New Ethereum talks, every Monday. The week's conference uploads by event, in your inbox.

Loading player…

EthCC 2: Alexandre Clement

Ethereum ParisWed, Oct 9, 2019, 12:00 AM

EthCC 2 EthCC, the Ethereum Community Conference is created by the community, for the community. It brings together great minds and forethinkers, and allows the community to grow. Follow us : @AssethFR https://ethcc.io

Transcript

hi everyone thank you for being there so Hurley so I'm Alex on limo and I start here I'm the product owner of fizzy the taxa which is a parametric I'll come back on that later flight delay insurance with automatic compensation of the customers in case of delayed we are using the blockchain in order to register transparently also that I used and processed to resolve as the contracts so today first I'll go over what's parametric insurance on a few words because I guess oh if you are very knowledgeable in the blockchain but maybe less on that on that side and my goal is to explain why parametric insurance is actually a great fits for smart contracts on the blockchain then we'll do a short overview of 50 de taxa a product and then see how we think we could improve this product in a blockchain environments so parametric insurance is basically according to the World Bank a product for which insurance pays out benefits for the loss of assets or investments based on a predetermined indexed for the loss of those assets so uncontrol traditional insurance relies on the assessment of the damage done so very quickly our lives it works first we need to predetermine a compensation for a given loss it's usually smaller and through traditional insurance then we need to find a parameter that correlates well with this risk and the threshold above which the risk is considered materialized so you have a few examples here for commercial property we could do fire of flood insurance with the parameter being the height of the water for crop insurance we could do something like a certain number of days of drought our astreus and surf our airline companies it could be volcanic eruption for instance so if the parameter which is the threshold we know we should compensate because the risk actually happens so it seems quite easy like this but it's actually quite hard to set up because it's directly dependent and the availability and accuracy of historical data for us to be able to mobilize the risk and also on the availability of live data it doesn't mean though that the process is aromatic and claimed less in most cases the customer we still eye we still have to notify its loss and ask for its compensation still it's considerably easier for insurers to check one threshold for one given index than to assess all the damages which means that in a catastrophic event for customers they will get basically their compensation a lot faster through this type of products and through traditional insurance and just a few figures to give you an idea why parametric insurance is a is an important factor for insurer in the last years a share of claims related and payouts from insurers related to catastrophic events has been doubled by almost three in ten years so this kind of product is considered a driver for growth and stability in a lot of regions where it could provide first injection of cash after catastrophic events for instance why our smart contracts a great fit first parametric insurance relies on an image table set of parameters that defines a risk and then the monitoring of a threshold that actually states the risk happened with the smart contracts you can make sure that there are no hidden rules inside the code no exclusions no small lines at the end of your contract and that the previous agreement is strictly in first the execution is also automatic and with cryptocurrencies you can have your indemnity in minutes and coming back to physi so if some of you haven't seen it yet it's basically websites in which you can get a quotation for a given flight in a few clicks you just input your flight details so this is the part where all the data science happens because we are able to price each flight independently depending on its risk of delay and then the interesting part at the moment of your subscription we will write all the parameters of your policy on the smart contract including the time of arrival that will trigger as in their meeting at flight landing we will use the smart contract to provide actual landing time do the comparison between the actual and scheduled times and take the decision whether we should compensate or not so at the moment this comes back to our backends to trigger an indemnity or not but what we would like to imagine is to have physi in an integrated atrium environment and this is what it could look like first the first step would still be to have as a customer coming to an ax a website in order to identify himself we could have a blockchain service providing us this but it still we'd have to go through AXA for identifying our customers which is a regulatory requirement based the customers flight detail we would provide him with the pricing and if you accept it we would sign a transaction and the customer would then be able to send directly on its own to our smart contracts so the signature of the transaction guarantees that the customer cannot change as terms of his contract because everything is written down indemnity is a premium is a flight and the scheduled time of arrival and once this policy is registered in our spot contract the smart contract itself would autonomously gets the time of arrival from an unchain service provider that would you know from his mouth contracts that would communicate with ours and once he has this time you could take the decision to compensate or another customer directly through a transfer of ATM tokens so now why do you think is the most interesting part how to improve in the coming years first of all cryptocurrency payments is not yet a reality it's actually quite complex to enables us with the regulatory framework we have for insurance the first thing we have to take into account is that was all exposure in insurance the exposure is what we could lose if all the policies that all the flights in that case that we are currently covering landed lights which means we would have to reimburse all to compensate all the current policies and we have to define what is the maximum value we could lose in case all flights would be lights so in order to initialize this contract we first need to provide a pool of eaters like let's say around 100 eaters and this is the exposure we are we agreed to cover until if someone buys the policy if someone buys a policy here is the pool will dip them decrease by the amount of the compensation we will offer in that case and the premiums will be stored in another wallet as the amount of taxes also has to be stored independently for us to be able to withdraw it at any time as it axes and insurance products that access us to the