Hashing It Out
Hashing It Out

Episode 88 · 2 years ago

Hashing It Out #88- FourSquare Max Sklar

ABOUT THIS EPISODE

Max Sklar is an engineer at Foursquare and is the head of a project called Swarm. John and Jaye speak with Max about this project,  machine learning, how to build an app in 2020, and can tokens/NTFS on blockchain be incorporated into Foursquare.

Links: Max Sklar

The Bitcoin Podcast Network

  • Slack

Donate to Hashing It Out

Discuss

Hey guys. This week's episode is brought to you by avalanche. Avalanche solves the biggest challenges facing etheriums, developer and decentralized finance or defy community, that is, velocity, security and time to finality under three seconds on the first decentralized network, resistance to fifty one percent attacks. With complete support for the etherium virtual machine and all of the tools that have fuel defies growth to date, including Meta Mask, web three, dot JS, my ether wallet, Remix and many more coming, avalanche will be at parody with atherium. For Defy developers that want a much faster network without the scaling issues holding them back. Get started today, building without limits on avalanche by going to chat dot ava x dot network, that is, chat dot a v X dot network. Thanks. Now entering incast met work. Welcome to hashing it out, a podcast where we talked to the tech innovators behind blocked in infrastructure and decentralized networks. We dive into the weeds to get at why and how people build this technology and the problems they face along the way. Come listen and learn from the best in the business so you can join their ranks. Welcome to hacking it out. I'm your host Jay Harrold and I'm your host John Marlin. Today and our show we have max far from four square. Max. Could you give us a quick intro Bo yourself and let us know what you're doing at for square? Hi, John Hij thank you for having me on. I've worked it for square well, basically since two thousand and eleven. There was a little bit of time that I spent away, but now I'm sort of on the innovation labs team, the labs team, and not exactly sure what our official name is, but basically we are a small group, three, four five people, depending on how you count, who are just tasked with using four squares location technology to build cool consumer products and and put them out into the wild. And I love this team because we could just build things quickly. We can just do very quick prototypes and we don't have to worry about the you know, we don't have to worry too much about the business case or the the short term money concerns, which I did have to worry about on some teams at for square, which weren't bad teams either, but I love the idea of just being able to build. It's it's a lot of fun awesome. Um, so maybe will. The question we asked on every episode is it's just what it brought you, brings you into the blockchain and crypto space, and what's your backstory? You Genesis story with that? Yeah, I mean I've I've been interested in this space for a long time, I think. I think it just goes with my interest a and like emerging technology, trying to figure out you know, back in the early part of the last decade, like two thousand and ten S, I felt like I totally missed the the the move to mobile in terms of consumer applications, and I was like, all right, you know what, I don't want to miss things anymore in the future, so let me start to make sure that I'm always reading up on emerging technology and what's going on and what people are building and what people are working on, and then I cover that. You know, I cover that in the local maximum, I cover that on my podcast as well, and also like my interest in sort of just human innovation and individual innovation, maybe I'll put it that way. Like I feel like monopolies are they get us into trouble. I don't want to to Clare like all monopolies evil or anything, but I feel like we have nowadays. Of course we have sort of our social media big tech monopolies that are causing huge problems and it's sort of a one size fits all and that I don't even want to get into those problems. When I say those problems with Facebook, I but everyone agrees ass problems with facebook, but people disagree on what the problems are. And then, of course, you know, there's problems with our money and our financial system. Don't forget, you know, that's the monopoly to so I looked at bitcoin maybe in I don't know, the the early part of the last decade. I wasn't like I wasn't like the first, one of the first people to look into it, but definitely people in in software in New York we're talking about it two thousand and thirteen, two thousand and fourteen, and it was really interesting kind of seeing...

