DAPPCON 2024: Making a Decentralized Consumer Price Index - Brenda Loya

Transcript

thank you everybody for coming right after lunch I am going to try and keep you entertained so my name is Brenda she said I am the CEO of teller but in my prior life I was an economist and I worked for the Bureau of Labor Statistics as um producing some of the pr principal Federal economic indicators such as the producer Price Index and the employment numbers that you see every every month and so with that in mind I think the last few years mostly crypto has been really focused on this intermediating the financial sector and I think we need to expand our reach to actually this

intermediate more than that including data that is has been forever centralized uh produced in a centralized way now we have a technology to be able to produce it in a more decentralized way and a more transparent way and I don't know if any of you are familiar with this quote but it's by Ben Franklin he was one of the founding fathers of the US and I agree with this um oops with his oops I'm sorry Which Way well I agree this is actually not his quote his quote actually says uh taxes um except death and taxes so but in this

world nothing can be said to be certain except death and taxes and I went ahead and made a correction to that because as we know if you have enough money especially in the US you can probably get away with not paying taxes you know but inflation it doesn't matter who you are doesn't matter how poor or Rich you are you face inflation and that is something that we cannot Escape um this is the official definition for the CPI and basically all it is it's um the government attempting to measure how much you know you're spending today for for the same item items that you consume and then how much

you would pay for them later and sort of measuring your purchasing power and how much your standard of living changes over time and this number is very very important because it only comes out once a month but it has a very direct effect on our pockets as consumers and sometimes I don't think people make the connection that this really random number that the government publishes you know it can be used for so many things like actually adjusting how much money you get paid every year in crypto we probably don't have that many claes uh for our wages but in general in especially the government they generally have a clause an escalation clause for your wages that depend on the CPI they also also adjust

uh government assistance payments to retirees and even um just you know um retirees and other payments that they do to basically help the economy that are based all of the changes for these are based on this very very specific number and they also do uh basically base their monetary policy on this number monetary policy affects us directly not just in how we transact on every day but also on the long term interest rates specifically if you're trying to buy a house in the US right now I think the rates for buying a house

are like six or 7% when in 2021 it was 3% and so they they do or they Implement these changes to basically dissuade you from taking a loan and hopefully pulling in the money supply and so these numbers is so important but ultimately we really don't have a way to check on it really right there are some paid uh paid systems or paid uh um ways of getting alternative uh numbers but there's nothing publicly available the there is no transparency in the government and the problem with that is that the CPI

itself has a really really mudded history when the CPI was the first commission by the US government was after they implemented the Tariff Act of 1890 and it was in an import tariff that was really high and this was basically the Democrats fighting with the Republicans and they commissioned this survey to try and figure out who was right about it and then probably play it out for the next election and then in 1881 91 they started collecting the data but they only collected data from white guys that were married and they lived in cities that was it that was the sample nothing else it took 50 years for the US

government to actually start or include blacks and Hispanic into the survey and then it took another 40 years for them to actually change from a basket of specific Goods because I I've let's be honest like so my background is Mexican and so I eat a lot of tortillas so that was probably not part of the basket of goods right so in 1870 in 1978 they actually went from a specific basket of goods to categories to help with this bias that they purposely had introduced and also it's uh important to not that in 1941 when they actually included blacks and Hispanic into the survey they purposefully under

represented those populations and that was open and you know known um they've always been very open about their methodology and over the years the main criticisms for it has have always been you know substitution bias lacked inclusion of new items and poorly measured changes in quality and all of these have an effect on the index but there's a lot of research and literature that have gone have poured into that what we don't really talk about as much is the conflict of interest that it exists between what is being produced by the government and um what their policy or their government agendas might be and this is very important because

maybe maybe we feel really comfortable about where we live and we really trust our governments and honestly having worked at the Bure of Labor Statistics for over a decade I think everybody there really does try and create the best numbers possible but it it doesn't you you still need to have some sort of oversight you can't just trust your government that change in any minute and one big very interesting example to me was in Argentina they've always had some political unrest but from 2007 to 2015 you know they were not go undergoing War there were not they have ports they have resources and they were openly cooking their inflation numbers they just would

divide them by they would only report a third of it people obviously knew those numbers were wrong right they knew like what we're experiencing does not match what the government is doing but it fit their agenda you know now we can go and get more loans from the IMF and the World Bank and you know if if it benefits a few people that got R are running the government another big example is Venezuela we still I don't think we know what's going on there but um there has been a lot of research done on that uh by Alberto cabalo and Berto bertoti um and they actually went back and recreated the CPI for Argentina and Venezuela and currently they started

with a project in MIT then later became um infl vera for Argentina and Venezuela and then now it's called price stats it's still under a pay wall or or you know they monetized it good for them uh it would be great if they just let us see the data and actually make it more transparent but they monetized it there's nothing wrong with that but I think we have the technology to make it better see bwe always uh goes I guess every time there's an election or like a big policy thing I think the last thing that they did in 2023 uh their government decided to base their policy on the previous methodology of the CPI instead of the new one it's just that just shows you how politicized this is and how how basically we really have no

