# IC3 – Crypto Bootcamp Project Presentations - Channel: [Devcon](https://streameth.org/devcon) - Date: 2018-10-07 - Duration: 15:37 - Watch: https://streameth.org/watch/yt-Mf8_seolcuQ - YouTube: https://www.youtube.com/watch?v=Mf8_seolcuQ ## Description Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on IPFS and more. https://archive.devcon.org/archive/watch/3/ic3-crypto-bootcamp-project-presentations A team member or lead of winning projects from the IC3-Ethereum Crypto Bootcamp present their projects and related current or future work. Speaker(s): Everett Hildenbrandt, Phil Daian, Loi Luu, Nate Rush, Haseeb Qureshi Skill level: Intermediate Track: Society and Systems Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum Learn more about devcon: https://www.devcon.org/ Learn more about ethereum: https://ethereum.org/ Devcon is the Ethereum conference for developers, researchers, thinkers, and makers. Devcon 3 was held in Cancún, Mexico on Nov 1 - 4, 2017 Devcon is organized and presented by the Ethereum Foundation, with the support of our sponsors. To find out more, please visit https://ethereum.foundation/ ## Transcript you'll miss that basically k is is a project for writing down programming languages in so it's a language building language is what I tell people and so we have built the languages C and Java and JavaScript and we're working on C++ and we have Python and we have a lot of other big languages and also some network semantics and stuff like that so basically once you write down your language in K you have both the formal specification in terms of a human readable document and you have a range of tools essentially that you can use for experimenting on both your language design as well as for running programs and things like that so you get a parser out of it an interpreter you get a model checker a deductive program verifier and we're working on some some new exciting tools as well I see three the EVM semantics was already finished actually it started as kind of a side project in a class and then it someone approached us and asked us to finish it and gave us a convincing amount of money to do so and the ic3 basically I just extended that semantics to include some kind of higher-level language constructs and EVM so it was just a few days working on that but basically I added things like while loops and if statements and assignment and local variables and structured control flow to the EVM and then just had a small compiler down to EVM itself and I called that language EVM crime the compiler is written all in K it's about 250 lines of literate K code so most of it's actually documentation so it was fairly compact and that just kind of speaks to how easy it is to build new languages in K and right now that project is being used to so we're writing tests for EVM crime essentially and then we're also extending it to kind of our goal is something like a Viper - - so that we can and very easily compiled VIPRE contracts into this EVM crime or this Viper - - thing and then the remainder of the compilation can kind of be well controlled and well understood and then also out of that which you'll get is all those tools for VIPRE contracts right you'll get program verifier and a model checker symbolic execution engine all those tools immediately provide for contracts as well and then hopefully also EVM crime will serve as a nice compilation target for other high-level languages one of the other things we were thinking about maybe doing was targeting lll if people are familiar with that there's another slightly higher level language and then that would kind of give us the benefits of having all the lll tool chains available to us but you know I mean we're open to suggestions in terms of how people would like to see that evolve and you know we'll keep on working on it and building more languages on top of EVM directly I think that's cool yeah so my project was writing secure smart contracts and we had probably the biggest team at the bootcamp I think there were like 10 people on it or something like that and it was sort of meant to be a dual purpose project both educational and constructive for the community so how many people here have heard of hack this site anybody use tech this site okay a few of you so I think a great way to train people train solidity developers and smart contract developers in the ecosystem on how to write secure smart contracts and had a right proper code that lasts a long time is by actually having them getting their hands dirty and breaking other smart contracts so the first thing we sort of did was we got all of our ten people together in a room and we had them write in the ERC 20 with no other instructions other than that just sit down and write in the IRC 20 and then sort of me and a few other people went through and audited their C 20s and pointed out to them all of the ways in which we could possibly break them after that we had them fix their ERC 20s and go through the REO dating process until they had correct D'Arcy 20s and then sort of the second half of the project was to build out this platform hack this contract so you can access it at hack this contract I and the way it works is there's five or six example contracts on there that are intentionally vulnerable you deploy them on the rinkeby Testament all you have to do is give it your link of the address and then have at it try to hack the contracts and see if you can actually break these things so future work directions there's a few features missing from the site which you'll see if you try to use it I still think it's totally usable and fun today so adding those in is obviously sort of priority zero after that I'd like to add in all of the wonderful submissions we had to be underhanded solidity contract contest which you guys heard about earlier today in the main room as well as examples of all the past tax so I would like new solidity developers to sort of come in and be able to reproduce the DAO hack themselves and write that attack contract and I think that by doing so we're sort of educating people on how you can write smart contracts