...all the crazy things that were going on and we're kind of thinking, well, what is this stuff? You know, is it is it really anonymous? Is it can someone hack into it? And you know, those questions were I'm people still ask those questions, but those questions were very on on certain back then. Like you know, I now I have a pretty ironclad guarantee because I understand the technology and I see the track record. You know, people still ask, well, why can't someone just take money from my bitcoin account? You could replace it obviously with, you know, a theory Mur or any other currency. Why can't someone just who's mining take something from my account and put it in their account? You know, people don't understand that well how the technology works. And even once you understand the technology, it works like Hmm, I still think there could be a trick to it because if you think about it, you know, even the etherium smart contracts, sometimes there's a trick to it and someone figures it out. I mean, look at the the Tao. What was that two thousand and seventeen where there was a theium smart contract and to grow really big and then somebody figured out how to extract the money from there. So it's it's, it's it's got just to think about, like what my thought processes were back in those days to what they are now. It's sort of change completely. I guess I'm so much more confident now in the space. It's so much more stable, even though it's maybe looks much more unstable now to people from the outside. Yeah, that's that's pretty cool. Seve ben watching it for quite a while. So it's now, you know, past the next having and so I guess the big deal. Time to look at bitcoin again. So I get the sense that you're looking at bitcoin a little more than a theium. Is there say, a reason for this? Well, I don't I wouldn't say that I'm looking at bitcoin more than the theorium. I mean, the first crypto episode I had in the local maximum was episode five with Christian Lunkvist, who is, you know, one of the smartest engineers in Atherium, and we talked all about smart contracts and I think that it's I think the only thing that I'm kind of waiting for is some application that I can kind of under stand and wrap my head around something. I can't understand the applications. It's just I don't see examples of people using them in the you know, in more than kind of an academic setting or more than the trial setting. But I really do see people using bitcoin in terms of just trying to preserve wealth or, you know, just as an investment. You know a lot of people who wouldn't have touched this stuff many years ago. Are asking you know, well, you know, can I invest it in the you know, in GBTC? Can I invest in Greyscale Trust on the stock market? And you know. So I just see more. I feel like people need to get bitcoin first before they can go in a theorem. That's maybe the order that I went in and that's the order that most people went in, because if you can't get how the main cryptocurrency works, then you're not going to get very far beyond that. I'll certainly, I mean I'm I'm a big proponent saying that. If you're going to look at this stuff, I think reading the ten page white people, which is still one of the best pieces of technical literature out there, even if you don't understand that it's very clearly written, really does help. But you mentioned something about a theory more, you know, academic setting key. Dive into that little bit more. Well, I mean I just think maybe I just haven't seen it, but I haven't seen any applications that have blown me away just yet, although I understand the idea behind the smart contract. I think there are going to be lots of applications that are widely used, but I just haven't looked into it for recently, like I mean, yeah, so there's definitely, I guess, like the the the response to that that I'll give for most of the thing community is defy. Why aren't you? Yeah, defy big big word right now. Yeah, it's big word. It's it's clogging up our the blockchain. So people are definitely using it. I can. I can see like my back, like I'm not I don't know, multiple layers of financial derivatives or a bit beyond me, and I'm starting to find them interesting for more just from its design perspective. But I'm curiously if we, you know, thinking about real world use cases for smart contracts, like is there something you know that you can see four square doing with it, making use of it, and maybe for background you could you could also just tell us a little bit about like what four square is up to these days and was...