regulator if policy closes and there is no compensation the value of C punch compensation we could have sent is provided back inside the pool and it's unblocked for someone else to book a policy with it if the policy Clause is but there is a compensation the amount is effectively removed from the pool and sent to the customer leaving us with a new exposure available of the amount left in the pool plus the premiums paid by the users and so this seems this might seem like a simple solution but it's actually quite hard to respect all the constraints the constraints of the regulator for an insurer in that case and now moving on to the next topic where we would actually like your help the value of the Oracle and on social networks it's actually one of the points where we are getting the most similar challenge and hence we would love to have your opinion to your opinion on this so for fizzy the Oracle is a guy providing us with the times of arrival of the flights the first solution is what we have today where we register the data in the smart contract as the data is publicly available and as you the customer is supposed to be on the flight you can match the data we used in the smart contract with what we experience or what he can find online through different other sources this is very cost effective it is rather transparent even though it relies on the customer verifying each flight independently it's also very scalable and efficient for the end link of edge cases because no data provider is perfect and you will both always have inaccuracies or gaps in the data so as I said issue is that it's really based on individual verifications what we've been requested a lot to do would be to get data via an Oracle ization of the external data provider which means paying a third party to authenticate or call to this flight data provider and saying that it does come from this guy and it hasn't been change in the middle the issue with that is that it's not very cost effective because the cost of identification compared is almost more than 10% of the basis price of our policies and also the ending of each cases is a lot more complex because you will directly get the data even if it's inaccurate transfers to your smart contracts effort solution will be to have the data provider directly on chain with a smart contract so this would remove the additional cost but still leave us with issue of edge cases and also one major issue is that this still does not exist or very few data providers are actually working on that today but as we are on stuff that doesn't exist maybe the first solution would be the most perfect one which would be a distributed unsane that a provider it's means someone who would cross reference data from different data providers like the one we use plus others but also maybe users using their GPS position through smart phones etc and it would also miss a state for them to display their recipes I mean by that that they should really explain how is it takes decision of showing one data instead of another depending of the different data sources they have behind and to us this would be the best solution even though it's still possible if we get data from the users for them to cheat and if you have any question and remarks on this we would be delighted to discuss that with you maybe if one of the four points doesn't make sense but to have your opinion and what we should do and how we could improve this Oracle and fizzing or maybe one of the points wasn't clear enough hello thanks for the presentation regarding the organization that's the price you've seen with our allies or have you look at other services like chain-link or predictive Oracle's so what we've looked at was always around 50 cents per transaction to identify the dollars but I think 50 cents is the price for Iraq Weiss specifically okay well I don't really know the price for the other project but personally I think a realization it's still may be the best fit even if it has a cost because you suggest solution for but I guess without the strong IOT connected directly on the blockchain you can still have misleading information people trying to corrupt the system or trying to impact it also you need to have a pool of people I guess and not necessary everybody I mean people are interested in insurance I work was working insurance but it's not like when the land the first thing that we'll do is to check their insurance contract I want to go on holiday and just don't care you know for once that's what happened today with normal insurance so maybe take he tech-savvy people we do that but the most common people we just don't care also for other solution in Oracle's you can use drive different I mean you can have a pool of data provider and there its life that's like start flightradar24 whatever and you can make never age or even life like labore you know you would when they calculate the liberates you remove some extreme value I mean that's um I guess hints to to make the contract a bit more decentralized because I guess what a solution three is pretty difficult to do unchain data provider I mean to me solution freeze rather easy because it's the data provider just has to push a smart contracts that makes a link directly between between him and the blockchain and just coming back to the first solution to meet relies a lot more on aggregating different data providers and on users that would be maybe here just to you know see if you have any edge cases or if there's something wrong but as of today it's maybe one of the part we had to work the most and fizzy as we want to compensate automatically based on the time of arrival we have to make sure the data is as close to the experience of the user in the plane and even with one data provider this is a lot of logic and automation because they of course never provide or sometimes provide a very precise data but even yeah but they also have estimated data etc so you have to logic to take the best one and also you don't know what is the actual data source they've used behind that maybe it's the airport maybe it's an ADB system that's live for most of the continents but not for all maybe it's directly from the National Organization for air traffic and to me the value of the force which would rely on the fact that you would explain why you would take one data over another because of the accuracy of the directors and I think that is something we're ready to pay for and an additional value to pay for it we're ready to pay for that about the source of the data that the external data do you unbarred airport's data from airports to your solution so is the data provider we use already aggregates several sources including airports and depending on the place on on the planet where it comes from data from airports can be the most accurate but not in all cases so nobody has a fifth solution we never fought about okay thank you I guess oh good [Applause]

Automatic transcript — names and jargon may be misspelled.