What is DeFi?
Editor's Note: This article comes fromChain News ChainNews (ID: chainnewscom)Chain News ChainNews (ID: chainnewscom)
Chain News ChainNews (ID: chainnewscom)
, author: Wang Guan, a blockchain industry researcher, has participated in the joint initiation of several blockchain technology community organizations and public chain projects, graduated from the School of Architecture, Southeast University, and published with authorization.
first level title
Recently, DeFi has become so hot that it can be called a phenomenon. Many people have talked a lot from the perspective of economics and Internet applications. In this article, I try to provide a different explanation based on "computation theory".
Ethereum is a representative project of the second-generation public chain. According to its white paper vision, it aims to use blockchain technology to realize a distributed computer with Turing's complete computing power on the public Internet. Compared with the predecessor Bitcoin, the most important upgrade is the realization of Turing's complete smart contract . So far, after five years of operation, a large number of smart contracts have been deployed on Ethereum. Whether this computer really has Turing’s complete computing power is worth studying and exploring. It may be the key to our thorough understanding of the Ethereum ecology.
Referring to the definition of the general Turing machine, if the state of each address ETH (Ether currency) in the Ethereum system is regarded as a coding method of the Turing machine character table, although no strict mathematical proof can be found from the white paper, we will follow the crowd for the time being. The smart contract based on ETH (Ether currency) code is regarded as Turing complete, so as to simplify the discussion of subsequent issues.
first level title
As we all know, most economic activities in the off-chain world operate on the fiat currency system, and the fiat currencies of various economies can also be regarded as Tokens that express and transmit value within the economic system to some extent. We can observe from the price data of various trading platforms that the mapping relationship between the unit price of legal currency and the unit price of ETH (Ether currency) on the chain has always fluctuated, and no natural anchoring has been achieved.
The reason for this phenomenon involves many dimensions, and one possible main reason is that the credit sources of currency value are different. The credit of legal currency comes from the state machine; while ETH (Ether currency) is quite strange, its source of credit is an innovative form accompanying blockchain technology: that is, through distributed power energy input (PoW mining), self-organized The way to generate credit, some specific discussions can be found in "
What Bitcoin Thinks
Most of the time, the rate of change of the credit mapping relationship between the two systems is not equivalent, which causes price fluctuations. The same fluctuations can also be observed between currency exchange rates of various countries and company stock prices. Looking further, even the Tokens running in different contracts on Ethereum have diverse credit sources, such as the huge difference between ETH (Ether currency) and USDT credit sources, so the price fluctuations between these Tokens are also exist all the time.
Although the Turing machine is essentially a mathematical model, it must be realized by physical equipment in real applications, because the classical information form used for encoding cannot be separated from the physical carrier. Most of the integrated circuits and transistors of computers and mining machines are used for this purpose. . Different from the physical form of a classic computer, in the Ethereum system, there are different types of mining machines (computers) involved in the calculation of the contract. Although there are differences in software and hardware, each mining machine is connected through the EVM (Ethereum Virtual Machine). ) output calculation results are consistent, so EVM is also such a Turing machine from the calculation principle, but it runs on the blockchain virtually.
One possible risk is that when we use ETH (Ether currency) as the code and use the Turing machine of Ethereum to complete the calculation of the legal currency Token problem under the chain, the code error caused by the price fluctuation between the two Tokens, affects the Turing-completeness of the computation.
This problem seems difficult to understand at first glance, but we can still discuss it from the structure of the general Turing machine and the principle of calculation:
An infinitely long paper tape, the paper tape is divided into grids, and each grid can write at most one character
A read-write head, the pointer of the read-write grid, can read, write and erase the contents of the grid, and can move a grid left and right
A status register, which is used to save the state and track the state of the whole machine (running, stopping)
A finite set of instructions, the behavior (program) that the read/write head should perform under certain circumstances
image description
Universal Turing Machine
When Turing designed the general-purpose Turing machine, he skillfully converted the calculation problem into a mechanical motion model. Of course, this mechanical motion can not only be realized through materialization, but also can be abstracted in the form of a fully informatized virtual Turing machine. Information is realized. For the coding confusion of ETH (Ether currency) relative to the legal currency Token calculation, we can also discuss the solution by transforming it into a dynamic problem. Abstract the chaotic fluctuation relationship that leads to ETH (Ether currency) encoding into a Turing machine calculation interference model of the external dynamic environment on the mechanical movement.