...focusing on. So I could tell you a little bit, a little story about four square, but I yeah, at the end of the story you might be disappointed, but four square start it out as a consumer application company and you would you go around the city or wherever you live and check into places and tell your friends where you are and you get little rewards for doing that. And when I was first hired in two thousand and eleven, I was working out the recommendation features where with, you know, kind of recommend good places for you to go. We have some very sophisticated tech behind that. Now we've moved more to being a enterprise software company where we built a bunch of software for these for these APPS. There's some software that kind of lives in these APPs and sdk called pilgrim, which sort of takes the information from your phone and translates it to, you know, the location where you are. So it started out being like, okay, can we suggests to people where they are so they don't have to go search for it, and then eventually it's just, you know, maybe people will just want to be we just automatically want to know where people are so we can feed them good recommendations, and then we look for other apps that were it makes sense from a consumer standpoint to have that in and then we have our location panel. But in terms of so one of four squares successor APPs today is swarm and that's where people kind of check in and they get like little coins and little rewards for it, and so you have account of coins and there's a questionable what are people going to do with these coins? I mean all these coins are are just numbers in a database. I. And so you can imagine that probably as early as two thousand and fourteen, two thousand and fifteen, people were like, let's turn this into a cryptocurrency. The problem, well, you can trade those coins now, but the problem that we had with that was, you know, first of all, we don't have any particular release schedule of these coins. Is kind of like the developers are like, Hey, I want to give people a reward for this. It's kind of a gamification thing. I should be able to give out coins. But if we did that in the future with what? If it's a cryptocurrency, we're going to have to think, wait a minute, how much money are we giving away? And so it's not about fun anymore. And then another problem is that, you know, all of a sudden it's no longer a game than that means that people have incentives to like, you know, spam the system and start, you know, undermining the game mechanics and, you know, trying to figure out what's going to get them the most point coins, not by doing it but by, you know, I don't know, spoofing their lat long and spoofing their Ip address and all that. So and then, you know, never mind the like tax nightmare and SEC nightmare. So I think we decided regulation nightmare. So I think we decided pretty clearly. You know, this was not a good idea for a gamification aspect, although I think there might be some games out there. We're a a you know, cryptoization of their coin might be a good fit, but I can't think of any right now. There are huge turtles to get over. That's really interesting. It seems very similar to this sort of current Web free paradigm, but maybe in the terms of, you know, back in the day, make something more like an open value network without enough sort of definition. And of course I could totally see this being open to Cibil attacks or, you know attacking and potential tcrs that are maybe built into the built into the platform. But I guess then the it would be within its own ecosystem. I guess if that was to be put in to today's thinking, you know, ten years later, it sounds like a little early. Would you change anything, or do you feel like it's still would suffer the same problems. You mean if I was starting a four square from scratch today, the swarm aspect of it or the starm aspect from scratch? Say? You know, I don't know. I think if I were building an APP in two thousand and ten I'd be like, let's build an API and, you know, let's put it out there and we'll grow rapidly and get funding and all that. That's not the world in two thousand and twenty. You know, APPs can't just explode like they once did, or at least I haven't heard of one for many years. Very saturated market. I think that today it's a good idea to think in terms of security first, and it's a good idea to think in terms of I don't know, how could you know anti spam?...

It's sort of like build into the system, like what can someone what can go wrong? I mean, I guess. I guess there's still a sense of in terms of the team that I'm in now. Maybe I contradict what I just said a little bit because, you know, we like to put out stuff for fun and see what people are are are doing with it. But I feel like today there's so much more. It's so much more exciting to say, okay, I'm going to work on a decentralized project, whether it's like bitcoin or using the etherium channel, and somehow, you know, somehow making use of the decentralized architecture is so much more interesting than the just usual kind of database and APP that we have now. But I I don't know, I feel like I'm all over the place. Helped me out a little bit here that maybe it was a it was a user problem in that there wasn't enough saying people using it, creating an enough liquidity to actually keep these kind of economic systems afloat in you know, walled off ecosystem right. I mean the question is it is a game. Is A game really an economic system or is it something or is it kind of like a simulation of an economic system? You know, because if once it becomes for real money, it's sort of changes it, it's sort of pushes it into the economic realm and then you have now it's not just about playing the game and enjoying the game. Now then it becomes about making money and that changes the whole the whole aspect of I wish we could give rewards for like, you know, a big part of four square at that time. It's still a part of it, although now we take data from lots of different sources, but it was, you know, about building the venue database, getting lots of making sure that we have all of the information and all the businesses and locations in the city, and that's largely crowdsourced. It would be good if we can give people rewards for getting. You're forgetting giving accurate information. But then again, you know, how how do you decide what's accurate and in a how changible? If it's yeah, how? Yeah, it's, and it's usually something that very few people can check. Like if there's some obscure landmark in a park and I say it's there, there are very few people who can check that or it would be very costly to check that, so we usually wouldn't. So then there's it's problem. Basically, yeah, we deal with here. Yeah, yeah, we're getting it. Yeah, what I was hearing is like, you know, you talked about it not wanting to make it about money, but but you do have like some kind of in game currency or reward system. And what was certainly, if you think, is like well, yeah, but people are playing like they're they're not made the playing for like whatever, like gems or whatever they're playing for. Right, it's still basically about some kind of strange, abstract reward, right. And since they're helping you, like, since they're helping build this database, should you not like want to incentivize them and actually reward them for that in a way that can be brought into the real world? Oh, yeah, yeah, I I would want that and I I think it would be a tough problem to design that. Like, I do think about this sometimes, like how could you design a crowd source databased where people are financially incentivized to provide the truth, to provide good information? And maybe, you know, some how we can. You know, you could get paid a lot if it's important information or if it's like, you know, information that other people are trying to undermine, or and you maybe don't get paid a lot if you're giving us information that nobody's going to want. But when you design that sort of thing there are a lot of like every time you have some kind of design, then you jump to the other side. You think about, okay, how can I exploit this to make the most money possible, and oftentimes the answer is doing something that you don't want done with your database. So I think to design something like that you'd have to really sit down and really think carefully about how you're going to do it to unlock that value. I mean, you know the bitcoin itself, the blockchain itself, at that ten page paper does it for money, exactly right, and it's worked, but it's it's not an easy problem to carry over to other domains. Do...