alternative right because it's just what they provide and then we just we just trust it antitrust laws exist in the US protect consumers we know the monopolies lead to basically companies not doing anything better for their for their consumers not improving their products and it leads to just a detriment of of the standard of living even in the long run so if we have that to protect us from monopolies and we know that they have a detrimental effect on us then it only makes sense that we should have something that protects us against just having a monopoly of data by the government and now we have open networks

that can provide that necessary competition to make sure the government we keep the government honest instead of um you know them trying to change the numbers based on their political agenda so in order to do that what I am proposing is um this thing that I'm calling the centralized factchecking object fact checked objectivity and it's basically a framework that we can use to leverage distributed networks to collect data in a decentralized way to um publish methodology in a decentralized way and also to disclose who's funding the project because there's a lot of times that whoever is actually funding

the research including universities um if the outcome or the conclusion is not what they expected because they found the research we may never see that that research basically because they fund it they're like we don't like it we cannot use it for our marketing materials instead of honestly being true to research right so if you actually if we do this we follow this framework basically it's super simple if you want to do some research project in in mik's CPI you publish the specification on a DSP like ipfs or something like that uh including the funding or the funding mechanism or you know in this case it's

a common good so that's how I would want to frame it um and then once you publish it you go on chain you register that hash of the the address of the ipfs where where this data is uh saved and then you start incentiv incentivizing data collection then you collect and publish the raw data again on a DSP and that's very important it has to be the raw data uh because then after you do whatever you need to do with your raw data you can go back and actually publish your methodology on a DSP as well the what having the raw data and the methodology open and available to anybody what this creates it's an open peer review system they can actually

keep you honest right and and and just make data actually better and beneficial for everybody for the CPI um the very big thing for it um Alberto cabalo and the research for Argentina and Venezuela was this research was done before the pandemic so even even when it was before the pandemic they used online prices to recalculate the index online prices and some local data they were able to get their hands on and they were able to come up with with pretty much the the correct trend of what the government had reported and that's how they came to the conclusion that the government was not

actually cooking the numbers at the Micro Data level but instead they would create the numbers and then just slash them by a third and so and and with their research they found out that online prices were actually very highly correlated with prices they could find around each geographical area so I'm basing this initial methodology on what their findings were and now that the pandemic has happened at least in the US it has become even more more prices are online about everything there's everybody uses insta card we can get all sorts of information online including I mean in Forever we've always been able to get the MLB or the housing

information which is one of the biggest portions of income where your income goes is your housing and your food and your clothing so all of these data can be collected on chain and be used to create a decentralized index I know the data is not perfect but the data is also not perfect for the government we as taxpayers we basically pay for the government to have people they clean up the data because people can misunderstand the survey can try and provide bad data but ultimately when you have a broad range of data when you have millions of uh data points it's easier to catch and to exclude outliers and that's what we pay the actual government to do through our taxes

obviously the government doesn't actually pay people to collect data and that's because you don't want self- selection bias to be included in there so it's it's the survey goes out people provide the data for free companies provide the data for free under the a privacy agreement basically and that's because you know companies don't want their uh competitors to know too much about what they're doing but honestly at this point everybody has online prices and that's why our verto cavalo and price stats that's are monetized system now where I have been able to create recreate these indexes on their own and so the next thing would be like classification of the data across the

different platforms that is collected outlier detection methodology and then using a really decentralized Oracle obviously to provide the the aggregated index back on chain and the reason so I'll obviously be using teller but one of the reasons you want something that is really censorship resistance is because as long as these numbers do not challenge the government's numbers I think we'll be fine you know as long as they act as a compliment that's fine but I think the moment that it becomes that it diverges too much from the government that could be an issue you know governments might not want to hear that and might not want to uh might want to censor that so that's why it's super important that everything that we use to

actually create this and make it transparent cannot you know the team can go away and the protocol can still function as long as somebody's willing to put you know a tip on chain to incentivize people to you know to pay the gas um to pay the gas and obviously for data scrapers you have to create them and maintain once they're created it's more maintenance and maybe we can you know do the maintenance with the AI in the future but it's a very simple system um basically it is three smart contracts interacting with each other the main contract where the registrations happen for registering the survey or registering the research then you get a research ID and then Reg registering the methodology that you use

and then each research so for the CPI it would have its own contract that is sort of linked by the research ID and then all the information is just submitted to that CPI contract and then when we want it aggregated we just go tip teller and Teller make sure that somebody you know their network of data reporters Aggregates the data reports it on chain and then it's available for everybody to use just like the CPI is available for anybody to use once the government publishes publishes it then we also have a small aspect that you know um to guarantee some sort of quality you know and to grow quality into the system

