Toward Ethereum 2.0 Sharding Universe by Hsiao-Wei Wang (Devcon5)

hello everyone I shower from research team and today my talk is about toward eastern 2.0 shutting universe and also before it's complete and delivered there is a gap between now and the end game so what we shall build for it so this is the well-known is there a roadmap that I don't need to go through again the allies quite simple all those faces are have its own girls and however found the decide to production let's actually like it's actually like this so how impractical is like a twisted a highway layer by layers and connected so the the courage of the current approach is much better than the early shoddy and Casper design like in the 2017 and where there were no beacon chain between the East one chain and is to chain and The Shining trains are actually they were depending on the expansion but however even with the become chain we still have to comfort the backward-compatible issues for it and for the design areas something sometimes is vertical in some highs is horizontal and that I recall that that in 2017 research King has received some feedback from community found the client thinks that we should have more incremental ism in the design philosophy that is to add more to add many small incremental steps step by step changes instead of a few large sounder drugs to compare this to different strategies and for example when we are building a city with incremental strategy we should start from the base construction and then build the small small houses and then nearby the small houses we build higher buildings the last incremental saying looks like and for the sound ins and with the southern implantation we would draw the blueprint very very carefully and start to contract and suddenly the castle just landed so I'd say that current the current approach in each row is much more toward incrementalist and design but there's still something that we need to think about so the disadvantage of the incremental decides that - in the part of you can see that in my waist on time to build the face and faces and let's also the Beco fallacy seeing can see they start a picot it has a very very good sense of smell but we've very limited I'll I sigh so it's possible they're praised just found out them but it sounds winked so like can see that the prey even if read it's very close to it so so is that in the local decide so that's why we usually have to go back elbows and to fix problem that are hiding somewhere and ok so classic and I organized survey recently the survey was for the external community and especially the developers and one of the question is what portion properties or functionalities do we sorry do you want to see implemented in East 200 or is to one point X so we got about sixty nine people answered a structure and I categorized them with some simple text below and the top three wishes there are scalability is proof of stake and get a free model increment improvement those three different requirements but at some point they are overlapping I'm sorry [Music] they are all rapping in the is to old design map so I select some of the topics based on the top wishes today to introduce to you and this three topics include the abstract and stretching and application and some significant changes I will have him English - all shouting universe so the first one is there one account abstract so it has been discussed in the East one point acts as well that in the abstract account model all contracts are for world cultures so there's no all count are forward contracts so there are two main strategy why is the after we only abstract the signature and another one is we do for abstraction include the nouns gas and payment and sorry and with the guaranteed transactions and also if we plug into the east to e es we will have the EE can have its own different execution environment and also each environment can have different account abstract model inside so it's very fixed flexible and very easy to plug into the one to each other another topic I want to talk about is the IP 151 five five nine it's the free market model change created by italic and Eric corner and so in this scheme we will have a dynamic pest free value and this value is addressed by the lights nation rate so if the ionization rate is is over 50 percent this special variable will be well increase and if it's sorry if the realization rate is under 50 percent this this very well be decrease so so since its update automatically it's allowed user doesn't have to do more work unless the quality service can help to define the fair free for users so it actually it's actually make a lot of you expert early and the current state crow so and also the past few is always bird this design rationale is to prevent minor or the proposer from manipulating the field in order to extract more fries from the user so generally it will be a more fair model Leon just a crow and before talking about shaking I want to mention the transparent future again some cosmic G is the finality to off all finality protocol folder is too old so the validator the duty of the editor has three step one is to make deposit to the east ones deposit country and the second one is when the block has enough has to vote for the current epoch the current change checkpoints and if the drip point got in enough votes we call that this checkpoint is justified and also if the checkpoint is it if the checkpoint has to justify descent does we call it this buck or this checkpoint is got finalized so that's the basic of a greater duty and we also know that the me misbehaved very data never got slashed so one is - - is that it will import wholly they were the top two little fellas from the protocol so after migrating the tutor for prefers tech system it will introduce a new economic model so : Meijer France consensus has been working on this economic models for a while and the model is based on the current forints faith I encourage people especially that Aramis to take a look he's a work and if to get feedback for the economic side so similar to the my poor we can expand that their work is taking poor in a few near future and if the military are not alive for a while and they can't finalize any choice so they will be penalized as well so based on that it's very important to have a stable networking environments for the proof of stake very leaders and the next topic is about the network latency so this message are the most overhead of the chatting network including the attestations which is the vote of the between the country and the shotshells and the second one is the translations so translations only in the shouting Network and also there are blogs are including the beacon block of their and also the shot blocks so um and we have costing so the shot block will point to the beacon block and the beacon park would refer to the shot block sometimes that's called a cross link so the crossing is included in the attestation data per shot shell and then God included in the prod in the attention so also the Shadrin has its own subnet this very clear but sometimes the cross shot and they're working communication is frequent so with the latest is to speak we have four million validators 64 start first epoch and six seconds become black time and we will lead to sixty four thousand individual signature on the big tension but the news is that list individual signature can be aggregate into the smaller message in in the shining Network side we have 64 kilobytes block data cap capacity and three seconds shot blocked start time and it were internally to about like 21 kilobytes per seconds in per shot so that's the overhead here and about the networking layers well it's - oh I used the p2p a stone-faced networking protocol so under the under the networking there it's the pure connection so the Pieta nation it depends on the pier discovery protocol which you are using and the true business logic is it is the message is forecasting with the sub the publish and sub subscribe system so we use the gossip star protocol to support has the sub Nate messages inside xiaochun's so yeah but imaging that if we have like 1000 sorry tens 1064 chakras and about 4 million spectators it won't be very large and also we have to think about the peer-to-peer network topologies so there are three possible cases which is the structured apologist and also the unstructured apologist and that includes flooding gossip epidemic and random work so the gossip is included in this category and also it's possible to help to combine a structured and unstructured ecology into the hybrid model so this is an example of how can we implement it simply so in this simple model the very latest are they can work as a message inside a shy by using the castle sub for a comb and only the proposal of the even slot of the given shot well they only had the only one who had to collect all those signatures for the given start so for the given shot so the overhead only inside Sundays and only the where data only the proposals have to exchange the aggregated signatures when a single one has to councillor issues in losing the data privacy and other nodes can need to trust the list proposals and if the promoter doesn't do their work well then the other the rest of the radiators will lose the reward from it so it's too easy another one is implemented by consensus research team it's called handle handle allows the aggregation of thousands of signatures just under a second and organize is it's it's done by organizing the connection in levels so it's more like a structure to decide it's doing well but we also need to think more about how can we utilize the hybrid model more efficiently so it's two is very cool it's a Collette but great work so I think Justin just mentioned about that very well and also highlight again so we have the EIP is research and magicians and is to speak this art of silent you can found more deeper knowledge about it and also there's a bounty initiate by adjusting and you can get five es if you found any but under the current we can change spec and finally I want to say that even with facing how the the concern protocol is almost done and there are some optimizations that kind hence are working on and generally an optimist or optimistic of what's happening and exciting about how can we establish this problem in the next few stages and okay so it's I'm co-hosting a breakout with like tomorrow morning and if you interesting knowing more about how the Europe developed her experience looks like in the phase one phase two stage and please come with us and join the breakout session and if you want to know more about these two discussions there's a chance that if you travel next after the Japan tour and you can come to Taiwan there's a conference called crossing and just having crossing Taipei okay that's the side credits thank you [Applause]