...you think it's hard when you even take the assumption that markets and everything is ultimately absurd and an anyone is ultimately able to probably do many things? Or is or does it? Or does that make it? Does that not change the design problem? But I'm sorry, what do you mean about markets like certain if there if there's many built in incentives in a reward structure and ultimately someone is always going to find a way to exploit it, right, and you don't want someone doing that with your database. Is is is that the problem? Or is it the problem that that it's always really hard to assume how people are willing to think when they enter into a system? Well, now, I think it's always a good assumption that you're going to have crazy spammers and you're going to have people who, like you know, attack your system. But it's I think it's I think it's a primarily a question of the economic incentives. But yeah, and we're going away again to now like a sort of crowdsource database versus a game. Those are like two different things. But but yeah, when you took come to the crowdsource database, I think it is a matter of incentives and I think that anything that could go wrong will. I mean Murphy's law applies. We've seen some spam and our systems that it's like no one can explain why someone is doing this and we still don't understand why someone is doing this, this crazy thing, and it's just people. I mean that's the beauty of permission list innovation, but it's something that's going to happen and if you have a centralized system in particular, you have to you have to be careful of that. Yeah, you mentioned sort of game and this and incentis winning games and we see a lot in video games loot crates and that sort of stuff. And I guess one of my problems with it is not that it's an interesting sort of rate limited token mechanism, es centrally right and incentil mechanism is it's very hard to exit the system with the value that which you take and then put back into the system. So you can't get there's it's almost like there's no reason to to, you know, have that value that works in your in your everyday life. And maybe that is is one of the fundamental flaws. So so it's walled off, like you can't most games. You can't walk away and be like, I play the maybe of the online poker you can, but for some game that you can't walk away and say, Hey, I made a thousand dollars today, I'm going to I'm going to take that out. You know, for most games that's not the reason why people are playing it. But if you could, then people would start turning that into their careers. Yeah, and one thing I just want to quit me mention is interesting how, when online video poker start to come about, there was this new game style that started because everyone's just getting all these like tokens and they just like put it all on their first hand. And then when you actually play real folker, people don't play like that. Except you did see a purity of time where people were bringing that style play onto the tables. And then now it's sort of falling out of favor again. Interesting, interesting, so Max, but like where would you where do you think that we should take this? You know, you've got a the huge following and the hashing it out podcast. That was kind of a joke. I actually don't even know what the numbers are. I know for Bess. So actually it's like they go up and down and I don't know who's listening, but I heard. So you have like a captive the focus audience here. Like what do you think that we should talk to you about that would be most interesting to them? Um, well, let's talk about maybe we could talk about a little bit about, you know, my coverage of Bitcoin and cryptocurrency on on the local maximum on the podcast. There was one episode that I did last year that was particularly interesting. I was proud of it. It was kind of hard to follow, but I was sort of proud of myself for diving into the numbers where there was like this this meme or argument going around that all the cryptocurrency mining is going to take over all the energy in the world by this year or whatever, and I at first I thought that was ridiculous and now I...