incorrectly and indirectly getting them to write smart contracts more correctly so I think it's a super-fun approach and I think you'll feel good if you log on and try to break a few of these things cool hey I'm Nate deploy on a project called peace relay so first of all to start off who here is sort of PTC relay all right a couple people the general idea is that there's a smart contract on the etherium chain that imports block headers from Bitcoin and once you have bought block headers a Bitcoin in a smart contract what you can do is you can provide proofs of transaction inclusion in those blocks to the etherion blockchain so an interesting use case might be something along the lines of oh you know pay me money to my Bitcoin address and if you do I'll give you ether in a trustless manner and so this is pretty much what we can do with this idea of relay contracts like crossed crossed NBA contracts so the project that we worked on please relay essentially we're trying to be that for any two EVM base chains essentially relay contracts between two EVM base chains over the course of bootcamp we built out one of these and we were actually the you know pretty big stuff here the first ever people to move Rob's tanea onto rinkeby right I know huge and the idea is that you lock an asset on one chain you can then provide a proof to the other chain using this piece real a contract that that asset was locked and what it does is it instantiates like you know setbacks token ERC token that backs the acid on the other chain you can move that acid around in the second chain however much you want and when you want to eventually move back you can burn it on this chain and then provide a proof over here that you've actually burned it you know other than that to cover a couple other interesting applications so I mean the immediate application is that you know you can move like et C or any other EPM you know back coin straight like neo or so like Gollum I'm sorry neo and killed him to interim and so that you know all the B central I can this on a thorium country you know between two constants in this coin so another application is that you know now you can actually you know move private coins right in in some private chain because of the you know interim Enterprise Alliance so many people are like during private blockchain with their own right so you can actually remove the asset in the private sale to the public chain and you know trade between this coin and other yes each what it opens the other interesting thing that you could possibly do is you could imagine kind of so the idea is like you know with peace relay one EBM chain can become aware of another EDM chain on the other interesting thing that you might be able to do is kind of point the piece we like contract that at the chain that it's actually on and that might not make much sense but kind of the ideas we can make the EVM more aware of what's actually happening in the EVM so for example as kind of a spin-off of this project one thing that kind of came out of the summer was essentially a gasps Oracle and so kind of you can imagine that like the AEM kind of becomes aware of how much gas is actually being used on the EVM the reason that's interesting you know you can imagine like you know if there's this much gas used if there's more than 50% of gas used in the last 100 blocks extend the time lock on this to eternal state channel time out interesting things like that right now peacefully is being maintained by two different team the first team is carbonyl group and that based and the second one is I guess consensus in the isolate project and from carbon dioxide we are planning to release some better version a piece relayed by the enemies this month or early early next month so please keep in touch any other question for peace really alright so in typical runner-up fashion I actually prepared a presentation so let me just run through that real quick so our project was front-running Bank art it was done by myself to see Qureshi Ivonne Bugatti and Preeti Katz ready so so quick so uh quick note so this research so we did this at I see three and then we published the research on October 10th at the same time we worked with the banqueting before we actually published to address the vulnerability and put out some fix so that it wasn't so vulnerable to this I just want to point out the bank routine was cooperative and analyzing attack pushing out of fix so want to give props where it's dude it's still in early space people mess up all the time you know blockchain develop is really hard so that was cool so what is front-running funding is a term that originated back in the stock market and in the days when trades were executed on paper so literally what would happen is a broker would receive a large buy order from a client and they might say this out loud so other people could hear and they would go walk over the desk to perform the trade a malicious trader would basically going on or see this is going on and literally run in front of them to put in a buy or sell ahead of time to profit and sort of sandwich their transaction in between now you can do the same thing on the blockchain if a miner or user sees an unconfirmed transaction in the mempool that they can profit from they can squeeze their own transaction to come in front so essentially you're profiting from a beneficial reordering of the transactions now as a miner this is pretty easy because the the way that mining essentially works is that you are sort of the leader for a moment you get to decide what goes in that block and so you can reorder transactions however you want if you manage to mine that block as a user however you can actually manipulate where you go in a block by changing your gas price and most mining clients will sort mine transactions by the gas price so if you come in higher than somebody else you'll come in before somebody else in the serialized execution so many blocks free market protocols are vulnerable to miner front-running but it doesn't seem to really happen very much in practice that we can observe plus miners