EthCC 2: Will Harborne

okay good morning I'm gonna be talking about decentralized exchange and particularly our experience of launching a noncustodial exchange five months ago so hopefully that'll be quite relevant to anyone else who's launching any decentralized applications it may be sort of something that may seem some of the stuff we learned is a bit obvious but also I think how we learnt it was really interesting so just to give a a background for what I'll speak about the first thing will be to introduce FNX and the context around what we're doing and our path towards decentralization and then I'm gonna speak more about what it meant to us to try and build something for a mainstream audience and what we thought the major challenges were going to be and then go into a real in-depth case study of FNX trustus which is our noncustodial exchange and how naive we were when we started building it and what we learnt from the process and now what we believe the biggest challenges will be for a decentralized exchange to go mainstream and how we might start to tackle those so the background is that FNX launched in 2017 as a centralized exchange so we have an advanced centralized exchange which is familiar to anyone who's used crypto currency exchange particularly focused on aetherium tokens and when we launched that the the utility tokens were just starting to do many ICS it was clinically the early part of of communities becoming really really important to a lot of new etherion based tokens and it was launched by a centralized company FNX and as part of our launch we also included the nectar tokens so this is an important part of where we were going which is that this wasn't sold but it can be ant by anyone who's using FNX for trading as a market maker so for adding liquidity and that's part of our journey which is to be centralized in two ways the first is along our governance so our eventual aim is to have F and X and everything it encompasses being governed by a decentralized autonomous organization of connector token holders and so far over the last 13 months we've been distributing those to anyone who's trading and therefore and they were already a couple of thousands of traders who owned these tokens and also any 50% of the fees that were and go into a pot which will eventually be owned by that Dao and used and so along the way we've therefore been developing voting tools and other aspects of governance but on our technology side which is what I'll mainly talk about today we've also been making that journey towards full decentralization in embedded comers and so we launched five months ago at the next trustless which is our stepping stone towards that that goal and particularly this was focused at solving and the liquidity issues behind trading in a decentralized way for the kind of higher higher velocity want more highly traded tokens so we split the problem into two which is that for the long tail of assets which are never going to have second by second price discovery because there just isn't the information and interest to trade them there are more suitable on chain solutions such as uni swap or other sort of asynchronous trading mechanisms but we wanted to build something which solved the issue for the higher velocity tokens and our next step will be to launch a sort of more fully decentralized exchange which is layer 2 scaling includes privacy and many other benefits but what we're learning from trustus is really shaping how we'll launch that and why are we doing this by the way of course we have a centralized exchange it's whether it's already got users but really what was important to us when we started designing this process was recognizing who the eventual customers would be for anything we built which runs on aetherium and we really see it as being the opportunity to serve a whole new generation of customers who are native to blockchain so companies or sort of protocols like melon port for decentralized asset management which can never use a centralized exchange and they want to interact only with systems which are also open systems and owned by their customers and so we've designed FN x from the start to be along those lines so when we started looking at this this problem of how we build for a mainstream audience for a decentralized exchange we considered it that there were already several decentralized applications out there some of them allow trading and the early adopters were already using these and so this is the typical adoption curve which I'm sure many are familiar with where you have your early adopters who are of course very willing to put up with all sorts of problems and bugs whereas your mainstream users need need to have a really strong use case it has to be really obvious to them for them to want to use these applications and you can't they can't put up with all sorts of barriers which you might see in an earlier stage product and that there's a big gap to cross there and we started to analyze what we thought it was and I think this was a very very commonly held hypotheses set of hypotheses at the time which were there were these three huge problems which had to be solved and and if we solved those it was pretend there was a potential to get to that more mainstream user user base so the first was lack of liquidity I think this is still touted as being the biggest problem for adoption of Dex's which for various reasons we're no longer convinced of and the second was low speeds so the fact that every transaction depending on how you design your system will need to go on the blockchain eventually for settlements and the third being a more complex user experience - the very fact of using blockchain and wallets and taking control means that you have to be responsible if you're in funds and and sign certain ways and so we started imagining what we wanted this to look like and this is what we ended up launching so FNX trustless has a very similar interface to essentially Isaac's change so and and by that I also mean a high high frequency limit order type exchange so this is very familiar except in the sense that you connect an existing aetherium wallet so you come along with a wallet which already holds aetherium and other tokens and connect and then can place orders but the interesting thing here was that the order book is also the same order book that you get on the centralized exchange so with we'd essentially taken the existing liquidity which is much higher on centralized exchanges and made it available to people in a noncustodial way trading from their own wallets and so that was pretty exciting to us and the way that we did that was essentially I don't think this diagram is very clear but essentially by providing a gateway into the centralized exchange so users lock up tokens and are then able to sign orders in a certain format and actually we use the 0x protocol format and can submit them through our gateway which then runs a whole series of checks on the orders and if they're if they're successful we'll add it to the same order book that would be used by someone who had given up custody and use at a centralized exchange so you therefore are at that point agnostic to how the orders were placed and