...definitely think that's ridiculous, but I wanted to kind of look into it a little bit because people were actually believing this, and so I covered that on the show. But basically what I did I sort of said, okay, you know, you have these block rewards and basically no one is going to invest more into mining than the block rewards themselves. So I said, all right, how much are the block rewards? And what if the cryptocurrency markets get to this enormous price that, you know, the highest price that anyone's ever predicted? And then I work backward from there and said, okay, that wouldn't be that bad. That would increase energy consumption around the world by one or two percent, which is a lot. But then you'll just have a half heavening after four years and then that'll cut in half and then another happening after four years and that will cut in a half again. So so yeah, I felt like I sort of put that argument to rest with that episode. So that was a fun one to do. What was the the highest possible value you were working with? So that's an interesting one. I used one. I think it was how Finny said in two thousand and ten that the bitcoin takes over the world it's ten million dollars. And my my biggest assumption was that. And let's say, I don't know, it's it's let's say that that that that also includes all of crypto. Okay, so I I sort of use that number and I said, okay, let's say it's that number in twenty thirty, you know, in ten years from now. And then that was the case where energy consumption. If assuming that, like people use the entire block reward on energy consumption, which is also a very dubious argument, people are going to invest a lot less because there's so much risk there and you know, maybe a few people will invest more than they should, but I don't, you know, it's if it does, it can't go on for much, for like too long. So yeah, so I use that and then I use twenty thirty as the as the year. I mean, that would be great. That would mean that in ten years crypto will have taken over the world. Or I don't know if that would be great or if that would mean that the world is in a very entered, a very dire state at some point in the next few years, which doesn't seem out of the realm of possibility either. But anyway. If that happened, yeah, energy consumption on that would increase one or two percent over current global energy consumption. But then once you have the having it be, you know, a half a percent and then so on and so forth. So it would be a very temporary situation. Another thing that we can talk about it I'm a little bit look, jumping around a little bit here, but I want to talk about the APP that we're working on at four square right now that we're going to put out, you know, in the next few weeks, because it's not really cryptorelated, but it's I'm excited about it because I've been working on it for a year and really we had it ready to go in March and then the world got shut down and we couldn't launch it, and that was called. It's called Mars Bot for Mars about audio. I think we're going to call it Mars Bot for are pods, and basically we sought everyone walking around the city with their air pods in and we can tell when someone walk walks past a particular store front, and the idea was, you know, let's play some audio when people walk by different locations. And so the thing that I'm site excited about, and I'm even excited about the spam here is you're going to be able to record audio and then everybody else with the APP is going to be able to hear that audio as they walk past a particular location. So there's going to be a lot of city tours and kind of like this. I just hand over, I just say hey, hey, iphone, I just want to I want you to play whatever, like whatever, some like this, this and this owner wants me to hear as in lor well, I mean, when we first put it out, it's only going to be a few hundred people using it, so it'll essentially be everyone, like you'll get really excited when you hear someone's thing, because it's going to be a very rare occurrence to start out. But it wouldn't be the business owner. It'll just be like, you know, anyone, and I think what we're going to do if it gets to be too big of a group is, you know, you'll have to subscribe certain people's channels, just like youtube channels, where it's like hey, you want to subscribe to my commentary on each individual tree in the park? I mean that's what I did to test out here in for Green Park. I did a little commentary on individual trees and I sort of was getting started to get starky about it. I was like you're at another tree again and you know, so you can kind of decide which types of content you you'd...

...want to hear. I think people are going to do interesting things with that in terms of like both city walking to tours and also maybe, I don't know, I feel like somebody could do something cool with like sound effects or you know what. Well, you're hearing like historical speeches and stuff as you walk by certain bits, certain places where those speechers would delivered. That would be super cool. Like what I would I so when earlier when I asked like, you know, looking forth Guargor with Crypto, my like my throughly the answer, I was the right answer that you miss but now you've got it is nonfondible tokens. So like what I what I want to do is I want to be able to by the corner of some school, some street or like the mellow park on the blockchain, and whoever owns owns that nonfundable token representing the mill of this park, they get to like set the the SNIPPET, but everybody has to listen to when they go there. I think like that. That seems like when I hear like the whole geolocation thing, nonfundables really come to mind. A traction. Yeah, I don't yeah, that's probably that's a good idea, but right. Yeah, the problem is every time we look into it it always seems like, you know, just implementing it in our database, as we normally would, is a better solution than having working in the cryptospace, which is disappointing for me, is someone who's interested in the cryptospace, but I feel like if I were to do something in terms of geolocation, it would have to be probably separate from from four square, but we'll see. I mean one interesting problem, the technical problem that is I think it's kind of solved, but it's very complicated, is like proof of location. It's a very complicated problem to solve, like can you prove that you're actually at a location? Because we've had we've gone through to down the rabbit hole of Oh, okay, well, you know, we'll post a you know, we'll post a qr code, a different in places. What's going to start me from taking a picture? Yeah, location is extremely hard because anything real world trying to attest to, you know, some sort of database, much less blockchain. Still requires some sort of transfer or or some sort of sensor network. But you know, you have to assume there has to be a reliability of its being correct. Maybe, you know, maybe we need to bring back reputation systems here. But there that that that is a hard problem. And I mean I guess four square these days is more of a date of provider, so I could see it being useful. But then, you know, I guess is the goal to then try and build a bunch of little APPs that incentivize the sort of game mechanics to ultimately provide these sort of sorts of data. But how do you then attest to its correctness? Yeah, I mean, well, look, we could let's talk about like location, location technology in general, because I don't know if this is the direction that four square is going to go down and I don't have that much influence. But you know, proof of Lokate. One thing I want to say about proof of location, and then I want to get back to your question, is just that, you know, anything that you come up with is you can always just figure out a way to like, you move it to another location. You almost have to get a code at that location and then send it out immediately and sort of prove that, okay, I only could have generated this code from, you know, these devices that are external because they came at me at the speed of light. Then I transfer them out at the speed of light. It's it's some crazy, crazy stuff, unless there's some like, you know, ingenious plans do proof a location. That that that that we're missing, but it's very difficult. And then back to your question. Sorry, what was the last thing that you said? I mean, I was just commenting that it that it's a very hard problem, you know, and if we think of this as a sort of data potentially going on to blockchains, you know that in itself is is like, okay, bringing back to we talked about earlier about bitcoin and mining awards versus speed market. Well, well, all this data doesn't necessarily make sense. If you're trying to have some sort of even if you want, even if you want provable or mostly provable location data, then okay, you can secure it on on some sort of blockchain, but...

