On The Brink with Castle Island - Michel Rauchs (Cambridge Center for Alternative Finance) on quantifying Bitcoin's energy use (EP.232)

Episode Date: July 26, 2021

We welcome Michel Rauchs, research affiliate and longtime contributor to the Cambridge Center for Alternative Finance which just released an update its Bitcoin Electricity Consumption Index featuring new mining location data. Covered in this episode:  Why Michel returned to the Cambridge Center for Alternative Finance The evolution of the CCAF's miner benchmarking efforts The original motivation for the Bitcoin Energy Index The relationship between the CCAF and Cambridge University How to estimate energy consumption from hashrate figures The meaning of the confidence interval in the electricity consumption figures How CCAF amassed the CBECI mining map Is the pool sample for the mining map representative? Methodological drawbacks with miner location assessment Why the CCAF is still reluctant to determine a carbon emissions figure for Bitcoin New evidence regarding the seasonal hashrate migration Is the Chinese crackdown going to lower the carbon intensity of the Bitcoin network? Is country-level granularity sufficient to determine the energy mix of mining? How the CCAF devised comparisons between the Bitcoin network and other consumers of energy Does Bitcoin get held to a different standard than other industries? Michel's experience talking to the press about Bitcoin energy consumption Has the narrative changed in the press at all? Michel's level of optimism regarding the decarbonization of Bitcoin mining This episode supported by:  Eventus, the leading global provider of multi-asset class trade surveillance, transaction monitoring and market risk solutions. Its award-winning trade surveillance platform is easy to deploy, customize and operate. Eventus is proven in the most complex, high-volume and real-time environments and supports many of the industry's leading crypto exchanges including Coinbase, Gemini, ErisX and OSL. Find them at onthebrink.link/eventus

Hello and welcome back to On the Brink. Before we start a quick word from our sponsors, do you know what Coinbase Gemini and Erasex have in common? They all use the sponsor of this podcast, Eventus, the global leader in trade surveillance, market risk, and transaction monitoring. Many of the largest crypto exchanges, broker dealers and trading firms in the world, use Eventus to improve the efficiencies of their regulatory operations and mitigate the risk of fines and reputational damage. If you're someone that wants to operate an institutional grade exchange or trading platform, you need to be running Aventus. Visit on the brink.link.link slash Aventus today to find out why 80% of the firms who take a custom

demo become clients. So today we're sitting down with Michelle Hawkes. He has been involved in Bitcoin research for really long time at the Cambridge Center for Alternative Finance. We interviewed Apolline Blondin, his former colleague last year when they came out with this Bitcoin mining map, which was this unbelievable project where they took data from mining pools and tried to geolocate Bitcoin mining activity became one of the most cited resources. It provided support for this idea that Bitcoin mining was overwhelmingly China-based. The Cambridge Center also issues this Bitcoin mining index, which is tremendously important in terms of being an unbiased and neutral third-party view of Bitcoin's energy consumption. A lot of other sources are

deeply inferior. And so Michelle's return to the center and they updated their data for the mining map, giving us a dataset stretching from late 2019 through to April 2021. And so there's some fascinating insights in there. So we talk about the methodology because it has been a little bit controversial. I really wanted to dig in to the pluses and the minuses of this data collection method. Now Michelle also wanted me to emphasize that the center, is independent, so they're neither a spokesperson nor an opponent of the Bitcoin industry. They report what the data suggests and not what they want it to be. And that gives them the credibility to actually have real impact in this sort of energy

discussion. And having known Michelle for a long time, I can totally attest to this. They genuinely are very fair operators, and they're an academic institution as opposed to an industry group or a critical outfit. And so I really love the work they do. Let's dive into it. It's a fascinating conversation. Brought down by bad mortgage investments, Lehman, which has 25,000 employees, will be liquidated.

The federal government loans American International Group, AIG, $85 billion. This is a different kind of market, and the Fed is asleep. The federal government is stepping it to stabilize Fannie Mae and Freddie Mac, the two mortgage giants that have been threatened by the housing crisis. The Bank of England has pumped 75 billion pounds more to Britain's ailing economy, with a new round of Concentive Easy. You print a couple trillion dollars and all of a sudden

people start to worry. So out of this worry, we have something called a Bitcoin. Bitcoin. Hello, I'm sitting here. Well, I suppose not here, strictly speaking,

but I'm communicating with Michelle Hawks, who is a research affiliate at the Cambridge Center for Alternative Finance. Michelle is the Titanic figure in the world of the Bitcoin energy consumption debate,

having previously spearheaded the center's efforts. Of course, they published some fantastic benchmarking reports, I think three to date. And then he left for a time and then came back. And we're all very grateful because Michelle has pushed out this new update to the Cambridge Bitcoin electricity consumption index, which now has coverage of Bitcoin from September 2019 to 8,000,000.000. April 2021. Anyway, we're all very excited about it. Michelle, thanks for joining us. Well, thanks for having me, Nick. Pleasure to be on. So tell us about your history with the center. You left, you came back. What motivated the return?