It should be noted that in actual situations, the coding confusion caused by fluctuations will affect the various modules of the Turing machine. Here we only focus on its core definition module——the transfer function of mechanical motion, which specifies how the machine moves from one grid to the next. Going to the next grid, for the convenience of discussing the problem, we need to simplify the model twice:
The first simplification: Although price fluctuations, as a kind of emergence of complex systems, are a random and non-periodic process, they only have one dimension and two directions. Assuming that we can convert all module coding chaos into a graph through a mapping relationship The random mechanical jitter of the displacement of the read/write head;
So we finally simplified the model to: a Turing machine whose read-write head is continuously disturbed by external Brownian motion, and its calculation Turing completeness problem.
It is an exaggeration to say, in this case, how much difference can there be between a soaked Turing machine and a monkey typewriter? The only effective part is the calculation of the internal problems of the Ethereum system. For example, when each project issues coins (ICO), it will be a distributed storage tool for Token status, or develop some Fomo games with asset attributes on the chain.
image description
A Turing machine (Aero) & monkey typewriter with a read/write head continuously disturbed by external Brownian motion. Source: Wikimedia Commons. New York Zoological Society
DeFi is on the scene, saving the water-soaked Turing machine
Compared with the internal calculation of the Ethereum system, the problem of calculating the fiat currency world is always attractive. The economic volume and social value are not the same, and the profits are considerable. Just look at the financial industry in the fiat currency world. Of course, solving the problems of traditional finance is also the original intention of Nick Szabo to propose the concept of smart contracts. There will be changes if there are incentives. The various structures built by most DeFi projects on Ethereum Layer 2 are essentially trying to resist (correct) this random fluctuation interference and save the water-soaked Ethereum Turing machine.
There are many DeFi projects, although they are all rescues, but each has its own method. Let’s briefly talk about the underlying principles of these solutions:
secondary title
Calculation and communication are not separated. If the price fluctuation is regarded as the channel background noise in the communication process, if the unknown reason cannot be accurately eliminated, the original signal power can be simply amplified and then amplified to achieve the goal of resisting fluctuation interference and accurately transmitting price information. It's like two tourists talking in a crowded square, deliberately raising the volume so that the other party can clearly hear what they are saying. In this way, the problem can indeed be solved, but it takes a lot of effort, so I have to eat more for dinner. The extra effort here is the extra energy spent to amplify the signal power.
This signal amplification process is realized in DeFi through on-chain oracles and public pledge pools. The on-chain oracle machine is responsible for continuously "grabbing" the off-chain price data into the on-chain contract for further calculation. In most DeFi projects, the on-chain oracle node itself needs to pledge ETH (Ether currency) to endorse the credit of the oracle data.
It is limited by the total amount of pledged assets of the oracle node, when it is not enough to generate enough credit scale to complete the business calculation. The price data credit predicted by the oracle node will be further amplified through a public pledge pool. The public pledge pool is open to ordinary users to participate, as long as you believe in the authenticity of the data on the oracle node and pledge ETH (Ether currency) in time to cope with price fluctuations. Of course, in order to cope with the unpredictable and violent price fluctuations, it must be over-staked, which is equivalent to increasing the signal power of the oracle machine, and it will inevitably cause some asset credit to be wasted. Even so, due to the sharp fluctuations in digital currency prices, in extreme cases, this mechanism will still cause downtime due to excessive background noise. Please refer to 3.12 Market.
Betting transaction oracle machine
The general form of the project is a group of on-chain VAM contracts that ordinary users can openly participate in. In response to price fluctuations, the two sides of the transaction enter ETH (Ether currency) into the contract in exchange for the put and call options represented by the original Token in the contract. At the same time, according to the algorithm, a third type of Token in the contract that is less volatile than the anchor target price is generated. Of course As the output of the entire calculation process, the third Token is sometimes invisible.
There are many variants of this model, but it can be regarded as a "gambling model" in essence. Each player uses his own price information (side information) to enter the contract, and through the estimation of the probability distribution of price fluctuations, he undertakes the corresponding Benefits and risks. This process can eliminate the entropy of ETH (Ether currency) relative to the fluctuation of legal currency Token. Compared with ETH (Ether currency) itself, the entropy rate of the third Token used as the calculation output in the contract relative to the price fluctuation of legal currency is greatly reduced, and with greater energy (credit) input, it gradually approaches the price of legal currency under the chain .
image description
Fundamentals of Information Theory, by Thomas M. Cover and Joy A. Thomas
The process in which both parties transfer their own ETH (Ether currency) into the contract can be regarded as a process of applying an external mechanical movement with the same size and opposite direction as the existing random jitter to act on the read-write head of the Turing machine. With the mutual cancellation of these mechanical movements (energy), the information entropy of the fluctuation of ETH (Ether currency) relative to the legal currency Token during the calculation process is erased. It is equivalent to a Turing machine working project, which simultaneously runs an algorithm for correcting the jitter of the read-write head.