...then that's tons of data. I guess you know what kind of it leads me to a question of from what we were talking before. How do you see layer to potentially eating up the fee market on beyond Bitcoin? Yeah, so we're talking, I assume, primarily about the lightning. Lightning working. Yeah, but there are others. Any call yeah or yeah, or even just a side chain thing, like I mean that was another question about that I got about, you know, about bitcoin mining eating up the world's where they're like, well, you're only using the blockchain rewards for to decide how much miners are going to pay. But there's also the there's also the transaction fee, and so the miners are going to be working for the transaction fee as well. And I'm like yes, but the transaction fee, somebody is actually going to have to be willing to pay that transaction fee to make their transaction and I don't think that people are going to be willing to pay trillions of dollars and aggregate to do transactions because there are so many solutions. So then that, which is that pushes so those transactions to layer to right eats up the fee market, which then is is downward pressure on on miners rewards, if there now, doubt if there's already, the block reward is so hard to the difficulty so hard that that's already downward pressure on the miners themselves. Yeah, yeah, the the block reward is still is still significant. I mean still, you know, most of what they're working for is the block reward now, I believe. But yeah, I think that they'll be times when transaction fees are very low and people are just going to go on chain as much as possible, and then they'll be times when transaction fees are very high and people are going to work on off chain technology. I think they'll just be like kind of the price of gas. You know, if the price of gas is high, maybe people will buy more efficient cars, they'll take less rides and things like that, or you use less energy or and if it's more than people use more injury. And so I just I think that, you know, people have it's sort of a slow moving technology and by layer two, like off off train transaction, that could mean a number of different things. That could mean like the development of a lightning network. It could just mean, you know, take something like coin base for example, where if you send cryptocurrency to someone else on coin base, they don't necessarily have to do an on chain transaction for that, which is you know, from a pure standpoint, not how you're supposed to use cryptocurrency, but there that's what people would do. But that's further downward pressure on on them, the on the fees itself. I mean I think there's a want to be enough transactions to satisfying miners, but I think that, you know, in my opinion, I actually don't think we really know what's going to happen. I think there's there's a lot more factors that were missing and I think taking a purely austen economics of view, in my opinion, other people on the show, let you listen. Other listeners will probably have a different opinion than me. That is sort of one of the undoing aspects of it. But that's not to say that these sort of austen economic models don't have place in a in a larger system. But anyway, just my take. I do think layer two is is very important. I think there's going to be ebbs and flows of it and yes, I think it's it's inevitable, but I don't know what it looks like. I don't know if it's the lightning network, I don't know if it's some confederation of side chains. I don't know if it's, you know, some centralized system like a bunch of coin bases. I mean, I guess I hope not, although I have used it in that way, but it could be. Could be anything to shift a little bit. Max. See Your your podcast, the local maximum, is about machine learning, an AI rite. Yeah, I cover machine learning, a ai a lot, but more generally. I mean, look, if I'm going to do a podcast every week, I'm basically like, let me just talk about whatever I'm interested in that week. Yes, otherwise what's the point, you know, what's the point of having a podcast? So I've branched out quite a bit. I started by kind of just interviewing friends of mine at work that are working on cool projects and asking them about it, and then I've branched out from there to talk about all sorts of you know, topics and emerging technology, what's going on in the news, and I sort of I try to bring machine learning concepts into it from time...

