This is episode number 621 with Philip Gradwell, Chief Economist at Chainalysis. Today's episode is brought to you by Datalore, the collaborative data science platform, by Zencastr, the easiest way to make high-quality podcasts, and by Bunch the AI-driven leadership coach.

00:00:22

Welcome to the Super Data Science Podcast, the most listened to podcast in the data science industry. Each week we bring you inspiring people and ideas to help you build a successful career in data science. I'm your host, Jon Krohn. Thanks for joining me today, and now let's make the complex simple.

00:00:53

Welcome back to the Super Data Science Podcast. Today we have the brilliant Philip Gradwell on the show guiding us through data science applications related to the blockchain and cryptocurrencies. Philip is Chief Economist at Chainalysis, the world's leading crypto analytics firm whose analysis is regularly featured in major mainstream news outlets. Previously, he worked as a Principal at Vivid Economics where he helped grow the consulting firm to 40 people, eventually culminating in the firm's acquisition by the consulting giant, McKinsey. Phillip holds a Masters in Economics from University College London and a PPE degree, that's Philosophy, Politics, and Economics from Oxford.

00:01:33

Today's episode will appeal to anyone looking for an introduction to the blockchain and cryptocurrencies. It'll hold special appeal for people keen to do data science with these technologies. In this episode, Philip details similarities and differences between analyzing cryptocurrencies and the established fiat currencies, his own crypto data analytics pipeline, how he develops data products for a wide range of users, including businesses, banks, governments and law enforcement, how the blockchain facilitates innovative computing and machine learning technologies, and what he looks for in the data scientists he hires. All right, you ready for this highly educational episode? Let's go.

00:02:16

Philip, welcome to the Super Data Science Podcast. I am stoked to have you here. I've been dying to learn myself about the blockchain and crypto, and now you're here you can elucidate everything for me and maybe even some of our listeners too. Welcome to the show, Philip. 

Philip Gradwell: 00:02:32

Thanks, Jon. I'm really excited about it as well. Just got a small set of topics to cover. 

Jon Krohn: 00:02:37

Yeah, we wrote down so many topics that we'd like to cover before the show, before we started recording, Philip and I lined up so many topic areas. This is going to be so much content for you listeners, and Philip is this amazing source of knowledge. I've known him for so long. So we met, let me do the math here, 15 years ago, I guess, because I started at the University of Oxford in 2007 and I think you were already there, right? 

Philip Gradwell: 00:03:05

Yep. 

Jon Krohn: 00:03:06

So you were doing an undergrad at that time and in PPE, which is a really interesting degree. I don't know if you want to tell the audience about that quickly? 

Philip Gradwell: 00:03:16

Yeah, so that's Philosophy, Politics, and Economics, which I kind of use as an excuse to do as much philosophy as possible and had to do some economics because I thought I needed a job at some point. And, I guess, that's how I ended up here. 

Jon Krohn: 00:03:30

Great. And then we've kept in touch over the years. It's been brilliant to watch your career blossom, and so I've known for the last five years that you've been Chief Economist at Chainalysis and then every time I come across Chainalysis in the Economist, which I read every week, I'm like, "Wow, Phil is really doing something amazing here," because the research that you're doing, the data that you dig up on crypto in the blockchain is making front page news in major international magazines. So I've wanted to have you on the show for a while and now finally here you are. Philip, where are you in the world? Where are you calling in from? 

Philip Gradwell: 00:04:11

So I'm currently in Brits in France, which may not be a place many people know, but I'm being a digital nomad for a bit. So I'm normally based in London, but currently on the west coast of France, just north of Spain. 

Jon Krohn: 00:04:24

Oh, that sounds lovely. I guess if you're being nomadic, you don't go to places that aren't lovely. 

Philip Gradwell: 00:04:33

When you got the choice. Yeah, you travel. 

Jon Krohn: 00:04:36

Awesome. So we were also supposed to be joined, in this episode, we were planning on having two guests on this show, which is quite rare. Since I've been host of the Super Data Science podcast for the last two years, there's only been one other occasion that we did that, but we thought that this would be perhaps another one of those interesting occasions. So we thought that we would have a Chainalysis colleague of yours, Kim Grauer, so she's Director of Research over there. And so she was going to talk about things like statistics on criminal activity in the blockchain or in cryptocurrencies, and she was also going to talk about crypto adoption statistics in general because she's the author of a very well-known crypto adoption report.

00:05:19

But Kim wasn't able to make the recording today in the end, and it actually, that ends up being a benefit to you listener because we're still going to have Kim on the show. We're going to talk about those topics in the future, but we'll be able to dive really deep into them with her and we'll also be able to dig really deep into what she does as the Director of Research at Chainalysis. And so that leaves us being able to dig deep today with Philip, who is the Chief Economist at Chainalysis. So Philip, what does a chief economist do in general, as well as specifically at a company like Chainalysis? 

Philip Gradwell: 00:05:57

Yeah, and so just to say people really should check out the episode with Kim. She's fantastic, will give a huge amount of insight. Kim's also an Economist as well, and we started working together and I think we must have flipped a coin and ended up with me as the Chief Economist. So yeah, what does a chief economist do? In general, a chief economist at a company is there to help the company, and often the leadership, navigate where the company is going and how to position itself as there are all these big macro changes. So in my pre-Chainalysis life, I did a lot of energy, economics, and climate change economics and I spent a lot of time with the chief economist of Shell and their job was to go, okay, oil prices might be this high or there's this climate legislation coming in, so we're going to have to change which places we go to search for oil and gas.

00:06:49

So they're trying to look into the future and understand how that might pan out and how prices for the company's goods and services could change. My job's a bit different because, I think maybe I have the title of being the first chief economist in a crypto data company, so we sort of had to make it up as we went along and so I did a whole range of things. When you're a small startup, you just help find offices, help set up teams, all those kinds of things. But really my day job has been to try and take a bigger picture look at all the data that we have and to go, how can we count all of the activity that's happening on a blockchain? How can we actually measure that activity? And in more recent years, how can we build products out of that understanding? 

Jon Krohn: 00:07:41

Super interesting. I understand that you do a lot of product development leadership in this role, yeah? 

Philip Gradwell: 00:07:48

Yeah, that's right. 

Jon Krohn: 00:07:50

So we will dig into that in a fair bit more detail later in the episode as to what that means to be leading development of a data product, but before we get there, I'd like to provide the audience and myself with more context on Chainalysis, as well as blockchains and crypto in general. So my understanding is that the mission of Chainalysis is to build trust in blockchains with crypto compliance and investigation software. So let's parse all of those words for our listeners. What are blockchains and crypto? 

