Editor’s note: This article from the value chain (ID:cc-value), author: Vitalik, Min Min, Elisa, compiler: a sword, the daily planet unauthorized release.
This is Vitalik in Devcon4 speech translation. English text from the recording and dictation, translation or will have any doubt, please forgive me.
The original link: https://www.youtube.com/watch v=Km9BaxRm1wA?
Congratulations on the 10 anniversary of the birth of the consensus!
Ethernet is 2 square the theme of my speech today, but I hope that the present is not only the etheric Fang 2 on technical points, including why we need to push the etheric Fang upgrade, we will do the upgrade, but also through which to go today.
So, what is the etheric Fang 2?
In recent years, some of the characteristics we diligently study and Discussion on the system, but also put a lot of energy in the development of the App on the end, they come together to form a harmonious world. These properties include rights proof algorithm, Casper; scalability slice; virtual machine upgrade, EWASM; strengthen the cross chain contract; improved protocol economic model, and contributed to the improvement of the logic layer very much diversified. There are so many important topics.
What did we explore?
Proof of interest (PoS) related research started in 2014. This blog I published in a month (Slasher: Punitive Proof-of-Stake Algorithm) describes a person called fines (Slasher) algorithm, introduced the idea of the basic algorithm in the proof of interest: if you violate the agreement and is found, (this can be proved by), you have to accept the punishment; this method can be used to enhance security. But then, as you can see in the slide, I believe that Slasher is a very useful construction, can make the proof of interest “mining” arsenal became more substantial, and provide us with a more powerful reason to switch to proof of interest. Of course, we have not done so.
Therefore, at the time, like PoS as our direction is inconclusive. But as time goes on, it becomes much clearer.
In 2014, first of all, we discuss a series of interesting and important ideas, such as “Proof of proof of work”, which is a kind of enhance the scalability of the proposed; “center spokes (Hub and spoke chains chain model)”, let a chain become the center, a series of chain as subsidiary this is actually a very early extension and slice proposal, try to improve the regional transaction scalability, but does not improve the global transaction (i.e. from one chain to the other chain) scalability; “super cube model (Hypercubes), the cube has 12 dimensions not 3, so we can achieve more than the center spokes model scalability.
Now, for many reasons, these ideas have been abandoned; however, people who have really made a great ICO to develop them, really happy people are willing to try (laughter).
So in 2014 we have not made the right progress?
Progress in 2014 years
Yes, we have。 We got a concept called “weak subjectivity (Weak Subjectivity)”, to some extent, this is a semi formal security model, try to explore the PoS bond forfeiture conditions, and under what conditions PoS is safe. At the same time, we are more determined, PoS algorithm has stronger security properties exist, so little money (Peercoin) and its variants is really possible. We have come to realize, there are some scalability with random sampling strategy can prove the algorithm combining the rights through, but we don’t know how to do.
At that time we identified a roadmap for Vinay Gupta in 2015 wrote a great post 3 month, 4 cited the main stages at the time line of the etheric fang. The first stage: Frontier, which we started a project stage; stage 2: Homestead, from alpha to beta; phase III, Metropolis (literally “metropolis”), that time is expected to develop Mist and user interface and enhance user experience. But since then, we focus on the implementation of encryption technology more support, and interface of ascension into the parallel target. The fourth stage: Serenity, realize the proof of interest. From now on, I would like to call the etheric Fang 2, of course I wouldn’t refuse to use Shasper to refer to. To avoid confusion, we can also be called “Serenity”.
Then, we will usher in a trough: we encountered a series of failures in solving PoS scalability, some of the core issues, Vlad quietly suspended all his work at Casper in CBC.
We are one of the most interesting idea is “(Consensus by Bet) betting consensus”: the participants for the next block which will be finalized once some people bet, bet, bet on the fact that they will become the consideration of other people to bet. So, in a sense we have a recursive formula, with the passage of time, more and more people will bet, which block will be finalized will become more and more obvious; in the logarithmic round later, everyone will own all rights are cast, so I have to to qualitative. Along the way we go quite far, it created a complete proof of concept, you can see (on the slide) to determine the process, and what the signature function. I mean, here we spent most of the time, but then the whole idea on the stalled, the main reason is that we know to which BFT heuristic consensus algorithm can really work safely.
The concept of Rent: the truth is that with the use of storage space charge high fees once, we can gradually collect fees. Therefore, every day, every block or other way, once a particular storage space occupied, you will need to pay a certain ETH. This is written in EIP 103, but it is really only for EIP 35, because 35 is the number of issue. This is a preliminary idea, try to form a formal concept. From then on, we’d better use rent this point a lot of iteration idea on how to.