...to time, including the local maximum. But a good example is like one concept that we have is underfitting and overfitting, and it's like, well, what's an example of an underfitting person versus an overfitting person? You and overfitting person is always going after the latest fad and whoever the last person spoke to them, they're a hundred percent on board with the last person they spoke to, and an underfitting person just is totally stuck in their ways and never changes their mind with new information. And so that's sort of the wisdom that we get from teaching machines and that's also a concept that you could use to kind of talk about teaching people as well. And so I'm really interested in those types of topics and we try to explore them a little bit as well. Yeah, I like that. I like that underfitting overfitting thing. It's totally a real world, useful mental model. Yeah, so, but thinking about, like you know, blockchains is being just large data sets, I'm curious what you think the the the biggest opportunities are at the intersection of machine learning and blockchains. Yeah, that's a that's a good question. I mean the the pessimistic one kind of scares me, which is, you know, are people going to use machine learning and statistical systems to sort of undo the privacy, the privacy features of the of the cryptocurrency universe? Yes, the answer is yes. Yeah, yeah, and and to what extent will they be successful. They will also be people who try to, you know, get around that as well. So that's I don't know, I'm a little concerned about that. I probably won't work on that personally, but I think there's, HMM, a lot of I'm trying to think of some more positive things that that can be done. I mean, you know, certainly in terms of routing transactions, and this is getting into like the the very low level technical stuff of it. But again, like, let's say I have block chain wallet and I want to do a transaction and I want to figure out what's the cheapest way of doing it. Well, I mean you could have a simple algorithm that maybe compares several ways, but if that space of ways of transacting becomes well, you know, it's each way is probabilistic, I don't know how much it's going to cost and there are hundreds of thousands of different ways. I can go here. You could build an algorithm that is statistical nature, that could, on average, get you the the lowest cost transaction to the fastest transaction. So that's another idea. Could you build? Interesting question is the could you build a statistical algorithm or like an or machine learning algorithm into the blockchain itself, into the etherium blockchain, and I know people have tried, but what are the applications of that? Is I still am I'm at a loss there, but I feel like something, something cool, has got to come out of that. Yeah, it's I mean I'm a bit with you there. It's hard to think of like what the what the cool stuff is. Mostly I just think of like hypercharge trading models and Better Arbi charge thoughts. But yeah, oring I've seen some like so one episode I had a friend of mine, Dan Pronavette, on the show and he talked about some of the stuff that he was working on at his company and one was like a kind of a budget setting algorithm where people would vote on where to allocate the budget and it would actually run some statistical algorithm over that and then automatically allocate the budget based on those votes. I don't have enough background right now to like explain how that happens. It's not as simple as like adding up the votes. There's some like Game Theory and social choice theory involved, but that seemed pretty cool. Thanks. I wonder, really I'm always I sort of wonder if there's like some way that the design space and include like enable scaling such that. Well, so what's longer put you can light like this is this is I'm super naive about machine learning, but are you know, is it possible for machine learning algorithms to be executed in a deterministic fashion so that they could perhaps be at least like verified on chain?...

