Rethink Trust 2018: Alexander Vlasov, Scaling Blockchain in Depth
Recording: Rethink Trust 2018: Alexander Vlasov, Scaling Blockchain in Depth
good evening everyone just in case I should correct all exams the city of the bank exfoliation of the main pancakes Vargas foundation is a non-commercial stichting registered here in Netherlands in what we do is 100% open-source today I'm going to talk about the scaling problem on the blockchain just for a little bit of context I talked about public law chains and mainly serial for now we base our solution for assumed later we can bridge to heroes or now or any other public book in which we find fits our philosophy a second requirement for is a scale his solution I'm going to present and is based on original plasma protocol introduced by vitalik and Joseph I think last August or July as the public blockchain it can be run on a requires smart contract was quite involved functionality and for now we can only speed up some specific sets of transactions later on we really hope that in four or five years we can build fully cheering full blockchain scaled and bound to the root chain but for now I will only talk about transfer of volume so first of all why it's the shift of trust the shift of trust we give some trust to one party so we allow one party to produce the blocks for us as we centralize all the data also processing etc we can speed up the production a lot right now our speed is 10 second 10k per second really cheap and we know how to scale further we just don't need it the second we let every participant in such subnetwork to verify every block and every transaction and so I know a few people mentioned plasma earlier today this main problem people should care about in plasma itself or similar sub chain is not some invalid transactions which can be cut and verified it's a data availability problem if the operator of some sub chain of chain of how we do call it just decides ok I will produce the blocks but I will never show it to anyone then you can do nothing with it and this is a largest problem in plasma and similar solutions after it we just guarantee that the supervision can be can be actually enforced and to enforce it we need a smart contract which works like a Supreme Court and such smart contract can only be run on a chain which is public where no one has a hundred percent authority or power of any kind well at least unless you make 51% attack and proof of work yeah to prevent cheating you need some economical incentives which trivial if you want to prove something and if later you can be challenged or disproved you should put some bonds it's not that interesting and the worst case scenario we should only always consider worst case scenario if something happens with such plasma sub chain operator but say myseif outage it's that data centers loss of everything or we just decide well we don't want to produce blocks anymore in the worst case scenario is you lose time you don't lose your money or assets or anything you just lose your time it takes two weeks to fix everything and move the SS back to the main chain well what can go wrong and send it such a chain such sub chain the plasma is interaction between the user and the block producer or who we call the master the client can do only three well thing client can try to double spend most likely the client will never double spend in a plasma chain or sub chain itself because the master or block producer can just prevent it but the client can use the time difference between two sub chains while the main chain and sub chain and spend on one and try to move from one to another so it should be prevented but it's three well it's really trivial but what can you go wrong if the block producer does something not appropriate and misbehaves well it's a long list and the right construction should prevent all of this I think I'm is the most difficult part that the plasma operator can just whisk hold the block and make some transactions but never show it to you if they're never shown to you you can never verify them so it should at some point be enforced that even if the operator misbehaves everyone who behaved correctly before and up to all the blocks which were available and everyone could verify this is a point of validity everyone who had some essence before this point can exit before something can happen with the blocks which never shown a little bit technical story involved later the day L 8 in a day but I have to say it's a crucial part so what's our current progress okay there are various options how you can construct such a sub chain we focused for now and the construction which allows you to divide the value so for some and for some assets which are not divisible the construction is even much easier and much simpler to implement which can be used in some areas for example I think logistics would be good you never can divide a shipping container well most likely you will not divide your shipping container but for now we focus on transfer value what we have right now as I'd say some numbers 10 key transactions per second which are produced by the producer and verified by verifier yes very fire requires some computation power but we can do it we know it we tested it and we can even do supply public demo if necessary Zuri verifying clients of Torah as Elsa non-tribal but much more trivial since the server the transaction cost as 0.5 cents and we can push it lower after some optimization of the server side it's purely economical value it's a value which gives us some money to sustain the system continue to develop and at the same time cover just upkeep costs of servers storage and everything ok well three data centers replication is released we can do so what can we do better yes we can first of all we can try to work on privacy for now every transaction on the most of the block chains I don't talk about the seed cash or Manero is available for blocks a variable for every one amounts transacted a public so you can see who sent how much to someone hurt you what party what we can do it can we can hide the amounts very strong cryptographic protocol which allows us to do it we just need the next work of a serum to implement it and push it just Tarara beta for now we don't focus on it so I project the horizon three to six months after the next decision fork if everything goes well so it works 9 to 12 months but we can do even better we can make a solution which doesn't require a trust to an operator to do it you need some very advanced cryptography someone already mentioned - earlier - Daisy key snarks the key starts a difficult philosophical topic for now they require at what's called the trusted setup so some parties come together make some keys and if no one has destroyed his part they can come again together reassembles the key and falsify any transaction I hope in one month there will be paper published which you can show the protocol how we can remove this part but still it's very difficult it's even difficult in terms of computational power for now a lot of optimization should be done to do it but I project two years of development for a first beta or maybe even for a first production we will see now thank you I try to be quick as possible if you have some question I don't know how much time we have left ok ok thank you first of all so any questions yeah a little bit too technical in wall sorry okay thank you [Applause]