Yeah, so essentially just after graduating, we're worried the thesis on the Bitcoin ecosystem. I essentially joined the center quite coincidentally as a research assistant. And that was just about the time where they launched a new, back then it was called cryptocurrency and blockchain program. And so after a year or so, the professor who was leading this left. And so I essentially took over the lead of that program and then significantly expanded it for over two and a half years, roughly. And then about two years ago, I just decided for myself that it would be interesting to venture a bit more into the private sector. And also I want to relocate back to my home country of Luxembourg. And so essentially I started a small consulting firm specialized in, well, you know,

the domains of digital assets, distributed ledger technology and all the other fancy names that the technology is currently being referred to. But yeah, I've always remained involved, I would say, to some extent, with the center's work as research affiliates. So I was aware of what was going on. I occasionally provided, you know, directional input and guidance and so on. But really recently, particularly with the, you know, surging or actually very heated debate about Bitcoin's energy use and the sudden inches by the media and policymakers as well on our tool, the CBCI, we essentially just got so many inbound requests. The team just couldn't handle it anymore. And so they needed to have someone that essentially could talk about all of this and actually also started really improving the tool. So, yeah, a few months ago, essentially I just got back and helped them also develop a broader digital assets research strategy that goes over multiple years.

So CBCL is one aspect of that. But we do actually have a lot more exciting research streams and projects planned for that. So, yeah. Wow. So were you at the center for the publication of the first benchmarking report? I think was that back in 2018, I want to say, possibly 2017? 2016, even. 2016?

Well, no, no, actually. So we started in 2016. Sorry, it was a long process, actually. So, yeah, it was essentially just the two of us, so Gary Coleman and myself. So essentially, I literally wrote it. And you polled minors for that one, if I'm not mistaken. Yeah, so it was the first time that we actually did an empirical benchmarking study on the industry and I think it was really the first that you know the first benchmarking study ever

really and so initially we started with four what we call market segments then which is exchanges payment service providers wallets and also miners and so we they have a special survey for each of those and so as a result we first got into contact with with some of the large mining operations and firms which interestingly many of them don't exist anymore today you know it's amazing I think about the role of the center in this energy debate, which I've been consumed by for the last year or two. And it's instrumental. I mean, you, the center provided some, you know, very profound. I'm not going to say anecdotal, like survey-based data through those studies. Then the CBCI, the index, came along and provided sort of some minor location data, your estimates of the renewable share of the network

in that third benchmarking study gets cited all the time, that 39% estimate. So, you know, so much of what the center has done has informed the debate. Was that something you expected to happen? Well, let's say when we do designers research projects, we of course hope that, you know, somebody will read them and it actually will have some impact right but um i think particularly in the case of cbc i were quite surprised um i think in in terms of just the reception that it has got and particularly more recently now as a debate where he has i would say captured the mainstream yeah the cbc i is probably the most cited resource on four journalists looking to put a number on bitcoin's energy consumption would you say that's the case i think it's fair to say um today

Yeah. So actually the funny thing about CBCI is really that it started as a sort of like, you know, side project by Applin and Tyn and myself, so my two co-workers, although Applin just recently left. And essentially, the only resource back then was Alex DeVries's Dig Economist Index. And while it introduced some interesting, you know, kind of like ideas around the methodology, we thought that that would be a different way. to do this. So rather than doing this economic top-down model, we wanted to do something more from the bottom up. So essentially a techno-economic model, I think you can call it, based on some work that has been done previously by Mark Beavent. And so essentially we decided to, rather than writing a paper that just gathers dust

somewhere on the bookshelf, we wanted to actually have a live tool that people could always get a go back to you, that we would then also. So essentially really contextualize all of this by adding comparisons, by really putting some context behind that. And also very importantly for myself to really put some educational content out there, that at least attempts to provide a fair and balanced view of what I think is really a very complex and multidimensional topic that really can't just be simplified to, you know, just a punchy headline.