In 2015, the scalability of the paper
This paper aims to form the concept of the two party patch and super two side pieces, but it is really too complicated. It will contain some very complex (escalation game) game intensified. The whole idea is inspired by the role of escalation in the legal system of the (I know Joseph’s philosophy of love using this analogy, but very much) I tried to put it into the base layer. Which also includes “depth”, that is to say, the state reversed if something is really wrong, then a large number of states can be rolled back depth in the future. Really very complex.
In fact, one of the fundamental issues that we cannot fully grasp, but gradually close, is the so-called “Fisherman dilemma” (Fisherman’s Dilemma). This is a very basic concept of film research, describes the expansion of state executive (scaling execution), the execution and data extension procedure extended availability (scaling availability) the huge difference between the. The fundamental problem is this: if you want to run the program, you can make people what is promised to answer, then you can challenge the game, using the dichotomy search to determine in which step who is wrong (translator’s note: every time run only half of the calculation process, compare the results to find the two sides in the end in one step get it wrong), confirming the fact that you can punish those who give wrong answers. While the data availability problem is: no matter how challenging games run, you can cheat, because you can not open any data until the mechanism to check if you have any public data, then the data as long as you open to examine the mechanism of it.
This has proved to be a great many scalability problems in the algorithm. I used to write a blog to discuss this issue, you can search the A Note on Erasure Coding and Data Availability, which talked about more details. This is one reason for the slow drag on our progress. But even so, we still are very pleased to be able to promote the etheric Fang to move on, we go in their own way!
Well, then, out of the The DAO event (laughter). This is not what, also no more questions. OK, well, another “Shanghai DoS” attack (the audience was a little slow).
In short, The DAO events, DoS attacks, and ultimately delay the many events and energy, even up to 6 months. However, we are still in progress, EWASM in advance, a virtual machine research in advance, we have an alternative version of EVM1.5. But we have a deeper understanding of the multi angle, better block chain morphology algorithm.
Then, we quickly made a series of significant progress. At this time, we have the harvest in many different fields: some are around, let us hope that the interests of the common layer more efficient; some on scalability, we hope to make the base layer can be divided into consensus; some are trying to improve the efficiency of virtual machines; some are on account abstract, abstract account allows users to choose which one to use signature scheme, which can obtain the safety after the quantum era, and make privacy solutions easier; and focus on the economic model of the protocol layer. Of course, all of which are still growing.
At a point in time at the beginning of 2017, we developed a protocol, and a modest Name: minimum fines conditions (Minimal Slashing Condition). The minimum conditions for the transformation of the traditional Byzantine consensus confiscated type PBFT; in 80s, Lamport, Shostak et al will create similar materials, but I simplified the algorithm and put it into the most blockchain environment.
The basic idea is as follows: in a block on the chain will continue to produce a new block, you can merge and continuous observation to obtain efficiency of a pipelined version: the start of each round you will add the data for this, but in the next round of people can confirm a part of these data; we can get great efficiency from the upgrade (translator’s note: This is the core idea of FFG, once in a round in a block of 2/3 verifiers agree, the block will be the parent block finalized).
The beginning of the minimum fines have 6 conditions, then I was reduced to 4, and a year later, we terminate the merger of prepare and commit, this is Casper Friendly Finality Gadget (of gadget friendly determination FFG).
Last year Devcon I demonstrate this new patch design, is to keep the main chain unchanged, then create pieces as Mount Layer-2 system at some point in the existing chain, then upgraded to be reliable enough to confirm Layer-1.
Then from the Vlad Casper CBC to the.
Casper FFG POC
Thailand time 2017 year 12 month 31 day 23:40, we happened to be in Thailand at that time, Casper FFG proof of concept was born. At that time we really want to make sure that a consensus on PoW/PoS hybrid technology is what in the end, and that a hybrid will use part of the concept of consensus from Casper FFG, inspired by a part of the traditional concept of consensus and Byzantine fault tolerance, load on the existing work proof chain. This is a kind of can let us fast turn mixed consensus mechanism, and due to the existing block chain impact is minimum, then we will theoretically be upgraded to the full proof of interest.
We also go far in this way, we also do a Python test client, VPS, server and computer communication to do different test.