According to Rove Randall's demonstration of the principle of eliminating information uncertainty in the article "Irreversibility and Heat Generation in the Calculation Process", we can now understand the energy (credit) input level and fluctuation elimination of this algorithm There is correlation, but there is no strict and complete mathematical proof for each DeFi project, so the accuracy of this calculation method ultimately needs to be evaluated with the life cycle of the DeFi project itself. Since the complexity of such problems may have touched the upper limit of the known complexity of human beings like the universal Turing machine, it may not be possible to find a mathematically rigorous proof in the end. The creator probably uses the life cycle of individuals to solve NP difficulties in a distributed manner. questionable.
Because energy is consumed in the calculation process of anti-fluctuation, the Token corresponding to the energy consumed should also be erased in the system, and the specific performance is that the assets of all parties participating in the game should be confiscated and destroyed. If all the values in the system are increasing all the time, it is also difficult for me to understand the calculation results of this process to be credible. To extend it, I always feel that the Ethereum system fee should be destroyed from the system along with the calculation, instead of directly participating in the secondary flow transfer to all ETH holders to pay. How can the state of such a running Turing machine be reset? After all, it is not habitual to think that the incentive logic of the Ethereum and Bitcoin system economies is completely equivalent.
In addition, for this DeFi model, due to the lack of a pledge fund pool to magnify credit, the calculation scale of off-chain fiat currency Token that can be processed will be limited.
There should be other forms of DeFi models that eliminate fluctuations, and some of the above two solutions are mixed, because there are too many projects, and the experience is anti-human, so I can’t learn to list them.
There are also some projects under the banner of DeFi, but the core is still a "crypto cat"-style asset Fomo game. Or some project tokens indicate that the goal is not to anchor the fiat currency world under the chain, and its asset attributes and dimensions are not discussed in this article.
first level title
calculated value
From the perspective of calculation theory, the once popular ICO can only be regarded as the completion of the ERC20 format data storage of the blockchain, not the complete calculation using Ethereum. After many projects issued tokens, large-scale on-chain transfers (which can be regarded as some kind of calculation process) occurred once or twice, and then most of the project Tokens entered the centralized exchange. The centralized exchange is the information black hole of the blockchain universe. It is difficult to say that the consistency of the calculations in it is difficult to say. It is even more difficult to use this type of economic Token to complete any valuable complete calculation.
The current DeFi popularity is somewhat similar to that of 2017. Although every time there is progress, in the open market, some people will use information asymmetry to create a lot of bubbles, but when the bubbles disappear, there will always be something left. With DeFi taking root, within a certain range of volatility of ETH (Ether currency) relative to legal currency Token, Ethereum has finally realized the original intention of the Turing machine that can "calculate the value of legal currency". For the first time, the blockchain captures the Off Chain value through self-organization, eliminates price fluctuations, and completes Turing's complete calculation. Isn't this the Killer application scene that the industry has been looking forward to?
Expanding, in fact, this type of model is not limited to the blockchain field, and has a certain degree of universality. In the traditional economic field, the problem of calculating the US dollar by treating the British pound as a Turing machine black box, if there is no exchange rate transaction to balance the fluctuation of the value information relationship between the two currencies, the calculation cannot get the correct output.
first level title
Assuming you've accepted the value of computing, let's talk about the limits of computing. If the Ethereum PoW mining process is regarded as introducing real physical energy into the blockchain, it will lead to a more orderly dissipative structure formed in a self-organizing manner within the system. Then the credit of the system relative to the fiat currency economy is the difference between the entropy of this ordered network structure and the network structure entropy of the fiat currency economy itself. However, DeFi completes calculations by consuming the entropy of this ordered structure in the Ethereum system, and at the same time inevitably leads to an increase in the entropy of the Ethereum system. Then the credit scale that DeFi can calculate is limited by the credit of the Ethereum system itself. All DeFi Token credits The total amount is a subset of the total amount of Ethereum credits. In the biosphere, the average efficiency is 10% to 20% in the process of energy transfer along the trophic level. I am very curious about the proportion of this efficiency in the Ethereum to DeFi ecosystem. I can only say that it must not exceed 100%.
Completing any calculation has a cost. Even if your calculation is valuable to my problem, it depends on whether I can afford it. Since all Ethereum contracts share an open public channel, the current DeFi activity has caused a significant increase in Gas fees. We don't know how many problems in the fiat currency world can cost such an expensive Gas fee, and still have continuous profits after the calculation is completed.