Yeah, the index is really, really stellar, I think, because it drives. arrives from an entity that is perceived to be neutral or genuinely is neutral. Well, I do hope we are. As opposed to the dig economists always had a bit of an editorial slant, if we're going to be honest. Regarding actually the center, I mean, so this is something people refer to you sometimes as just Cambridge University. What is the actual relationship between the center, the judge business school, and then Cambridge itself? Oh, that's a really good question, actually, because I had no. no idea before I joined how complicated it actually was.

So, well, Cambridge University, you know, it's existed for centuries. It's a very old institution. But actually Cambridge is, I think, together with Oxford and Durham, and the only college at university in the UK. So essentially it consists of 31 autonomous colleges and the oldest dates back to, I think the 13th century. But now it gets complicated because the university also has over, hundred different departments that are part of six different schools.

And so our center, which was only founded in 2015, is part of Cambridge Judge Business School, which itself is a department of the University School of Technology Administrative Group. And there's a lot more complexity in that. But so I don't think it's fair to say that we speak for Cambridge University with our studies. although of course we do hope that we captured a generous spirit there but I think it's worth separating between those two still although we are a part of Cambridge University yeah that was that was my intuition as well so I'll be careful not to refer to you as Cambridge so just but I thought that was important to note because there is a lot of conflation so referring going back to the

index your methodology so you know of course there's this this um you know potential introduction of error when you move from a mere hash rate estimate to an energy consumption estimate because we don't note the precise composition of machines active on the network. So tell me a little bit about how you go from a hash rate figure to analyzed terawatt hour figure. Yeah. So essentially our bottom model, when you start with the hardware, so the technological equipment, so essentially A6, we do have a list of a thing.

it's over 100 basics now with their power consumption, their efficiency and everything. And essentially, we need the key, the key center variable, well, not variable, but actually, let me rephrase that. So I think really the entire model is based on the profitability thresholds. So everything that lies beyond that thresholds just isn't profitable to mine, just purely in electricity terms. And so that way... Sorry to interrupt. And you're assuming a certain cent per kilowatt hour price. Oh, absolutely. Yeah. So this is the critical parameter, which we've set by default at five US dollar cents. This is based on conversations we have with minors. It's been used in other academic

studies. And of course, you can't really have, you know, a global average electricity price. So So it's at least, I'm going to put this. Yes, it's a big assumption, essentially. But that is confirmed, I would say, by the conversations we have with other miners and other academics. But that being said, we do recognize that the importance of that parameter. And so on our website, we actually let users of play around with the slider. So they can see how the model adjusts if you actually change that assumption.

And the intuition there is as electricity gets cheaper, the implied electricity consumption per unit of hash increases for the Bitcoin network. So the idea is that we, so we do calculate for every single day a set of profitable hardware. So all the machines that given this five US dollar cents parameter, plus then also of course all the network variables, so, total hash rate, Beckcom price, mining revenues, including transaction fees and everything. Yeah. And so in terms of, because you actually give more of a confidence interval, I would say. You've kind of an estimated number and then a hypothetical range. Like what do the min and the max refer to on the track?

Yeah. So I think it's important to highlight again, this is a model, right? So you can't really know Bitcoin's electricity consumption for, you know, a variety of reasons. So what we do with our model is we call it the best guess estimate, but of course, as you know proper academics, we wanted to also provide a broader, at least purely theoretical range just to show people, you know, what, where at least it could fall and what would be completely not even unrealistic, but not even possible with the current equipment that exists. So essentially the minimum and the maximum figures there are really just for.

for the user's reference in a sense of, it couldn't get worse than that, or better than that, in the sense. And it's based on two assumptions that are very simplistic and very unrealistic for that purpose. So the lower bound actually assumes really that all miners would be using the most efficient hardware,

which is just incredibly unrealistic. And the upper bound just does the opposite. It says that miners would just be using the least efficient hardware, but as long as that hardware is still profitable and purely electricity terms. Got it. So on the one hand, it's basically saying the theoretical maximum electricity consumption of the network would be if all miners were using ancient, largely inefficient hardware, which produces relatively few hashes per unit of electricity. But as we know, like this, just a mix of machines active on this. network and miners are always replacing their fleet and things like that.

Exactly. And it's not just, you know, less efficient, but we need the least efficient equipment model, which is even more unrealistic way. So in, I guess it was 2020, um, you, in your absence, the center published this mining map, which to me is one of the most interesting pieces of data I've ever come across, frankly. Um, I think it was Apolline that Speer at that or at least she came on the show to talk about it at the time. I want to say maybe April 2020 and then there was kind of a fallow period when the map wasn't updated and that caused a bit of consternation at least for me because people were citing the data as of those current as of April 2020 in terms of the mix, the geographic mix of miners. And then now to my, to my

you know, to my immense joy and satisfaction, you've, you've updated it with new data from pools, and we now have a data set running from September 2019 to April 2021 in terms of trying to locate miners. So I guess, first of all, thank you for doing that. I know it's a lot of work. And second of all, like, you know, just tell us about how you went about getting this data set and, you know, how we should be interpreting it. Yeah. So it was a very long process that actually dates back, you know, a few years even. So we mentioned at the beginning this first benchmarking study we did on the ecosystem. So that's really when the, you know, relationship building started in the first place.