At the same time, we also made a lot of progress in the block. We continue to improve on slice technology, we in the 18 year 3 month in Taipei organized a collective rest. Here a lot of ideas began to emerge, how to achieve a sub area block chain looks become clear. To look, it becomes clear.
In June, we made a very difficult, but I personally think that in the long run very valuable decision. We found that: Well, here we have several teams in mixed interests proved that they are trying to Casper FFG for the intelligent development of an existing contract chain, make an adjustment to the bifurcation of choice rules; on the other side, there are a few pieces of complete isolation of the team, to a verifier or developer said management contract (later the chain was renamed patch management contract), development system on this basis; they are not between exchanges. In this side, eventually becoming more and more clear: let the core slice system on the PoS chain will bring a higher efficiency for us; into a contract on the PoW chain, we have to deal with the overhead of EVM with Gas mechanism, to cope with the unpredictable PoW block time PoS; + partition can bring faster block time and a series of benefits.
Then we realized why, here there is a team in the development of PoS, there are a PoS project? Why do we want to repeat the work? Then we decided to work together on both sides. A lot of work before the decision to a death sentence, but profound: we will work together to develop a technical specification, the same protocol, and we can also make Casper PoS and slice complement each other.
All in all, and have a purpose, there is another side, then we try to tell how to merge the two results, we should spend some time in the starting point, and then to a proof of interest and area block chain. (applause)
At the same time, we spent a lot of time to discuss the bifurcation selection rules, finally we are getting closer, more and more in-depth understanding of divergent selection rule; although GHOST algorithm is the beginning for the proof of work design, but because of a series of reasons, we propose that the rights for re.
Justin began to study of delay function (Verifiable Delay Funciton, VDF), we have a studio at Stanford, Justin has been there many researchers with cooperation.
About how to realize the abstract and rent, we also formed more ideas. For some reason, we decided to rename the storage storage rental and maintenance costs.
In terms of research, we are studying the inter slice transactions, such as csoss-shard-yanking, we generalized some concepts in traditional distributed system, using asynchronous cross patch in the environment; and resource pricing papers, including the optimization of cost market, how to charge the storage and maintenance costs, the trade-off between different design. Casey also wrote this article to discuss the inter slice transaction synchronous assumption. Of course, the CBC study also extended to the field of the patch, and, this really is not Valding, Vlad hate the word. We have developed in the etheric Fang 2 development strategy is: to promote the use of multiple clients to the center of the pursuit of creative development. This is not just because we have to go to the center of the pursuit, this is a very practical strategy, is the other team would have bet you do well; in addition, the Shanghai incident let us wake up, if a client of a problem and no other client, the entire network can run better; by the way, also let the development of the whole ecosystem more does not depend on the fund itself. Foundation of real development of the client is a Python client, Python has inherent limitations in performance, it also gives you the incentive to run created by Prysmatic, Lighthouse, Status and Pegasus client and other such as bamboo shoots after a spring rain as the emergence of the team development of client.
The upcoming Serenity
Then, I have to say, the upcoming, cannot be called Shasper. Serenity is about to set sail!
What is the Serenity?
First of all, Serenity is the fourth stage after Frontier, Homestead and Metropolis. Of course Metropolis is divided into Byzantine (Byzantium) and Constantin Bo (Constantinople) two stages, and Constantin Bo soon started to upgrade.
Secondly, it is the combination of us in all these different areas into 4 years, including Casper (not only is a mixture of consensus, including 100% pure Casper slice, EWASM), and the concept of all of these protocol research.
Then, this is a new block chain. From the data structure is so, but it will be with the current workload that keep in touch block chain. So, for example, the PoS chain can perceive PoW chain block hash value, you can transfer from Ether PoW to PoS chain chain. This is a new system, but also a set of interconnected system, the long-term goal is: when the new system is stable enough, the application of all existing block on the chain can be stored in the contract a slice on the new system, then there is a EWASM written with the EVM interpreter. This point has not been finalized, but from the current point of view, this is almost our roadmap.
Again, Serenity will become the world’s computer, rather than intelligent mobile phone in 1999, only 15 transactions per second, can barely play snake.
Finally, Serenity will continue to the center, we also hope that in some dimensions, it can be more to the center than the existing block chain. For example, as a beacon chain verifier (from the current point of view) your storage requirements may be less than 1GB, today you need 8GB to store state than, or like storage Ethernet block chain online rumors that Fang’s need for 1.8TB, will be much better.