so there's going to be a provider reputation system that is going to be based on the ratio of the data that they commit there's a commit reval scheme here just to incentivize or to disincentivize mirroring but there's going to it's a function of the commit commit reveal ratio as well as the data quality so how many outliers they were able you know if their data is full of outliers and it's just really bad data then their reputation is going to be lower and you know I'm still taking um feedback on this if you guys think of anything else that I can use I got some ideas yesterday about using um this website that keeps a historical state of

the blockchain and so we could use that off chain to do a measure of how good their data was or how good their data scraper was um so that the CPI would be the first thing that I would try and then the next index will be the Big Mac index and this would be actually the data I would want to collect it on chain because it's not as big as a CPI the CPI if you want to collect lots and lots of data data points you're going to need a DSP you're going to need a place to put it because it's just too much to hand handle for the blockchain but for the Big Mac index you can actually collect that data for specific geographic area and if you collect I mean the minimum in statistical terms is always like 30 but

if you collect at least 100 you can come up with a pretty good number and that and if if the research is small enough that you can actually collect the data on chain that's what you would want to do because once the data is on chain if you disagree with my methodology by all means you know you have the raw data go and register methodology number two for the CPI explain your rationale behind it and uh tip the the teller report data reporters to go aggregate it and report it on chain and in this way it creates a really robust way of interpreting this data and also it keeps us all honest because it's a an open peer review if you don't disagree if you disagree with

my methodology that is okay like we can just clean it up we can make it better as a community um and I have one more minute uh we have the tech to create alternative economic measures to those provided by the governments to ensure their changes are actually neutral and not for any political agenda and I think we are at a point in time for crypto that we need to start not just thinking about how we can make money and this intermediate Finance but how can we actually create things that are that are necessary for for an economy independently of governments as well so with that if anybody has any questions I'm happy to answer any

um thank you so much Renda for your talk and yeah as as she said do you anyone have any questions in the audience haveone done there hi thank you for the great presentation I have a question about the AI um classification system you mentioned just if you can give more color around that and also the in incentivization and mechanism you use for the protocol yes so there's two incentives going on here there's going to be the the first incentive is actually uh the contributors that initially unfortunately or until I get enough funding it's going to be my team but it's uh it's supposed to be sort of like

a dow uh for the contributors it's going to be reputation based for what you know they develop the scrapers and whatnot once the scrapers are developed then that's just an ongoing maintenance that is just it's it's going to be the Dow sort of distributing some funds I would not at this point have not I don't think I want a token I would want just you know you as a community you contribute and that becomes the Money Pot that goes to that then there's the other incentive mechanism that goes to actually um incen iing the the Oracle to aggregate the data and provide it back on chain and for that that would be part

of those funds and the way that I see it and honestly that we I am still working those Kinks out is if you as a community really believe that your government is not doing a good job this is where you would um you would make your voice heard so it is a common good I realize that it's going to be really hard to just find and and you don't a fun but you can actually the government currently or everything is collected for free it's paying for the gas that's going to be the cost if you're actually collecting from people but you cannot really pay for it because then you do this or you create a self- selection bias where like

only poor people that want the money are going to provide information and so that is it's counteractive to what you really want to create um and then for the AI thing um because the government really has a lot of people looking at at these specifications and classifying in different categories like that's why you spend most of your time in when you're um working as an analyst or as an economist statistician uh it's that and then alar detection you know like but that takes a lot of time when you're scraping the data there's going to be some data that's not going to be used but it's just more efficient than hiring a bunch of people to just do that on a monthly basis the government has the

luxury of having way more money but that's one of the reasons that we would use that sorry thank you do we have any more questions we have one up there thank you hey um what do you think about the quality of the current numbers from Western countries like United States or Europe Germany the official numbers I mean person I I for at least the US I think they're pretty good honestly I think they're they're pretty good they do publish there's not just one CPI number by the way there's like several and there's the only thing is that you do have to go into each of the

methodologies to really understand which one you want to use and that is not easy to understand it's also not as easy to to claim that that methodology is what was applied because we don't have access to the micro but I haven't worked there for over a decade I can assure you that at least the analyst they're doing the best job that they can with the data that they receive because it's mostly cleaning it up but um a lot of the the details are are you know the devils in the details so for example like for the US like they didn't include any blacks and Hispanic in the survey so did that really um reflect

inflation at the time probably not and I think that's where like the things go you know depending on how the how the survey um is distributed how it is actually asked even there's like a whole science behind how you ask a specific survey question so that it doesn't bi us one way or another and I think they they do have a lot of people that are trying to do their best at that um when there's political unrest that's when I would probably say I I don't know cuz you you don't really know what kind of political coration is actually being or exerted on that institution on that governmental institution and that's just really hard

to measure but yeah I think I thank you so much once again Brenda um it's been amazing having you talk at dcon and yeah another round of