Philip Gradwell: 00:08:26

So blockchains are like a shared database, they're this common ledger that anyone can contribute to. And say you had a normal database where anyone could just write onto that database, it would be chaos, you would never know what its latest state is, et cetera. And so blockchains have a consensus mechanism that agree on what the current state of that database is, so that's in a sense all it is. It's a database that anyone can edit and then it has a distributed consensus mechanism to make sure that, that database is correct. 

Jon Krohn: 00:09:07

Nice. Okay. That's crystal clear. That is perhaps the clearest explanation of blockchain that I've ever had. So yeah, just a database that anyone can edit with a distributed compliance mechanism to make sure there's consistency across those data. All right, so that explains what the blockchain is, Philip, then how is crypto related to that? 

Philip Gradwell: 00:09:28

So people need to do a lot of work to make sure that database stays consistent and they need to be rewarded for that. And so Bitcoin, you might have heard it called proof of work, so there are these miners who are running computer algorithms to say, "Yes, this is the actual state of the blockchain," and they have to spend money. Those algorithms cost electricity and hardware to run, and so they're rewarded with a Bitcoin. 

Jon Krohn: 00:09:56

Got it. Okay. So you can have all manner of blockchain out there, it's a database that all these different people around the world are continuously running computations on to verify the current state of what those data are. And, as you're saying, all of these computations require electricity to run, a lot of people are using GPUs, graphical processing units, which are also popular for, I mean they were originally created for rendering graphics, particularly complex 3D graphics on a computer, but they've turned out to be very practical, they're capable of doing thousands of parallel simple linear algebra operations like matrix multiplication operations. And those thousands of parallel operations end up being the kinds of operations that we need to do to, say, train a deep learning model in AI, but they also are the kinds of operations that we can be doing to verify the state of a blockchain, yeah? 

Philip Gradwell: 00:10:59

Yep, that's correct. So Bitcoin is to blame for machine learning people having more to pay for their GPUs than recent [inaudible 00:11:07]. 

Jon Krohn: 00:11:07

Right, exactly. I've had crazy situations where a few years ago I was trying to buy this 1080 Ti GPU, which at the time was state of the art consumer GPU, and I'd have to, while everything else, all other computer components, I could just order them and they showed up at my door, I'd have to find a physical location that's like a 45 minute Uber ride away from, so I live in Manhattan, I'd have to get an Uber to deep Brooklyn to go to a computer store where they'd have a few of these 1080 Ti GPUs and then I would only be allowed to take one out of the store. But I did manage to claw together a deep learning rig by doing that. 

Philip Gradwell: 00:11:56

Nice. 

Jon Krohn: 00:11:58

And it was expensive. I understand it's not quite as bad as it was years ago. 

Philip Gradwell: 00:12:03

Yes. Exactly. So now, obviously, the market, as we're speaking, isn't as high as it used to be. And also Ethereum, which is the other big blockchain next to Bitcoin, has moved to a proof of stake consensus mechanism, which means it doesn't need all those GPUs. So the industry is trying to work out how to get around it. But we've dived into some of the advanced topics, I was trying to do the intro explainer. 

Jon Krohn: 00:12:26

Yeah, I mean I think this is interesting stuff, especially because, so one of the things that made me skeptical about Bitcoin in particular for a long time, as being a ubiquitous currency of the future, is that it didn't make sense to me that the currency of the future would require enormous amounts of compute to maintain. So I don't know what the latest stat is, but I remember reading a while ago, it was like the amount of compute used to process Bitcoin is equivalent to the entire economy of Austria or something. And so you might actually know better where it's at today. 

Philip Gradwell: 00:13:02

Yeah, I mean that's about the electricity usage and it certainly is energy-intensive and that's not a great thing. I think though people should think about, A, Austria's a great place but actually doesn't have the largest electricity usage. People say it has the amount of electric usage of a country and you're like, well, actually some countries are pretty small compared to a US state, but that's by the by. The other key thing though is that energy is not just for making transactions, it's actually for protecting and storing wealth, protecting and storing information. The electricity usage of AWS is huge and the amount of social energy and effort we go into shoring up fiat money is enormous. So there's a really difficult workout what are you comparing it against. 

Jon Krohn: 00:13:52

Yeah, you just used a term though that we should make sure listeners are aware of, you said fiat money. 

Philip Gradwell: 00:13:56

Yep. So that is the dollars that you use every day or the euros that you use every day. 

Jon Krohn: 00:14:02

So they're currency that are supported by federal government. 

Philip Gradwell: 00:14:09

Exactly. 

Jon Krohn: 00:14:11

So crypto is supposed to be completely different from that kind of idea. So you've got fiat money, which is what we've been used to over the last centuries in trading with. And then now over the last few decades we've had this emergence of something other than fiat money for, I guess, the first time, I don't know, you're an economist and you might know the history of this. Is this the first big alternative to fiat money, I mean, since bartering? 

Philip Gradwell: 00:14:38

Yeah, so there were asset-backed currencies, so we used to have the gold standard and the supply of money was limited to the amount of gold there was. So that's slightly different from fiat. Fiat really now is guaranteed by the government and the society that backs it. And then cryptocurrency is backed, essentially, by code and by that consensus mechanism that we started talking about. So we've never had something quite like this. 

Jon Krohn: 00:15:14

Today's show is brought to you by Datalore, the collaborative data science platform by JetBrains. Datalore brings together three big pieces of functionality. First, it offers data science with a first-class Jupyter notebook coding experience in all the key data science languages, Python, SQL, R, and Scala. Second, Datalore provides modern business intelligence with interactive data apps and easy ways to share your BI insights with stakeholders. And third, Datalore facilitates team productivity with live collaboration on notebooks and powerful no-code automations. To boot, with Datalore you can do all this online, in your private cloud, or even on-prem. Register at datalore.online/sds and use the code SUPERDS for a free month of Datalore Pro and the code SUPERDS5 for a 5% discount on the Datalore Enterprise plan. 

00:16:05

Cool. Well, so I was describing this kind of proof of work thing as potentially being a downside to Bitcoin, the amount of energy being used, but then you described how it doesn't have to be that way. You could have another mechanism called proof of stake, which yeah, there's been this huge merge lately happening in the Ethereum space where Ethereum flipped over from being proof of work, more energy-intensive to being proof of stake, less energy-intensive. So that, to me, makes cryptocurrency seem more viable, as a currency of the future it seems like it gets rid of one of the downsides of crypto. And so could you go into a bit more detail as to what the difference is? How is it that this proof of stake mechanism is able to be less than 1% of the energy-intensiveness of proof of work? 