once they match together you pull out the signed orders afterwards and lazily match them and settle them on blockchain and on ethereum so the big benefit there is that you still get the high liquidity and instant execution of trades but I aren't waiting for the slowest settlement which then happens later on ethereum so kind of with those various steps we solved the liquidity issue we and solved the fact that you could you could now have concurrent trading because we would know that by the time your next order went through your although you might not have had a settlement yet your balance would have updated and you've also got no sign ups privacy by design because you're using your own wallet rather than creating a new account so essentially we were pretty happy at this point six months ago and we'd built this we were ready to launch it and we thought we'd solved essentially the the three biggest problems and we're ready to see if we could get huge adoption on this platform and we launched it and I think as you might expect things were quite slow to begin with but there was a slow pickup in number of users and and volume over the first month or two with a lot of typical bug fixes user various improvements and by sort of end of November early December things are really starting to pick up so in terms of volumes we were seeing 1 million than 2 million and then we reached a peak of 8.1 million dollars a day of trading here which was a huge proportion of trading fire sort of Dex's and at least noncustodial trading but we were ready to get congratulate ourselves in having solved these problems but as soon as we started to look more deeply I think although there were a lot of transactions happening the number of users were still actually only I think reached a peak of 250 daily users and nearly all of those once we looked at what they were actually doing through trustless we're doing arbitrage operations between us and other exchanges rather than being anything like sort of more mainstream users so these were very specialized people they understood blockchain and they understood Dex is very well and they also were using multiple centralized and decentralized exchanges so there was still really the the early adopters although they were very sophisticated and I think this is the the reality that then sort of broke through our naivety quite hard was that even looking at all these other applications on ethereum the the number of daily users once was also started to path and match between what they were doing on trust us and whether also interacting is still less than a thousand really and it's mainly a small group of people the same small group of people and very enthusiastic and knowledgeable I mean this was hammered home particularly in our kind of one of our first months when someone managed to get about $800,000 stuck in one of the contracts temporarily and they what was the messaging us trying to find out what had happened and eventually we got it out about 12 hours later and this guy was then very happy at all thanks so much can I have a the next t-shirt I love using it having waited 12 hours to get his money out stock and that kind of is really I mean that could only be an early adopter given that he also missed many arbitrage opportunities while I was waiting and everything else so the reality of what we started to see and when we kind of sat back down and started to analyze all of this was that we were still looking right back at the beginning of that early adoption curve and to really cross the mainstream that there's a much bigger chasm that we'd have to cross and it's really based on these five elements of innovation but we started to measure and the way we did this was by taking users of our centralized exchange and also actually other centralized exchanges and asking them to sit down with us usually by sharing their screen and use ethnics trustus for the first time and suddenly we started to get this really rich set of feedback and I think as far as I know this is one of the most comprehensive view series of user interviews that's been done in this space and so a lot of these guys and tip and we tried to choose people who were already traders so they knew and usually had been for a while so they knew a lot about cryptocurrencies they'd been trading cryptocurrencies for a long time but they hadn't necessarily and surprisingly often hadn't ever actually interacted with their own wallets or made transactions directly on ethereum and the first thing we saw really strongly was that we didn't have a really clear relative advantage so in our view the fact of having custody yourself was a huge benefit and that itself would you know it was orders of magnitude enough to get people to come and use this and that's not the way they saw it at all they weren't clear on what the benefit wasn't for them to hold it versus using a centralized exchange it also wasn't very compatible with how they were currently holding their funds so even those who did have a theory in wallets would typically tell us that for example they used my for wallet as their wallet and they didn't realize that they could use that anywhere else that they could extract a keystore file and all private key and had never done so and the fact that also this was much more complex and difficult to try because also your funds are at risk as soon as you're using it and the benefits weren't very observable so the real challenge that we kind of eventually having summarised all of this saw was this frictionless society that everything in the last two series of decades has been about removing friction removing barriers from UX removing choice and actually removing wherever possible the need for the user to think at all and blockchains are clearly a very strong reaction to that and it's causing people to take back some of that control it's just the power back to us as consumers from a companies but that obviously has really strong effects when you design a system which doesn't take into account what people are used to and I think this is a well-known statistic that illustrates it really well but in - even back in 2009 every additional 100 milliseconds cost Amazon 1% in lost sales whereas you know you can only imagine what now the cost of having to have a meta Mouse pop up and for several seconds and block times in 15 seconds would have on people who actually want to do very quick trading and we're trading actually has a financial value depending on how quickly you can do it because you may or may not miss an arbitrage opportunity and we started then to to look at how we actually build a bridge across this and where we would need to start and again coming from these user interviews the the strongest starting point that we saw was going to be education and I think that's very well you know that's spoken about a lot in this space but actually from an exchange point of view that's something completely alien to us to have to think about actually not just educating people about how to use our platform but also how to use aetherium how to get a wallet how to set up and start even interacting with the space and we we kicked this off really by looking at what our real value proposition is and