In addition, because of this only public channel, some bubble-blowing projects will intentionally amplify fluctuations, making noise and interfering with calculations for profit. Perhaps smart contracts that support encrypted calculations can eventually solve such problems, but there is currently no solution.
first level title
Finally, let’s talk about the "dark cloud" of this article. At the beginning of writing, we acquiesced that compared to the state of each address ETH (Ether currency) in the system, the smart contract of Ethereum is Turing complete. But the actual situation is that apart from external power input, Ethereum, as a blockchain public chain, does not have more coercive forces from the outside world that can effectively penetrate into its system to correct or interfere with calculations. This kind of Turing completeness achieved through self-organization is like a Turing machine affected by inertial force. If the system cannot guarantee sufficient energy input redundancy, the reliability of its calculation results is still doubtful, and of course it is not completely impossible to do effectively. Calculation, but the proof process will be more complicated, and the credit scale that can be carried in the calculation will also be reduced accordingly.
To give an example, along with the calculation, the order degree of the Ethereum system is lost, and why the ETH Token used to represent the value of the system has increased. It feels like DeFi didn’t give any money after the meal, wiped their mouths and left, and then all the ETH Token Holders paid for it together. I always feel that the Ethereum transaction fee should be destroyed.
Regarding the value that DeFi can bring to the Ethereum system, it seems that it is still ambiguous. I lack data statistics research here, only a simple overall indicator. That is, no matter how it is conducted in the middle, in the end, the Ethereum system as a whole must "eat" more electricity (energy) in the mining activities, so that the order of the system is further improved, and the entropy of the network structure is reduced. The greater the positive correlation between this increase in mining energy consumption and the activity of DeFi, the better. If not, it is likely to be a "positive feedback" that causes the system to collapse faster, rather than an "innovation" that promotes a more orderly system.
first level title
spread out
Divergent talk about a few big brain problems.
Regarding the Fork of the DeFi project, because the DeFi contract requires public credit, it must be an open source form. In addition, the current public chain information storage is in the form of unencrypted data, which is equivalent to running an open source algorithm on a public computer. So once the algorithm can capture value and generate profits, the exclusive rights of this algorithm will be challenged. The traditional world is a pirated copy, and the blockchain world is the emergence of Fork. Speaking of Fork, it is essentially a self-"reproduction" behavior of blockchain projects that we are familiar with. It is analogous to nature. As long as there is additional energy supply that has not been captured in the ecological niche, the species will definitely fill it up by multiplying more individuals. Redundant energy space.
The contract copyright on the public chain is essentially a new form of "Open IP", so fork cannot be eliminated fundamentally, but the contract ontology still has the opportunity to use the early bonuses to establish a user-scale network advantage. Since there is no legitimate and effective means to stop Fork, it is better to follow the laws of nature, provide convenience for Fork, and actively breed more contract "individuals" to occupy the corresponding ecological niche. At the same time, pay attention to laying the groundwork in the contract, so that all Fork individuals and Ontology share data and Token with the contract. In this way, cross-contract communication and computing are realized, as if it has become a socialized animal population, business expansion and robustness will be improved, and there are more opportunities to realize data cross-chain between various public chains.
Talk about the relationship between DeFi and CeFi.
Similar to the blockchain and the classical Internet, the term CeFi corresponding to DeFi has also been lit up recently. CeFi has a wider extension, and the most typical formats in the blockchain industry should be considered exchanges and stable coins (USDT). Regardless of the specific form, the most significant difference from DeFi is that CeFi credit is not realized by the self-organization of the blockchain, but comes from the endorsement of the project party organization, which literally means centralized credit injection.
Centralized injection does not mean that there is no credit (the entire human society is also a generalized self-organization), on the contrary, most of the time the user experience is "fast and good". It’s just that everyone can’t judge the credit risk of the project based on CeFi’s limited public data on the blockchain. Unknown risks cannot be priced. Things that cannot be priced are not called risks, but uncertainties. When facing CeFi, users either don’t believe it or believe it all. Once they encounter risks, the level is not a gray rhinoceros but a black swan. The jargon is called "poor system robustness".
Therefore, users will divert assets according to their preferences in different dimensions of risk and cost. In the end, DeFi and CeFi will capture the value of different dimensions and scenarios. When the market sentiment is overconfident, all comers will be rejected; when the market panics, the value will flow to the most certain asset form. So for the current DeFi bubble, it is likely to evolve that DeFi cooks, ETH collects juice, BTC eats, and the ignorant and fearless pay the bill.