And so we maintained those relationships with different miners, mining pools and so on, really over the years. And so that really you could establish a certain level of trust, I would say. And so as a result of that, at least it was easier for us to get access to that data, but then still from actually agreeing to participating and actually getting the data and making sure that everyone submits the data at the same time, it isn't the same former, you know, just adds a lot of complexity and further delays, even though we're just dealing with for now four different mining pools.

So it is a quite arduous process, but we're very grateful for the mining pools that participate, that they're contributing really to this research project. Just tell us a little bit about the methodology for this mining map. I mean, how are you obtaining, you know, geographic estimates of where miners are based? How does that work? Yeah, so there's different methodologies that you could use, right? So it's the same as with the electricity consumption. you can't really know for sure weather-based unless you do a purely bottom-up approach where you would actually track down pretty much every single facility,

or at least like the largest, you know, responsible for 90 plus percent. And I'll just say you about the complexities of already coordinating between four different parties when it comes to data sharing. So you can imagine just having hundreds of different facilities all needing to share data or we needing to employ dozens of people to track data. them down. So that was not really an option for us. So we decided to actually explore alternative methodologies and starting with a top-down approach. And so we realized that mining pools would actually send out shares. So work shares to the hashing facilities that are connected,

I would then send them back based on the work that they've done. And so in that process, at least some of the mining pools would collect the IP addresses of those hashers. And so rather than us having really to keep track of individual facilities, which also for frankly, just confidentiality and privacy reasons is not something we would actually want to do and really have that data set. But so through that, we realized that we could actually really directly get that data from the mining pools in an aggregate way. So we would have absolutely no individual hasser data, which we actually not need at all, but actually just the aggregate. individual pool distribution that pools are collecting on our behalf and generating on our behalf before they send it to us. So currently we do have four mining pools that contribute. So it's BTC.com, pool in via BTC and foundry. And essentially what they do is internally they

restructured their technical infrastructure and processes so that they can, so that they collect these are peer addresses and then aggregate them on their end. And then they share their individual pool distribution, which is based on essentially country-level data and also for China, the provincial level data, with the absolute amount of hash rate that they have for each of those regions on average for a given month. And so then we receive those individual pool distributions, we aggregate them on our end after we did some data verification checks. And essentially, then we look at, okay, so how much is this interim? of total hash rate and then from there we extrapolate to the whole network.

And so currently, at least for the periods that we examined, so from September 2019 until April this year, that is pretty much between 32 and 35% of the hash rate. So about roughly one third that is contained in this dataset. So, you know, I don't know if people really appreciate how Garvey Gantuan a task this was. I mean, we're talking about mining, which is one of the most secretive industries on the planet and trying to coax these pools into sharing reasonably sensitive data. I mean, it's fully anonymized, of course, but still, like sharing the geographic breakout of their clients. Of course, it's good from an information perspective. It's good for analysts like me,

but it's also, I can understand why there'd be some trepidation around that. I don't know if people appreciate how difficult this was. But I'm very appreciative to the pools to do it. And special thanks to Foundry, who were also a new contributor this time. So thank you, Foundry. Yeah, and a big thank you from our end as well. So actually, most pools have been very open. And the problem is just, often it's either problems in terms of technical issues.

so their infrastructure is not ready or they're not collecting that data. And so changing all of this just for this, you know, what was back then, I would say, a minor project as just not a big priority. And then, of course, now more recently, just in terms of the government crackdown in China, and of course, the big hash rate migration debt, well, all the hash rate that's on the move now, of course, they had, you know, different priorities. So that is all to say that in the future, of course, we hope to increase the sample. and to onboard additional pools.

So we've had many discussions with some of them, and I'm quite confident that we will manage to onboard additional ones, so that we can actually increase really the validity of that analysis as well. So in terms of the sample, this is, so there's a number of common critiques, of course, of the mining map methodology. We'll dive into them. So one would just be sample bias.