Philip Gradwell: 00:17:00

So the key idea in all of these kind of proof of systems is that you need to make it expensive for someone to attack the system, and that's the core idea. And so in proof of work you have to do more work than half of the other people because you need to have 51% of the ability to do that work. In proof of stake, you need to have 51% of all of the money, all of the Ethereum, or at least the Ethereum that is staked. And so what that means is that people, I'm sort of simplifying, but they put their Ethereum in and essentially say the next block in that chain, that's the right one.

00:17:45

And if anyone wants to challenge them, they need to have more assets and say, actually, we kind of have more votes because you're kind of giving the assets they hold each of them a vote. And that is what secures it, because if they're wrong, if everyone else kind of shows up and says, no, you attack the system, they would then lose what they had put at stake. And so it's all about making sure that the system is hard to attack and you can either do that by expending money or just putting money up. And so proof of stake is just about putting money up rather than actually burning through it in terms of electricity and computation costs. 

Jon Krohn: 00:18:31

Cool. So with these very large, with these heavily adopted cryptocurrencies, like Bitcoin and like Ethereum that probably most people have heard of, it would be very difficult for somebody to accumulate more than half of the assets, all of the Ethereum assets, or do more than half of the compute. 

Philip Gradwell: 00:18:55

And by the time they did that, they would be so invested in the system that they probably wouldn't want to blow it up. 

Jon Krohn: 00:19:02

I see. Because that was going to lead me to my next question, which was then is there a bigger risk of attack in a smaller or a less adopted cryptocurrency? Like somebody could get 51% maybe of some very small crypto thing, but then I guess there's still this situation that you just described is that, well, now that person's really heavily invested, so why would they tank it? 

Philip Gradwell: 00:19:21

Yeah, I mean it does happen and I think, honestly, why does it happen? Sometimes people do it just to prove it happens. Why do people do denial of distributed, DDoS tax on various websites? Hey, who knows? People will do things like that just for fun. But yeah, so it does happen. Sometimes you can make some money off it. Sometimes people are just doing it to test the system. And one of the fascinating things about crypto is that everything is a bug bounty because there's a financial incentive to go and test the system and because you can go and get some assets and so on. And so it's basically the world's biggest bug bounty, it's just running 24/7. 

Jon Krohn: 00:20:05

Got it. Super interesting. 

Philip Gradwell: 00:20:07

That makes the system more secure. 

Jon Krohn: 00:20:09

All right, so we've talked about fiat currencies versus cryptocurrencies. You trained as an economist or as a philosopher, political scientist and economist at Oxford, and then you also did a master's degree in economics later at the University of College London, more commonly known as UCL. And so when you were doing that economics training, did you study crypto at all or was it entirely studying fiat currencies? 

Philip Gradwell: 00:20:39

Yeah, it was entirely studying fiat currencies. 

Jon Krohn: 00:20:42

Yeah, so that's super interesting. So describe that transition. How did you end up deciding that, "Oh, I'm actually really interested in this other monetary system?" And then, I guess, a follow-up question, and I can remember to ask it again if I've already asked too many questions, but what skills that you developed as a traditional economist, as a fiat currency economist, are still applicable as a crypto economist? 

Philip Gradwell: 00:21:19

So I think probably the first thing was my journey into crypto, so I actually did my final exams at the end of my undergraduate in 2009. And so the financial crisis was in full swing. And actually my macroeconomics exam paper, so that's like economic select the whole economy and there's micro, which is on individual interactions, so the macro exam paper said you can answer these questions as if the financial crisis had not happened because it basically, yeah, it's absurd. Like, what? 

Jon Krohn: 00:21:57

Wow. 

Philip Gradwell: 00:21:59

And because it just invalidated the last couple of decades of theory. 

Jon Krohn: 00:22:03

Oh my goodness. 

Philip Gradwell: 00:22:04

And so I spent a lot of time thinking about an economy as an unstable system rather than a stable system because that was basically the only way to understand the financial crisis, as a system that's in this kind of constant chaos and flux and also maybe a system that's open to some challenge. So that was part of that journey. I then just always had a deep interest in technology. I think I've read Techmeme every day for the last 15 years and that's where I heard actually about the Silk Road. So the Silk Road was the first .NET marketplace. 

 

00:22:45

And a lot of people listening might be like, this sounds like an interesting story, but I actually went through it for professional interest because I was an economist and you read that people can go and do some bad things like buy drugs on the internet, but they can do it. And so as an economist you're like, this was the thing that was never possible before and now it is. And this technology, Bitcoin, is enabling them to do that. And so as a student of economic systems, it's fascinating. And so that was the first time that I learned about it and it's genuinely academic perspective. 

Jon Krohn: 00:23:25

We didn't have to do all the drugs that we got on the Silk Road, we just had to buy them as a proof of point. Just like the proof of stake, the people that are destroying cryptocurrencies by taking 51%, just had to see that you could do it. All right, that's super fascinating. All right, and then so I imagine that a lot of what you learn, so there's probably specific economic principles, like you're saying, there are economic principles that were invalidated by the crash in 2008. So there must be other things that you learn though in these economics degrees like econometrics, statistics, psychology, those kinds of subject areas probably transfer to studying cryptocurrency and maybe anything... I mean, to some extent, understanding statistics is going to be useful, I mean, I guess I'm kind of answering the question here, but I know that knowing statistics is useful for answering any kinds of questions related to data. And so that presumably is still faring you well. 

Philip Gradwell: 00:24:32

Yeah, absolutely. And then actually what I'd add to that is the extra things I learned were around complex systems and about network theory or graph theory and actually, so absolutely understanding statistics, understanding how to write equations, a lot of economics is basically writing equations, which you can translate to algorithm design. And then there's thinking about systems as complex systems and understanding network structures because certainly when you come to a blockchain, we talked about that database of transactions, what it actually is, is it's what we call an address, which is a person and a wallet, they make a transfer to another person into their wallet. And because you have the complete record of all transactions, you have this network of transactions. And actually, that was the key thing that drew me to it, is going, wow, there is this complete set of data on an economic network, there must be something fun in that. 

Jon Krohn: 00:25:36

Oh, wow. Yeah, that's super cool. Yeah, I can see why that would draw you in. That's fascinating. Okay, so I think now we understand fiat currencies versus cryptocurrencies, we kind of understand what you do as an economist or what kinds of skills are useful as an economist. So to dig into Chainalysis a bit more, before we get into what you do there specifically, let's talk about what the firm does. So how does Chainalysis build trust with the software they build? So they build software that allows crypto compliance and investigation. Tell us about that. Give us some context. 

