BTC Exclusive Interview | Dr. Fan Long from Conflux: Cross-chain is a Subset of the Oracle, the Core is to Grant Real World Data Access to the Chain

12 May, 2019

On May 10th, Conflux Media Day, Conflux Founder and CEO Fan Long, CTO Dr. Ming Wu, and Research Director Dr. Guang Yang gathered together to introduce the recent development of Conflux and answered questions from everyone.

When I interviewed Fan in February this year, he told me that Conflux aims to accomplish what Ethereum was not able to achieve. In fact, Conflux is based on PoW and realize high security while ensuring high TPS without sacrificing decentralization. On April 4th, Conflux launched its test-net. According to the official data, the current TPS is stable at 700 or higher, significantly higher than other PoW-based projects, such as BTC (7TPS), ETH (15TPS), Grin (30TPS) and BCH (100TPS).

On the media day, I asked Dr. Long Fan whether everything that can currently be done on Ethereum can be done on the Conflux main-net? He said with great confidence: "Conflux can do everything that can be done on Ethereum, and Ethereum’s trading volume of one whole year can be processed within one-hour with Conflux."

Not only in regards to the throughput rate, but also in regards to the time necessary for confirming transactions, Conflux also made a breakthrough. Ming said, “While it takes Bitcoin 10 minutes to confirm one transaction, it only takes Conflux about 30 seconds. This will inevitably lead to a wider range of practical applications on Conflux." He also told us that Conflux's main-net will be launched from Q3 this year to Q1 next year.

The following is a live interview with 8BTC.

8BTC: You mentioned that when the Conflux main-net launches, there will be some corporations and partnerships to be released at the same time. Can you tell us what kind of application we can expect?

Fan: The current blockchain world is like a “casino”. You basically exchange your money into chips to participate in gambling. When we look back at the ICO boom, we find that it is identical to the pass the parcel game. Why? Because the blockchain world and the real world are disconnected.

We want to utilize Conflux's high throughput rate to connect the real world and the virtual world. For example, if I want to start an insurance project for PM2.5 on the blockchain, won’t this be related to the real world?

But the question is, how trustworthy is the data on the Blockchain about PM2.5 in Beijing? In fact, after we solve the aforementioned problem, applications can be launched on the blockchain. How do we insert information from the real-world on the blockchain is the key to this. Once we have breakthroughs in linking real-world information with the blockchain, we will have more meaningful applications that can be launched on the blockchain, that is our goal.

8BTC: Linking real-world data to the blockchain is currently being processed through oracles. What is the difference between what you do and the oracle?

Ming: There are still some problems with the mainstream solution of blockchain oracles. It’s common practice to deploy a new distributed protocol around the main chain, then execute the protocol with some designated nodes and finally deliver it through the smart contract of the main chain. The distributed protocol is mainly used to collect real-world data. But I think there are some problems. Distributed protocols are made of a small collection of nodes, which is not secure enough, and for this reason current oracles are designed with very complex distributed protocols to punish a malicious party. The delay in real-time processing is very long because the process has many steps. We also want to do it through the oracle, but we will use some new methods to solve the existing problems.

Fan: Currently, the blockchain oracle is only a term that has not solved any problems. Ming can tell you more about it in a very professional way, let me simplify it. First, the player who provides the data is called an oracle. For example, if we write a smart contract for PM2.5 insurance, then the distributed protocol running on Conflux will be executed by all Conflux nodes, and therefore, no one can do anything evil. However, if the data is only provided by one person, it is meaningless.

For example, if we find the Beijing Meteorological Bureau to provide this data, it seems to be very reliable at first, but this means that it has become trust towards a centralized data-set that is stored on the blockchain. If the Bureaus station server is altered, then all our efforts on decentralization become meaningless because we only have a single source for the data.

That is why we need a decentralized oracle, where different parties vote collectively to decide what PM2.5 was yesterday and remove the highest and lowest point for the average. This is is the most fundamental idea. We must figure out how to guarantee real-data through game-theory and build a model based on it. We need to know what the penalties are for players that provide false information in a collective. For example, if you buy an insurance with a pay-out of only 1,000 yuan and have a far higher penalty for cheating, then you can believe that the players will not take this risk to lie.