can only front-running blocks that they themselves happen to mine right so unless we're coordinating among a big mining pool if I'm a malicious miner I only get the front run and it happened to mine a block which is very very rare for any minor however if your protocol can be front-run by random users that's really bad does that means that any user who's sitting on the blockchain looking for transactions can run everything in front run any transaction that they want to so banker was known to be vulnerable to minor front running as many marketplaces are but we demonstrated that it was vulnerable to user front-running meaning any user on the network could front row and Bank row transactions and make money so how does Bank work for those of you who are not familiar Bank is basically an automated liquidity provider using a smart contract it gives you deterministic pricing by maintaining a constant reserve ratio okay so long story short if you buy the price goes up if you sell the price goes down that's that's the law of the land for Bank Corp okay so what that means is that if I know someone's gonna buy I know the price is gonna go up great so the attack is pretty simple you launch a guest node and you write a little bit of Python script to basically monitor the mem pool and see what transactions are going in the network if you can if you can tell that the transaction is one that qualifies as being worth front-running which that's basically that's doing then all you do is you increment your gas price a little bit more than what they paid so do you can see 1g way more than them then you you launch the transaction and if all goes well you should have run around them so running simulations we showed that over a two month period of July and August with 100 each principle you would have produced $35,000 in return the front-running transaction on the main assuming that you know that the market or the people in the network didn't react which is greater than 100 percent ROI over that period of time we also showed that because Bank is a protocol that can be used by many different tokens the smaller the token mark you have the higher your ROI would be for front-running that market and so if there were a large number of tokens that were using Bank run at that time you would have been able to make even more and so we also front ran a single transaction on the main net just to prove that this was in fact doable which we did we returned the Delta that we made to the person who originally made the transaction and then didn't run any further so how can you mitigate this kind of attack so this is kind of endemic to blockchains but there are several ways to try to mitigate it so one thing you could do is force all buys or sells to specify an acceptable price range so you could say okay I'll I'll make this transaction but only if the price offered by the market maker is P plus or minus some Epsilon otherwise cancel the transaction if you do that this can't be front run by them the users because you're guaranteeing yourself a certain price however this leaks important metadata if someone front-runs you and then your transaction is canceled they kind of know let's say you were gonna buy you know a million bank right well okay maybe your transaction canceled but they know someone's trying to buy a million Bank so you know that bank card is gonna go up so if they buy a bunch of banco and they know eventually you're planning to buy they know that they can just wait for your buyer to come in and then they can sell so there's also significant limits your throughput so if if lots of transactions are happening on on bank or then it's very hard to ascertain how wide that epsilon should be so that your transaction doesn't sort of get cancelled and occu ously right so otherwise you just have to accept some margin of front-running that you're okay with if that epsilon you know ideal obviously if you're buying you're okay with the price going down but you wouldn't to go out so it's not a great compromise but it is better than nothing however what you can do is you can set a maximum allowable gas price for the contract so essentially what that does is it puts an upper limit on what people are allowed to pay to try to get in front of line of each other right this is nice because it can't be fronted by users however it's still vulnerable to minor front-running cuz minor can do - can do whatever they want it's not it's not really you know they're not beholden to accepting these rules so it's also vulnerable to transaction flooding by malicious users but this is expensive and difficult in practice and not really something that people have seen so this is a solution that bank or settle on it's a decent compromise between you and security everything so happens in one transaction just a few extra parameters you have to send out now the ideal solution would be some form of commit reveal so in a commit reveal these are well known for voting schemes in a commit reveal you first commit to the trans actually going to make by sending a hash of the transaction plus some nonce then you reveal the transaction announced in a subsequent block unfortunately for a commit reveal to work it has to span multiple blocks to actually make a trade so this is kind of undesirable right like you know in Spanish two blocks things can change it's just kind of a slow crappy UX but this is the safest scheme now doing this naively is still vulnerable to certain attacks so Submarine sense was actually worked on by lawrence Breitenbach phil diane ora jewels and florian tremor but were they proposed a scheme for accomplishing this securely which is which is a lot easier with the create to opcode which is offered by EAP 86 which is not gonna come out in metropolis unfortunately so we're still waiting on that we're also exploring other practical commits reveal schemes that are possible with with current op codes so that is front running there was me me prithi Ivonne social thanks to re Jules Ito been Tawfik Karl flourish for helping us the research and of course to ic3 Thanks [Applause] thanks guys [Music]