Philip Gradwell: 00:26:19

So remember I mentioned that on the blockchain there are all these transactions from a crypto wallet to another crypto wallet. We map that data to real world activity. So what I mean by that is anyone can have as many crypto wallets as they want, and there are lots of businesses on the blockchain and they have even more, they can operate tens of millions of cryptocurrency addresses. And you need to actually understand, okay, are these 10 addresses, are they controlled by a single entity? Are these hundred million addresses controlled by a business? If you don't see that, then you just don't know, you're in this huge sea of just letters and numbers and you don't really know what's going on. So Chainalysis maps all of that data into real world activity, like this is the amount of Eth that is flowing out of Coinbase into, say, Kraken, another exchange. 

Jon Krohn: 00:27:16

Right. Okay, cool. So Chainalysis, so you were describing moments ago how part of what drew you in to studying cryptocurrencies as an economist was that you have this complete ledger of transactions that have happened. And so what Chainalysis does is it takes advantage of this rich data source to be able to provide mappings of who owns a cryptocurrency, where cryptocurrencies are flowing between, what exchanges they're flowing between. All right, I think I get it. And then so how does that help with compliance or investigating things? 

Philip Gradwell: 00:27:58

Yeah, absolutely. So a key thing to understand though is blockchains and crypto are still anonymous. So you don't know that the crypto address that I have on the wallet on my phone represents, it maps back to me. 

Jon Krohn: 00:28:14

Right. It's not like a prospectus for a publicly traded stock where it's like, oh, you know that this asset manager owns this amount of Google stock. 

Philip Gradwell: 00:28:23

Exactly. 

Jon Krohn: 00:28:24

It's just an anonymous wallet address. 

Philip Gradwell: 00:28:26

Yeah. Except for when those addresses are controlled by a business that's an open platform. And so that's a thing like an exchange or actually a darknet market. And so those darknet markets, they all have crypto addresses that they operate out of, and you can map those and then you can trace the connection between that darknet market and that crypto business. And so if you're doing compliance, what you're really doing is a source of funds check. You're asking, is this crypto that's entering my business, where did it come from? We don't so much care that it came through a private wallet, we just care that its ultimate origin was actually this darknet market. 

Jon Krohn: 00:29:12

Right. Got it. Wow, that's super interesting. I had no idea about that. This is great. I'm learning so much. Okay, so then if you're able to track those kinds of things, if you're able to track that some crypto has flowed into or out of one of these dark markets, then what can someone do with that information? I guess, I mean criminal investigators must be interested in that, so I guess they try to somehow sleuth out if somebody's a major player in one of these dark markets, they try to sleuth out somehow who that person is or who that organization is. 

Philip Gradwell: 00:29:59

So what they do is they try to understand the businesses that, that person interacts with because that gives them sufficient evidence to then go to that business and say, "Look, we have sufficient evidence under the standards of law that a crime is being committed and therefore we'd like to know what information you hold on that account." And that's how they then can make that link between the blockchain world and the real world. And it's not just darknet markets, you might have heard of ransomware. So where people go and lock up a company's computers with a virus and they say, "Unless you send us some Bitcoin, we're just going to wipe it all," and people will pay that. But of course there's an address that they've sent that Bitcoin to, and then you need to know where does that Bitcoin go to? Where do people cash out? So we help investigators then follow that. So it's all about following the money to a point where you can get extra information. 

Jon Krohn: 00:31:01

Historically, creating a studio quality podcast remotely was a hassle with many different applications required. Now, thanks to Zencastr, making high-quality podcast episodes like we do at Super Data Science is a piece of cake. Zencastr bundles together all of your podcasting needs. 4K video and lossless audio recording, editing, distribution across all the major platforms and even monetization, all-in-one intuitive web application. To have recordings as high-quality as Super Data Science yourself, go to zencaster.com/pricing and use the code SDS to get 30% off your first three months of Zencastr professional. Okay, cool. So criminal investigators can make use of Chainalysis software and analytics. Is there also a use case for commercial clients? 

Philip Gradwell: 00:31:53

Yeah. So if you're a cryptocurrency business, you sort of have the same almost responsibilities that those investigators have that you have to check. 

Jon Krohn: 00:32:00

Right. You don't want to be unwittingly, because I guess if a whole bunch of incoming, let's say, you didn't know that all of the revenue you're getting was from a shady source, that is a huge risk because then someone could knock on your door and be like, "Hey, all of the money you've got, that's actually been stolen from a ransomware attack, so we need that back." So you're like, "Oh, no, I've already given away all these widgets." 

Philip Gradwell: 00:32:29

Exactly. But I mean actually worse than that, you would be, in the US you'd be in violation of the Bank Secrecy Act, which is the legislation that all financial institutions have to comply with. So you can't just say, I don't know, you have a duty to know. 

Jon Krohn: 00:32:45

Oh, okay. Super interesting. Okay, so it sounds like there's a fair bit of trustworthiness that is baked into the blockchain by combination of the immutable ledger that is more or less guaranteed by proof of work, proof of stake. And then in addition, there are tools like those that Chainalysis offers that allows us to build trust to have some sense of security as to who the counterparties we're dealing with are. 

Philip Gradwell: 00:33:15

Exactly. 

Jon Krohn: 00:33:17

So are there any other innovations coming out that will help make the blockchain more trustworthy in the future, or are we already in a pretty good spot? 

Philip Gradwell: 00:33:27

So there's a lot of transparency about what happens on the blockchain, and I think the industry is in a constant debate about how much of that transparency is good. So what do I mean by that? I describe to you that you can see everything that's happening on a blockchain, so you can see that this address transferred some assets to this address. You don't always know who controls that address, but maybe it would be good if we could make transactions without revealing that they were made. Some people want to be able to do that. Of course, if you do do that and you send a transaction for nefarious means, then that's not a good thing.

00:34:07

But in our traditional world, we don't have that full transparency. We have a degree of privacy. And so people are trying to work out how to bring more of that privacy onto the blockchain. That's kind of one area of active research that's out there. But honestly, a lot of the technology development at the moment is about getting more transactions on the blockchain. People want to be able to do more and more things. So we're kind of in this stage at the moment of increasing the bandwidth, and that's actually where most of the effort's going. 

Jon Krohn: 00:34:41

Cool. All right. So yeah, let's actually talk about that. Let's talk about kind of volume. So there's a huge amount of volume that happens in the financial system period. So including cryptocurrencies and fiat currencies, there's an enormous amount of transactions that happen every day. Foreign exchange, commodities being purchased, stocks and bonds being purchased, commercial products being purchased at a convenience store by a regular consumer, there are just a bewildering amount of transactions that happen every day. So how is crypto asset analysis different from analyzing other kinds of financial data? So as an example, in terms of the volume that we were just speaking about, my understanding is that there are only about 19 million Bitcoins in circulation and somewhere between 200,000 and 400,000 Bitcoin transactions per day. And Bitcoin is the most popular cryptocurrency as far as I'm aware. So yeah, that kind of scale, it seems to be dwarfed significantly by the amount of outstanding shares on public markets, for example, or the amount of transactions that happen of shares or currencies on public markets. So yeah, I've given you a huge topic here now to explore. Go ahead, Phil. 

