing companies. This is difficult.

One last thing to explain. We have announced the throughput and latency of Ping An Yitong. In the case of 6-core 2.3Ghz, the average throughput of a single node is about 1.5, and the 24-core 2.3Ghz is about 50,000 TPS, and the maximum delay time is less than 0.01 second. Everyone says this is so high, what's going on? I can tell you the secret today. As early as 2015, we were discussing a question when we were at IBM. Whether the alliance chain needs blocks or not, although the blockless framework was not adopted because of PR, but later the IBM Zurich Research Institute Some of them also did blockless PoC on the super ledger Fabric? In fact, many of the newer public chains in the world now use blockless frameworks. Ping An's blockchain has two prominent differences: the first is a fully encrypted framework, and the second is a non-blocking framework. All transactions are processed in real time, which is why we have high throughput Under the circumstances, the effect of no delay can also be achieved.