So, in other words, yes, so there are some algorithms that involve random number generators and there are some that don't. The ones that do, though, you could always set the sea eat off the random number generator and so it's sort of deterministic which numbers are going to pop out. And so you're saying have someone runs some algorithm. H just say it's like logistic aggression, or someone gets some parameters off offline and then publishes them to the Blockchain, but it needs to be confirmed that those are the parameters that you actually get when you execute the algorithm. I suppose you could do that. I don't see what the application is, though, right now. This is one area where I am I'm disappointed in myself right now. I'm like I'm lacking the creativity in the vision to come up with one, but I just I'm at a loss here. I actually like something I thought about earlier. So Jai, I actually we both work in security, which, you know, is you brought the Doo beginning. Yeah, and so, you know, findings and stamping out bugs is is what we think about a lot. And and there's also we've seen people recently like playing around with gpt three and actually getting it to write like or identify some super simple bugs in boring smart contracts, you know, kind of like play toy contracts with set up just for learning. Do you know anything about you know, like security analysis using that's been done using existing machine learning techniques, or very little? That's a little outside my my area, but I do have a question for you. So you've looked at the Tao and so do you think that? Do you think that in the future these types of, you know, massively multiplayers smart contracts are going to be a thing and this is just kind of a blip where it's like yeah, you have to be really careful about how you write that, or do you think they're just unworkable? Good? I mean, I think we both have different answers, but I think it depends on a how you look at massively multiplayer and you know, I think on aggregate there's going to be a lot more of these used. And I mean how you define a Doao is is dependent. Right now it's being defined as a funding mechanism, but that's not necessarily what it's always meant to be. I think there's going to be a lot of these little sort of micro economies that output something that's generally universally usable, that has some sort of intrinsic wish to participate in. I think right now we're seeing a lot of that in these defy systems and we're seeing a lot of that and very interesting economic games. I just I think we're so a little white ways off to finding like where those really take hold. But I guess the short answer is yes, but the long answer is, wow, there are so many vulnerabilities that that are that get missed. unlet's almost like you have to see a whole new standard of like, you know, rigorous code testing. That is largely unheard of from today's software developer. Yeah, absolutely, I mean like those those techniques and methodologies are out there. Like something I used to hear get battered around more is, you know, people would start ranking smart contracts and they would say oh, you know, like this is this is mission critical. We have to like be like NASA. But then they just start logging stuff at the wall and kind of like free writing smart contracts, maybe writing tests later, not not really testing the whole thing, testing the happy paths. So it feels to me like it's I think it's absolutely possible, within the right constraints, you know, keeping your contract simple, using verification methodologies and you know, we recently talked to Ethan from informal systems, who's actually pioneering a techniques called verification driven development for some form of a verification and other like fuzzing and symbolic execution and stack analysis tools. Like I think the space is really really pushing that stuff forward because of the incentives. But even as like the security...

...community is creating those things, we feel like we're sort of well, like shouting into a void kind of and and to me it feels like it's not a technology problem at all. Really it's a cultural thing and this fit the space is full of like it's early adoptors who are excited about things, people would probably were working at startups five years ago. Yeah, that's some break things and it's and shifting that mindset is is really, really hard, especially when you are effectively operating a startup and and your CEO as some cool feature ideas that he'd like to see out in the next two weeks like these. Yeahs right, these are not, you know, airplanes to hold many people like these are not just like a a little meat up spot. Exactly. Yeah. Now, I totally identify with that because my sort of my sort of go to, well, like you know in the in the group I'm in now, is to move fast and break things. I like just throwing things up and seeing it work and and keep going, and that's that's the way that I like to work. But I know that if if I'm working on something and something goes wrong and like a hundred million dollars goes missing, I know that I'm not going to I would decline to work on that. I would either have someone else do it or I would be able to change my my my methodology a little bit and sort of figure it, figure out a way where I can fit in while doing that, or find some section of it that that that I can work on. Would that's that's maybe less mission critical, but I see why developers get into that. That mindset of, you know, just throwing things up, or it's not just developers, such as engineers, as the whole startup mindset, which works in a lot of situations, but it could be it could be very scary or very dangerous. And situations like decentralized organizations certainly, and you know, I think we need to remember is decentralized organizations and decentralized interaction is something that a lot of people are really not used to. Right. It really does take person's ability to see some things amiss and and pitching in a way that that is potentially useful and productive, and one hopes that the growth of this whole thing is an underlying incentive to do so, but that's not necessarily the case. And then you you layer on top of that this sort of startup culture and start up mentality and it has a recipe for potentially leaving a lot of people with a bad taste in their mouth. Yeah, yeah, I have to agree. Well, I think that's probably a good place to tie things up. Thanks so much for talking to sit in Max Working People find out more about you and follow you. Thanks. Yeah, well, definitely subscribe to the podcast. It's the look called the local maximum and you could even go to the website local Max RADIOCOM and get the entire archive, or you could just go to your podcast APP and basically every week I try to either talk to someone interesting and entertain you a little bit and make you smarter. So that's my goal. And we learn about emerging technology, we learn about crypto in some cases, and you'll learn a little bit about machine learning, and so we have a great time and you know, that's that's basically what I like people to check out. I'm also on twitter and Linkedin and all that good stuff. It's very easy to find me. Thanks for Kinda thank you so much for having me.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (127)