how we make it really clear to people and the fact that it's actually always a trade off so it's not-it's no good at saying if you use a centralized exchange you could get hacked you could lose your funds I think that's been a lot of the messaging that's been going out about this sort of thing but that doesn't connect with people and hasn't really communicated the real trade-off that there is which is that on the one hand by using something like trustless which is non-custodial you get back control and that has a lot of benefits and you trust as that means you've got no signups no loss of access to funds its permission this access but you also have therefore an extra responsibility and there's no there's almost no getting away around that fact which is that you therefore must protect and backup your wallets and there are no safety nets and if we try and sort of make the case without mentioning that stuff it's usually it usually doesn't work you have to almost present it as a trade off and the fact that it's not sort of a win-win necessarily it's a different value proposition and actually to get people to come and use these as mainstream customers those safety nets which actually they they may really want need to be almost given in another way or or even just through education about what the risks are and it's the second huge one which again is very well spoken about is wallet on boarding and so we were lucky to then get some really useful statistics every time a user came in connected of what a a what wallets they were using but also what the conversion rates were four different wallets as in how many people from coming to the site managed to get all the way through to actually trading and one of the biggest things was the eighty percent failed to actually connect to wallet whether that meant that they didn't have one or just didn't want to but this is also an automated process if you had-even have metal masks often so the process actually for a typical user coming for the first time would be if you don't have metal masks and most didn't at least of the traders that we took them from centralized exchanges you have to go to another site install the extension come back get some aetherium on it and wait 10 or 30 minutes and by the way you should have protected your wallet and back to that because otherwise anything that's on there is potentially at risk if you lose your device and then come back and trade and and of course we saw very different conversions depending on what wallets people were using as well and and so we started experimenting with various different other options so one of the first ones we did was Portus which has no installation so from within trustus you you press the button and you can create a new account within and password which is a much simpler flow it's also very familiar to people and that works cross-device so if you connect for the first time on your laptop and then use your iPad you can still use it without having to worry about exporting and importing your private keys or whatever else we saw really positive and interesting results just from simple things like that so from before adding this and after adding it we ended up we now see about on average about 14 and a half percent of traders who actually use Portus and also the majority of those are people who didn't already have a theorem wallet so the nice thing is that if we then look at conversions of people who come who don't already have a browser-based wallet then going through meta master versus going through Portis to ending up being able to trade is I think a very significantly improved conversion and I think there's all sorts of projects working on different ways of connecting but that's probably still our biggest bottleneck and surprisingly the thing that's holding up a lot of what we're doing at this point and so where are we going next with this so part of the reason why we are spending so much time right now on trying to make these these investigations into where the barriers are and trying to test out and experiment with removing them is that what we're building at the moment and what we're investing most of our time into planning is the launch of our a lot launch of our kind of more fully decentralized exchange which will happen in 2019 so that will be a layer to scaling solution and it'll have performance and scalability to rival what's possible potentially on a centralized exchange in many ways as well as having privacy and solving some of the other front-running issues on maintained aetherium but actually what's blocking we think adoption isn't going to be solved by the fact that this could be a hundred times more scalable it's going to be blocked still by the UX and by the education factors and by a lot of the other things which actually can't really be solved by us on our own and so everything we're learning currently from trustus we hope to share and we're also very willing to share any of the statistics that we have on wallet connections and wallet onboarding for anyone who's working on improved solutions for that sort of thing and very much want to discuss if anyone else has similar learnings or has found improvements for for that side of things and hoping that we can start to really share a lot of the the user interview work that we've done with other gaps and so yeah that's that's it from me on on trustus and then if anyone has any questions happy to answer [Applause] so I think this so that this wallet connect Universal login there was a presentation earlier from our Jean connect as well all of these I think are very very promising but the danger is so already we have six six wallet connection options within our UI that's a mind-boggling amount of choice if you're someone who doesn't already have an aetherium wallet and already actually we're starting to see the effect of having too many options is putting people off again because how do I choose you have to start pointing them towards one and recommending something which we don't really feel we want to be quite agnostic and if we started to integrate all of those as well we're gonna come into even bigger challenges and some of those I think are better are like much more smooth and quite fantastic user experiences actually but unless there's a standard and one that gets very widely adopted with we're gonna have even more of a an issue with then having new 10 or 15 different options and asking a user to try and choose one any others or handover to Jordi Oh what was the retention of porters users I haven't had it running long enough to say for sure but the I mean part part of our messaging around it is also partly that for many of our users it may make more sense once they've experimented with it insider-trading to start using a hardware wallet anyway and actually for most of our most people here doing large volumes through this to trustless end up using ledger or treasure and I think compared to other - that we've looked at in terms of wallet statistics we have a much higher ratio with Hardware wallets and I think that kind of makes sense so it's partly a lot of this is ok come and try it using these but once you get familiar with it and enjoy it try try with a hardware wallet