The expectations of the stage
Phase 0: beacon rights proof chain. From a certain point of view, PoS beacon chain is a block chain warehouse, storing all the information, but it is the silence. The beacon chain will have some verifier (validator), the verifier will run the PoS algorithm; it is between the test network and the main thing, it is certainly not the test chain, because you can pledge the real Ether in and get real Ether rewards, but it is not the main chain, because there is no application, so if it collapses, people will not kudiehanniang, won’t cry like Shanghai DoS attacks.
Phase 1: as the slice data chain chain. This is the piece some role. This is a simplified version, that is to say we do not state slice, but the data sheet; you can put the data into the chain, then try to run a state of their own execution engine. In fact, the most easily slice data. Therefore, if you want to do to the center of the application in the block chain, you now have scalability, but does not have the ability to execute state so the development of intelligent application of contract and all those complicated things. (translator’s note: this sentence should have doubts. The accompanying video link in the original twenty-third minutes).
Phase 2: EWASM state transition. This includes: support virtual machine, support account, contract, Ether transfer sheet, etc..
Phase 3 and beyond: iteration, adding new technology.
PoS pure consensus.
A faster synchronization confirmation time (8-16 seconds). Note that because of the bifurcation selection rules and signature chain beacon mechanism, a beacon chain confirmed between hundreds of verifier message delivery, from a probabilistic perspective, in fact equivalent to hundreds of times on the chain to confirm the block existing Ethernet work proof workshop. The synchronous transmission in the message under the assumption that already can be regarded as a block will be finalized.
In the confirmation of synchronization under the assumption that the 10 – 20 minutes to achieve economic uncertainty (Economic Finality).
Virtual machine to execute faster by EWASM.
Ideally, reach 1000 times the scalability.
The innovation of Serenity
To strengthen the protection of privacy. We have done a lot of work. For example, in the Byzantine stage, we activate the elliptic curve arithmetic, elliptic curve, pre compiled. Barry White have done a wonderful job, use Layer-2 to protect the token transfer, voting, reputation system in privacy. This work can be borrowed.
Cross patch trading.
Semi privacy (Semi-private chain) chain. The idea is that if you want to develop some applications and data remain private state in a part of the user, you can still put all the data into the chain, you are only in encrypted form put it up, or upload the hash value, zero knowledge proof, can decide.
Proof of interest on innovation. This is a thing we always concern. When Casper CBC enough economy in expenses, will be added to our roadmap.
In Serenity, we will prepare everything to upgrade to STARK. For example, STARK is used to verify the signature polymerization, erasure code (erasure code) to check the data availability, may eventually be used to test the correctness of the implementation of the state.
Cross patch transaction form stronger.
Single confirmation faster. To confirm the time from 8 seconds to lower.
The intermediate goals
The final stabilization function of Layer-1
Considering the use of Issuance, as well as the cost of……
When people want to guarantee provided by the protocol layer (guarantee) and get more and more able to attribute the consensus
Considering the problems of governance
The upcoming update
In the final issue before, we will do what updates?
First of all, the protocol layer on stereotypes. Has been paying attention to the GitHub on the Ethereum and the 2 main specifications, beacon content should chain code until the protocol layer is developing very fast, but very soon we will get off the ground.
Continue to be developed and tested. It has 8 lanes and 2 Ethernet protocol implementation under development.
Across the client test network. Some people show very much hope to see across the client test network in the next year 1 month run up. Of course, I also very happy to see that this year 8 months we have seen the hope. Here is a brief review of the development history of the etheric Fang 1. The white paper from the concept to the official release, the interval of 19 months, with part of the reason for such a long time in technology that has not yet finalized before we tried to create cross client compatibility, we have agreed to do the test, test network, wait for the agreement to change, agreed to release new test network, and then wait for the more changes of the agreement, so repeatedly 5 times. In this matter we have learned a lesson, we do not need to focus entirely on the cross compatibility of the client, can wait for a real close technical specification was finalized again. I think we can release candidate technical specifications is not far off, at least in part of the state does not include execution is so.
The last is the security audit. Here what the audience that security audit is important? (audience) what the audience that the security audit is not important? (who think that the world is in perfect silence) is the lizard (Lizardman) control? (laughter) well, third people also smiled for a second, awesome (translator’s note: mean to believe that the world is the lizard man control, than many people believe that audit is not important).
The last is the issue. Do you think the issue is not important? (no response) well, then all who love the most political candidates is a lizard man? (laughter)
Well, so, the issue is that we have been expecting for a long time and in the past 4 years of continuous struggle for milepost. And I believe it will not be too far away. Thank you！