Philip Gradwell: 00:36:12

And maybe I'll answer that by almost talking about the Chainalysis data pipeline. What does it take to do the type of work that we do? 

Jon Krohn: 00:36:20

Great. 

Philip Gradwell: 00:36:21

So first of all, you've got to go and get the data from the blockchain and there, which means you have to go run a Node and plug-in to this peer-to-peer network and you can then download this history of transactions. For some of the more recent blockchains, you have to run what's called a full Node, where you are basically capturing all the information rather than just some of the summary information. You have to keep the complete history of all the state changes. And for some blockchains that can be starts to get quite large. You're talking about gigabytes and gigabytes of data, even getting into multiple terabytes now for some of the largest blockchains, at least when you uncompress it, because you don't just want to look at it in its raw blockchain form, you want to map that into some kind of more common concepts.

00:37:19

So you talked about a transfer, but a transfer is not always a transfer. It can be different on a different blockchain. So you've got all of these blockchains, you want to map them into a common schema, you then want to pull them into an environment where you can analyze them. And then what Chainalysis really specializes in is adding these maps. These addresses are controlled by these entities, so we've got to constantly work out what that map is and then apply it and then great, we've got a data schema, which is like this entity made these transfers to this other entity. But you can start adding more metadata on top of that. So we can start saying, well, what if actually there was a business that made a transfer to a entity we didn't know, who made a transfer to an entity we didn't know. It's kind of important to know that, that second-hop entity was connected to that business.

00:38:14

You can start adding more information in like you might have had a decentralized finance, there what people are doing is they're not just transferring or holding an asset, they're swapping an asset or they're minting a non-fungible token. So now your world of transfers starts to get more verbose. There are more verbs. And so your data types are now starting to really explode. And then you've got to find out a way to efficiently analyze this all and serve up answers in a product where people expect a real time response. And so you're right that the scale of the challenge is very different from the entire economy, but you're also places you're getting your data sources, the different, the places you've got to, you've then got to apply that map and then you often have to run algorithms over the top of it to describe the data in a richer and richer way and- 

Jon Krohn: 00:39:13

That sounds fascinating. 

Philip Gradwell: 00:39:14

Yeah. And for also an industry that's not been around that long and there was limits to funding and limits to no one else had ever done this, so you got to solve some problems for the first time as well. 

Jon Krohn: 00:39:25

Yeah, that sounds like a particularly interesting part of what you've built up as the Chief Economist at Chainalysis. So the amount of data that needs to be analyzed, you described this process of capturing data from the public blockchain to start before any of those other pipeline steps that you mentioned like decompressing the data or enriching the data, so just capturing those data initially from a blockchain source, those chains in recent years have become, I don't think it's unreasonable to use the adjective, exponentially larger. And so the amount of storage capacity that you need on the machines to analyze those data needs to grow exponentially, and the amount of compute might need to as well, unless you can come up with tricks. 

Philip Gradwell: 00:40:20

Yeah. 

Jon Krohn: 00:40:22

Right. So that must have been a really interesting problem to grow with, especially as Chainalysis has enjoyed growth around the same time. So when you would've started at Chainalysis five years ago, Chainalysis would've had much less funding, but also the data sets that you needed to be processing were relatively smaller. And then so as the popularity of cryptocurrencies exploded over the last five years, you too enjoyed series A funding, series B funding, series C funding, and those later funding rounds enabled you to presumably hire more data scientists, hire more data engineers, have more cloud compute resources. Tell us a bit about that whole journey. 

Philip Gradwell: 00:41:13

So I start five years ago and we're like, "Great, let's go look at all this data," and then start talking to the data engineers and they're like, "Yeah, so we represent it all in a custom built in-memory graph database that's written in Java." And we're like, "Oh, how do I get the data out of that?" And so go towards my new data scientist, I'm like, "Can you learn some Java and work out how to get that?" And yeah, it was kind of insane, but that's what the technology was at that time. It was actually a really fast way, as in a very responsive way, of querying individual bits of data on the blockchain but it was very bad at querying lots of data on the blockchain. 

 

00:42:01

So I described earlier, Chainalysis is all about following the money from one wallet to another. That's very different from the type of data science workflow that me and my team have been trying to do, which is saying, let's look at this as a whole economy. And so what's really benefited my team is as Chainalysis has grown, thinking about, okay, let's get everything out of custom built in-memory graph databases. Let's get everything out of bare-metal servers that we used to run and get it all in AWS, so you can use some modern tooling. And then you have questions of, okay, it made sense to put it all in the big SQL database, but actually now the blockchain doesn't fit in that, so what do you replace it with? 

00:42:49

And those have been the kind of evolutions that basically every time I think blockchain technology in the amount of data there has kind of got bigger, we've had to raise some more money to try and solve those problems, because it really started in this kind of crazy small way. And now, I think you can do, people look at the blockchain and they're like, "There isn't that much data here. Google, Facebook have solved it." But it's like, "Yeah, but they have Google and Facebook size engineering teams and budgets," and you've just got to grow the [inaudible 00:43:21]. 

Jon Krohn: 00:43:21

Oh, and you're not saying that Google and Facebook have solved the blockchain analysis problems that you're attacking, you just the scale of the data, you're saying Google and Facebook are dealing with orders of magnitude more data, so therefore, come on Philip, this isn't a big deal. 

Philip Gradwell: 00:43:40

Exactly. 

Jon Krohn: 00:43:41

But then your counter is, "Well, we are of a small fraction, we are orders of magnitudes smaller than Meta or Google." So it still is big. 

Philip Gradwell: 00:43:50

One day. 

Jon Krohn: 00:43:52

Yeah. 

Philip Gradwell: 00:43:52

Yeah. And I think a really interesting part of that, I talked earlier about graph theory and thinking about the blockchain as a big network. Actually running queries over very large graphs is something that's still hard for Google and Meta. Google Maps is one of the most mind-blowing pieces of engineering in the world, and it took them decades, ultimately, and billions of dollars to be able to solve that graph problem of how you get someone from A to B. And actually on the blockchain, you can do even more complicated things because we have, we call it temporal graphs because the blockchain changes over time or the set of transactions do. And those are problems that are at the very frontier of applied math. And so... 

Jon Krohn: 00:44:42

Wow. 