All the current theories on oracles are not perfect. Everyone just talks about it, and they are not systematically connected with public chains. Also, there are not many cases. In addition, other public chains don't have the capacity to process the necessary amount of data, but Conflux has, so we are thinking about how to design a good oracle based on Conflux.

8BTC: Conflux had previously said that it targeted to be a better Ethereum, so does that mean when the main-net is launched, it can do what Ethereum can do? And what can Conflux do that cannot be done by Ethereum?

Fan: Everything that Ethereum can do can be done on Conflux, and our smart contracts are compatible with Ethereum. Internal experiments re-ran all historical transactions of the first year of Ethereum. Three to four million transactions in the first year were processed by four to five hundred nodes on the Conflux test-net with 2000-4000 TPS. Therefore, the first-year transactions of Ethereum were processed within one hour by Conflux.

So, what can we achieve that Ethereum couldn't? When the throughput is increased by 2 magnitudes, the transaction cost will decrease. For example, when submitting a transaction on Ethereum, the handling fee may be around 10 RMB when the network is congested, and 3 or 4 RMB when there is no congestion.

So imagine what would happen if there was a taxi application on the blockchain? People all have mobile phones, we press a button to transfer our location to the blockchain to indicate that I want a car. The driver will have to update his position every few minutes when he drives, and then the system will send an order according to his position. If you upload a position for 5 RMB, the driver will also upload a position for 5 RMB. In Beijing city, the average cost for a taxi-ride is also thirty or forty on average. Using the blockchain to make a taxi-ride on the Ethereum, even in the most ideal case, will cost about 15 RMB, and you wouldn’t be able to upload your location frequently and only have one chance for the location-upload, how is this a feasible solution?

If Conflux is launched, the throughput can be increased by two magnitudes and the transaction cost per unit can be reduced by two magnitudes in the same situation. So what are the things that Ethereum cannot do but we can? We can put real-world information on the blockchain and use our throughput advantages to generate new applications.

8BTC: We have noticed that there are a lot of DeFi projects on Ethereum, but there is no good cross-chain solution at the moment. These smart contracts can only interact on Ethereum. Do you have a solution?

Fan: Cross-chain is actually a subset of the oracle. What the oracle does, is to bring all the real-world information to the blockchain, and the cross-chain brings the information from one chain to another, which is directly related to the oracle. We are advancing on this matter and we can definitely do this.

8BTC: The concept of blockchain is too wide-spread, and the discussion that people have is usually not on the same level. We all know that the Internet has seven layers and we know what each layer does. But for the structure of blockchain, it seems that no one has a clear structure, do you have any thoughts about this?

Fan: We have never officially discussed it. I think an underlying public chain like Conflux has the most fundamental protocol for reaching the consensus more efficiently. On top of this, there may be some grafting for security trade-offs, like layer 2 solutions we have discussed. On top of that will be a connection layer for applications.

Suppose that Conflux is an operating system, there may be some accelerators on it, like layer 1 and layer 2 solutions. There may also be some drivers that are not related to the operating system but are important, e.g. an oracle. The oracle is not a part of the public chain, but applications need the oracle to function. One layer further up is the application implementation layer, there you will find applications that you can think of.

Ming: The seven layers of the Internet you just mentioned are the protocols for the network system. But in fact, when we discuss each layer itself, they are hierarchical. For example, when we talk about applications, the software will utilize several layers. These layers have nothing to do with the seven layers you mentioned. Blockchain systems also have different hierarchical relationships. For example: nodes need to communicate, there is a P2P network layer, on top of that there is a consensus layer and a storage layer. If we discuss some specific technologies, we can certainly talk about each layer for quite a while.

Fan: Let me add one point, in addition to the acceleration of performance, there is also the need for staying anonymous, this can also be a layer on public blockchains. For example, we all have data, but I only want my data to be verified by you but not known by you, in this case, we can build another layer to do this, because it is quite inefficient to do this on public chains, and we can hardly achieve what we want with the existing technologies.