Philip Gradwell: 00:44:43

There's even some problems that if you could solve them, you would really do great work on the blockchain that even the big tech companies aren't thinking about. And that's because we have this richer set of data that no one else actually has. 

Jon Krohn: 00:44:58

Looking to take your career to the next level, but not quite sure how? Well, check out Bunch, the AI leadership coach. Bunch is the easiest way to learn critical career skills like giving feedback, resolving conflicts, and communicating with confidence, and you can do it in just two minutes a day. Bunch is not a one-size-fits-all course, but a fully personalized learning journey. You learn daily from a global community of coaches, managers, and executives from companies like Calm, HubSpot, and Twitter. Download Bunch for free in the Apple App Store, search Bunch leadership, that's B-U-N-C-H leadership, or simply check out the link in the show notes.

00:45:37

Cool. All right. So you've put the work into this data analysis pipeline that you described where you're capturing data from these public blockchains, you decompress the data, you enrich it with other related data like NFTs that are being created, I'm probably not using the right verbs, but are associated with some crypto transactions, and then your team applies algorithms on top of all of these data in order to be able to summarize it in some way that's useful and actionable for users. So a long time ago in this episode, we described how one of the hats that you wear as chief economist is leading the development of data products, so it sounds like we're now beginning to touch on that. Could you fill us in on one or two concrete case studies related to data products that you built? 

Philip Gradwell: 00:46:38

Yeah. So I lead the, what we call, the investments set of solutions. So Chainalysis, we talked about doing investigations and compliance, and honestly, they're the core of our business. So following the money when there's been a ransomware attack or doing those anti-money laundering checks on the source of funds when it comes into crypto businesses, that's the main area for Chainalysis. But, of course, crypto is an economy, it's a financial asset. And so surely all this on-chain data can help people understand that world.

00:47:11

And the way that I look at the data is there's a lot of people who, say, might trade in oil futures, like an ETF, right? On energy prices. And that's going to be informed by the actual physical world of energy infrastructure. So how much is coming out of this well? How much is in this tanker? How much is in this storage tank? Now the on-chain data is that physical commodity world of how much is in the tanker, how much is in the tank? And on top of that sits all these people paper trading because they just think the price is going to go this way or that way. And so we have this kind of fundamental data set which should have some relationship to that world of prices and trades in an order book.

00:48:02

And so one of the big things we did was try to understand how is all of that crypto kind of flowing across the map? So how much is actually sitting in this exchange versus this exchange, and how much is moving on a particular day? And a lot of crypto also moves through, what we call, private wallets, so that might be like you might have a crypto app on your phone where you hold your crypto. Can we describe the behavior of those private wallets? Because we actually don't have any other information about them, we don't know the name of the holder. And so literally just trying to build that map of this is the total flow of Bitcoin or Eth or any other asset on a particular day and which type of business or private wallet it's moving through, I guess, was two years of my life. 

Jon Krohn: 00:48:51

Cool. I mean, yeah, it sounds like a lot. So two years is probably a short period of time with those accomplishments. Cool. So then how do these analytics get surfaced to an end user? I guess you have like a click-and-point user interface where somebody can select Bitcoin and decide to break down the data into the amount of outflows from major exchanges or something like that? 

Philip Gradwell: 00:49:27

Yeah, absolutely. And I think this is actually one of the most interesting things about my journey, at least for me, is I was an economist, I wasn't a product manager, and the product manager mantra is go and talk to your customers and listen to what they want and then go build that. And I'll be honest that I was like, "But this is this brand new data set, how do customers know what they want? Maybe let's just kind of go work that out." And what I've realized, after some hard lessons of not quite getting that user interface right, is customers maybe don't know the solution, they don't know the data that'll solve what they need, but they do know what they do every day and they do know what problems they face. And so you've really got to think about what does your customer do every day and what frustrations do they have, and then you can go and work out the data that solves that problem and how to deliver it to them. 

00:50:32

So if they're in a trader or someone in a big financial institution, maybe they don't need that point-and-click UI, you can just send them data. But you need to think what decisions are you driving, and actually most decisions are driven by people, and it hurts as a data scientist, but most of those decisions are driven by people who aren't necessarily in this technical roles and descend all day in the data, and actually they do need a nice dashboard with a really clear workflow that basically says, your problem is A, the solution is B. So for example, now we're building a product that helps people understand their customers on the blockchain and building that for marketing teams or customer success teams and crypto businesses. And there we're taking all of this, honestly, very complex data and boiling it down into these are your customers and this is how they behave. 

Jon Krohn: 00:51:32

Super cool. That's fascinating. All right, so I now understand cryptocurrencies, blockchain, I can see what you do as the chief economist and this data product development leader at Chainalysis. Let's talk a bit more about data science applications that are related to the blockchain. So I've come across things like smart contracts, data providence architectures, and privacy preserving machine learning. So these all seem to be use cases that blend data science approaches, not just data analytics approaches where we're summarizing data, but data science approaches where we're building data models like machine learning models and relating those to the blockchain. So what do you think about those kinds of developments that I've listed? Smart contracts, data providence architectures, and privacy preserving machine learning. And then yeah, maybe you have other kinds of use cases that blend data science in the blockchain. 

Philip Gradwell: 00:52:41

Yeah. So smart contracts, there's a lot to say on them. The others, there's sort of, I don't know, less to say. I think... 

Jon Krohn: 00:52:52

Right. 

Philip Gradwell: 00:52:52

Blockchains have a lot of promise, but they're at this really early stage that we haven't really worked out exactly what is possible. And so people will go, "Yeah, let's build in some machine learning program that runs on the blockchain." And they might have three examples of that, but there's basically nothing else beyond that. Smart contracts, I think, are worth talking about a moment more. When we were talking about the data model of a blockchain, I mentioned that people can transfer assets and they can hold assets. That, for example, is all that you can do on Bitcoin because it's a very simple language. It's not a turing-complete language, the sort of scripting language that underlies the protocol. While something like Ethereum, you have a turing-complete language, which means you can write any software program and have that run on the blockchain. 

 

00:53:46

And so when people say a smart contract, that is just a computer program, and what that has started to do is it's added more verbs into the vocabulary of crypto, so you can swop, you can mint an NFT. Technically, you can do anything, you can make something has arbitrary complexity. And that's fascinating, and that's why people are like, "Hey, we could run a machine learning model on the blockchain." It might be very expensive and inefficient to do it on a decentralized network compared to a centralized one, but you could do it. But people are starting at that simpler level and maybe we'll get more advanced from there. But today it's a bit like a kid on their first hundred words, they'll grow to be an adult who has 20,000 words in a vocabulary, but currently there's like 10. 

Jon Krohn: 00:54:40

Right. Great analogy. Okay, cool. So basically what you're saying is that other than smart contracts, the things that I listed are more hype than practical at this time? 

Philip Gradwell: 00:54:52

I mean, hey, maybe listeners are out there in the ecosystem and they found their application and they should for sure send it in the comments and be like, no, this is real because maybe it has turned up in a corner of crypto. That's the other thing about crypto is the innovation is crazy fast. 

Jon Krohn: 00:55:04

It's changing quickly. 

Philip Gradwell: 00:55:05

Yeah. 

Jon Krohn: 00:55:06

Yeah, you don't know absolutely everything that's happening on blockchain all the time, Philip. Come on. 

Philip Gradwell: 00:55:12

Ah, I try. 

Jon Krohn: 00:55:15

Come on. You're reading Techmeme every day. Okay, cool. So let's dig in a little bit. So we were just talking about the relationship between the blockchain and data science, and so it sounds like that's very nascent. It has an infantile vocabulary, if you will. And so potentially huge amounts of opportunity there for our listeners to be identifying new opportunities to be blending data science and blockchain. But so on a related topic, in your product role, how does that relate to data science? So what's your relationship like as this data product leader? How do you work with the data scientists at Chainalysis? 

Philip Gradwell: 00:56:06

Yeah, so I work very closely with the data science team. I'm much less hands on keyboard. In fact, I think they try to stop me being hands on keyboard, but my job is really to help set the direction and the questions that need to be answered. And the other key thing I actually do is, it's very hard to know if you're right or wrong when you're doing analysis on the blockchain because no one's ever done it before, and so I also help bring in some of that QA experience because I've just spent a lot of time working with the data and understanding it. So there's really two key roles, what are the questions that we should try and answer, and how do we know we got the right answer? 

Jon Krohn: 00:56:49

Very cool. And so you just mentioned that you're not hands-on very much, but do you happen to have a tool that you like that maybe our listeners should be checking out that they haven't before? 

Philip Gradwell: 00:57:01

I mean, one of the reasons why I said my data scientists don't like me being hands on keyboard is I love Mathematica. Which if you've never played around with it, it's got a beautiful interface, it has great graphics, and for people who think very mathematically, you can express yourself literally in an equation, but its performance doesn't really work and no one else can use it. So yeah, I am now banned. 

Jon Krohn: 00:57:26

Mathematica is the language that underlies WolframAlpha, right? 

Philip Gradwell: 00:57:30

Exactly. It is. Yeah. 

Jon Krohn: 00:57:32

Right. So Stephen Wolfram developed, I guess, he developed this mathematical language as a way of doing exactly what you just described, being able to express yourself very richly with a mathematical vocabulary in a programming language. But, I guess, you're saying that the reason why it's not particularly widespread is it's not very computationally efficient? 

Philip Gradwell: 00:57:57

I mean, it has got better over recent years, but it's not as integrated into the modern tech stack as Python, basically, or Spark. 

Jon Krohn: 00:58:08

Got it. Well, so that could be something if listeners haven't come across Mathematica yet and you've been looking for a programming language where you can express yourself more mathematically, it sounds like Mathematica could be something worth checking out. But even if you're just curious about an easy-to-use UI that's built on top of Mathematica, you can check out WolframAlpha, which is freely available online and allows you to use natural language to make all kinds of interesting data-related queries. So the head of product of my company was actually just showing me one today. It was something like, tell me the length of this movie, whatever, tell me the length of Gone with the Wind expressed in dog years. So you can use this WolframAlpha search engine to ask these kinds of semantic questions and blend together different data sets and it figures out how to map them together. 

Philip Gradwell: 00:59:10

And there's a lot of beauty in it. It also has some of the best technical documentation, I think, you can come across. 

Jon Krohn: 00:59:16

All right, Phil, you've led this outstanding exhibition of blockchain, cryptocurrency, data analytics, and data science related to the blockchain. If a listener is out there looking for more to dig into, what resources do you recommend for listeners to become expert at crypto or blockchain analysis themselves? 

Philip Gradwell: 00:59:39

So they can actually follow Chainalysis, and Chainalysis actually has a public academy where you can go and learn the core concepts around crypto and blockchain. So that's definitely one thing to check out. If you are getting to that more advanced stage and you want to go hands-on with the data, actually really recommend June Analytics, which gives a sequel-like interface to go and query on-chain data and there are a big set of public dashboards that you can see how other people have done it. There's a big community on Twitter around this as well. So that's kind of the core places I would stop. 

Jon Krohn: 01:00:16

Nice. And then so I guess that question that I just asked was kind of thinking about somebody who already is a data scientist and is looking to shore up their crypto or blockchain capabilities. But then, let's say, hypothetically you were getting started in your career right now, so instead of already being an established chief economist at a tech startup that's doing crypto analysis, if you knew that you wanted to do that, if you knew that you wanted to be analyzing crypto data for a living, what kinds of subjects or courses would you study to be best prepared for that? 

Philip Gradwell: 01:00:55

Yeah, so I would definitely learn some math concepts and some stats. It's going to be essential. I would also think about taking some finance courses, although I would try not to get too suck down that rabbit hole. I actually think the future of crypto and blockchain is, potentially, less in the financial applications and more in things like social media content moving to the blockchain. So just a lot of people can get pulled into the finance side of crypto. I actually think the future might be a bit different. So I do the math, the stats, some finance, and then I would try and understand the psychology of why people might want to try and get into these new systems, these new ways of doing things. So psychology might be good. I'd always give a plus one to some philosophy, just to put that out there. But anything that helps you think about the world as a complex system and find the signal in that. 

Jon Krohn: 01:02:05

Super interesting. All right, math, stats, a bit of finance, some psychology and some philosophy. Brilliant. And I realized I just kind of said something that sparked a question for me, which is, so I mentioned how you're now established as this chief economist at a tech startup. What's that like? Are there many chief economists or economists at all at small startups? 

Philip Gradwell: 01:02:39

Yeah, I think I was particularly lucky. So Jonathan, who's one of the co-founders, was also an economist, and I think he was just like, this feels like an economy and a problem we should solve. And that probably is a little bit unusual, to see economists at tech startups, but I think it's more and more common and, honestly, it's a lot of fun. So if you're not necessarily a data scientist listening to this, but you're an economist, then my advice is to actually go and think about how to go and apply that in a tech company. I mean, Google very famously has Hal Varian as their Chief Economist, and he wrote all the textbooks and they did amazing economics on the AdWords auction. So the fact that all the auctions to display an ad to people are settled in milliseconds and Google makes a ton of money out of it, largely comes down to them.

01:03:30

But there are avenues in computer game companies like games are economies. Uber certainly had a chief economist who invented surge pricing, so whether that's a good thing or a bad thing, but that's the type of thinking they did there. In real estate tech, there's a lot of economists there as well, so obviously you got to try and think about how those prices are going to move. But yeah, it's a fun world and a lot of economists otherwise end up in consulting like I did at the start. And there's something really fun about going and building products that has, honestly, it's just the thing you should do, I think, if you're an economist. 

Jon Krohn: 01:04:12

Super cool. Yeah, I've read recently that the big tech companies at least have been hiring way more economists than they used to. So it used to be the case that the distinguished faculty that tech companies were poaching until recent years from universities, from top universities, were computer scientists or statisticians, machine learning experts, but that in the last couple of years there's been an explosion in them also poaching the top economists from top universities. So yeah, you do seem to be, you are leading a trend, Phillip. 

Philip Gradwell: 01:04:54

I try. 

Jon Krohn: 01:04:57

Awesome. All right. So if somebody wants to work with you, if they think that you sound brilliant, which you are, and they want to be doing data science alongside you at Chainalysis, do you have any roles open right now and what do you look for in people that you hire? 

Philip Gradwell: 01:05:11

Yeah, we do have roles open. 

Jon Krohn: 01:05:14

Oh, awesome. 

Philip Gradwell: 01:05:14

Even though the crypto markets are down, Chainalysis are still hiring really strongly, which is awesome, and there's a huge amount of data science to be done. And really what we look for is we do need people who will understand a bit of the data engineering as well. Our data science teams run pretty closely with our data engineering teams. You do need a good appreciation for the sort of statistical methods that underlie a lot of data science because blockchain and crypto analytics is so new, you can't just put it all through a machine learning model and expect to get good results. So you're going to need that deeper level of understanding.

01:06:02

And something that I've often found is people who have studied two slightly different topics have often done quite well. So whether that is network theory and operations research, or it's a bit of software engineering and statistics, or I guess in my case, some very mathematical economics and some philosophy. That often stretches people, I think, to kind of be willing to go, "Okay, this is a new area. I'm less scared of just the blank sheet of paper that's in front of me because I've had to do that a couple of times." 

Jon Krohn: 01:06:42

Right. It probably also allows you to identify interaction terms between two different disciplines that very few people in the world would be able to think of or appreciate. 

Philip Gradwell: 01:06:59

Yeah. But I shouldn't say, that may sound like a high bar, that's not on everyone. You can totally have just studied one thing and come in, but you've just got to come with that, looking for those interaction points, thinking deeply about something. 

Jon Krohn: 01:07:11

Cool. All right, Philip, this has been an amazing episode. I always like to wrap it up by asking you if you have a book recommendation? 

Philip Gradwell: 01:07:19

Yeah, so one of the good books I've read recently is called Lying for Money by Dan Davies, and it's a book about financial fraud, which is fascinating because it goes from the small frauds that people might do every day and the mechanisms and psychology of them to the kind of big frauds that just change an economy. And it's just a great way if you in that world of thinking about crypto, thinking about fear, thinking about a complex system that might be unstable, it's kind of just a great insight to that. It's not a rule book or a guidebook on how to do fraud, but I think if you can look at the world through that type of lens, you might start to go, "Ah, maybe the world could be different in that way or different in that way." 

Jon Krohn: 01:08:07

Awesome. 

Philip Gradwell: 01:08:07

That's why I liked it. 

Jon Krohn: 01:08:09

Cool recommendation, Philip. All right. And then you've had amazing information to share with us, succinctly and eloquently, all episode long. How can listeners follow your work after this episode? 

Philip Gradwell: 01:08:22

So I'm on Twitter at @philip_gradwell. That's, honestly, the main place I'm at. I'm also on LinkedIn, but feel free to reach out. Always happy to chat about crypto and data. 

Jon Krohn: 01:08:34

Nice. Yeah, we'll be sure to include those links in the show notes. All right, Philip, thank you so much. This has been an illuminating episode for me, hopefully for many listeners as well. Thank you so much for taking the time with us. 

Philip Gradwell: 01:08:46

Yeah, thanks so much for having me, Jon. It's been great. 

Jon Krohn: 01:08:54

It was delightful for me to catch up with Philip after all these years, and I'm so happy to now have a much better understanding of crypto and the blockchain. I hope you learned a ton from Philip too. In this episode, he filled us in on how chief economists help companies navigate through the winds of global macroeconomic change, how the blockchain is a database anyone can edit with distributed consensus mechanisms to verify the data's accuracy, how crypto is provided as a reward for doing that verification work, whether it be done through proof of work, as with Bitcoin, or through the less energy-intensive proof of stake approach that was recently adopted by Ethereum. He also talked about how he understands user's problems and the decisions of theirs he's driving in order to devise useful data products, and how Ethereum is turing-complete, facilitating any computational task on the blockchain such as smart contracts and even machine learning models. 

01:09:47

As always, you can get all the show notes including the transcript for this episode, the video recording, any materials mentioned on the show, the URLs for Phil's social media profiles, as well as my own social media profiles at superdatascience.com/621. That's superdatascience.com/621. If you enjoyed this episode, I'd greatly appreciate it if you left a review on your favorite podcasting app or on the Super Data Science YouTube channel, and of course subscribe if you haven't already. I also encourage you to let me know your thoughts on this episode directly by following me on LinkedIn or Twitter and then tagging me in a post about it. Your feedback is invaluable for helping us shape future episodes of the show. If you'd like to engage with me in person, as opposed to just through social media, I'd love to meet you at the Open Data Science Conference, that's ODSC West, so that will be held in San Francisco from November 1st through third.

01:10:37

I'll be doing an official book signing for my book at Deep Learning Illustrated, and we'll be filming a Super Data Science episode live on the big stage with the world-leading Deep Learning and Cryptography Researcher Professor Dawn Song as my guest. In addition to the formal events, I'll also just be hanging around, grabbing beers and chatting with folks. It'd be so fun to see you there. All right. Thanks to my colleagues at Nebula for supporting me while I create content like this Super Data Science episode for you. And thanks, of course, to Ivana, Mario, Natalie, Serg, Sylvia, Zara and Kirill on the Super Data Science team for producing another sensational episode for us today. 

01:11:11

For enabling this super team to create this free podcast for you, we are deeply grateful to our sponsors. Please consider supporting the show by checking out our sponsors' links, which you can find in the show notes. And if you, yourself, are interested in sponsoring an episode, you can find our contact details in the show notes as well, or you can make your way to jonkrohn.com/podcast. Last but not least, thanks to you for listening all the way to the end of the show. Until next time, my friend, keep on rocking it out there and I'm looking forward to enjoying another round of the Super Data Science Podcast with you very soon.