So, you know, the allegation that, pools, some pools have a particular geographic focus. Maybe the pools that are in your sample are disproportionately catering to miners in a certain region. I mean, for instance, Foundry, you know, tends to do a lot of marketing in terms of being a North American pool. I mean, what do you make of that? Like, how representative would you say the sample is of the Bitcoin network at large? Yeah, that's a good question and something that we're well aware of, which is why we try to onboard more pools, right? But so the initial three pools, they join for all Chinese pools, but already that have a quite international audience. And the interesting thing is if you actually look at the distributions and also the evolution of a time, of course there are differences, right, between the different pools.

but the trend in the overall direction was pretty much consistent across across those three pools as well, which to me at least is an indication that this is to some extent at least indicative of a broader trend in mining as well. But that, of course, being said, there are reasons to believe that perhaps some countries might be understated given that some of the large mining facilities do actually partner with the pools that are not in our samples. So this is, of course, always an issue if you're working with samples. And so we always try to increase, you know, the sample size as a result of that. And so perhaps I should say at this point that we actually set up an API endpoint to make it easier for mining pools to actually share the data and to automate all of this.

So also for new mining pools that might be interested in contributing, it would be an hopefully more streamlined process than, than it used to be. So to any miners, many pool listeners listening, please contribute this data is invaluable in terms of understanding the Bitcoin network. With another critique or comment, I suppose I had, looking at the new release of the data,

has Germany at 2.8% of the network and Ireland at 2.3. I think we can probably both agree from our understanding of mining, there's probably not a lot of mining occurring in these Western European countries with relatively high electricity costs. What do you make of the presence of these two countries in your sample? And what does that tell you about the methodology? Yeah, well, it clearly shows that the methodology has certain limitations.

Actually, I would even add Beijing to that list as well, and there's probably a few others. but the reason for that it really is, well, at least for Ireland's pretty clear, it's the heart of ICT in Europe. So most VPN connections go through there. So it's not very surprising. There's quite a lot of servers in Germany too. Now, I wouldn't say this is about mine is actively trying to conceal the location, but it's just about, you know, standard practices of using VPNs or proxy servers in some companies. And this is really, and I want to make this very clear, just like all,

models, you know, they're always based on assumptions and there's always trade us with that. So the key assumption we did at the heart or at the heart of the mining map is really the fact that IP addresses constitute an approximate indicator or accurate, I should say, relatively accurate reflection of the actual mining location. Now, if that assumption doesn't hold, well then essentially the whole analysis doesn't, you know, doesn't fly anymore. And so we are definitely aware of that. But I think given the few outliers that you can see there, and again, we're looking into different ways of filtering those out in the future.

But if you actually look at the numbers, it's still the broad trends are still there. They have been confirmed by all our discussions that we had about mining trends with different stakeholders. So I'll still think given all the limitations of other methodologies, I'll still think it's a quite acceptable trade up there. But that being said, of course, we're looking into alternative methodologies as well to just complement this. So also reaching out to some of the large mining facility operators to start doing a more bottom-up approach as well. Just on one hand, to essentially validate the data that we get from the mining pools. but also on the other hand, when it comes to future power mixes and the environmental questions. So we have a better understanding about these as well.

Yeah, I don't think it's a deal breaker by any means. And I think people that understand the methodology are prepared to accept that there'll be some mining shares that are routed through VPNs. I guess one question would be, you know, as this data becomes more, well known and as you continue to add pools hopefully you know do you think that there's an incentive for miners to actually conceal their locations in the presence of the of the cbc i data site if they kind of feel surveilled if they're in some location where it's politically sensitive where maybe mining is banned and they don't want it to show up in the data site is that something that you or have sort of thought about?

That's certainly one option. I think the other option would be to actually deliberately trying to influence a given narrative or change the narrative. Right. So we're definitely aware of this. It wasn't too much an issue before since CBCI would say a fair share of attention, but it wasn't worth news. Now with the debate that's been going on for like the last six or seven month,

what has intensified over that period, that has definitely changed. And as a result, I think this might become an issue in the future. And so this is why we're actively looking into alternative methodologies to at least complement our current approach. It's kind of like quantum mechanics where observing the data changes the data. Yeah, it gets very immediate. Classic data problem.

What do you make of, so there's been a big change in the data set. And we haven't even incorporated the post mining migration or, you know, the event in May 2021 when China basically turned off all miners domestically. You know, your data set ends in April. But even so, it's very informative. It shows the U.S. really growing its share. It shows Russia getting pretty significant. Kazakhstan, really significant.

What are, you know, Iran is on this? there with almost 5% of hash rate, pretty interesting stuff. What were your big takeaways from just the country by country developments? Yeah, well, I think the obvious really takeaway, which actually surprised me quite a lot, is the continuous decline within China's relative share that actually studied a lot earlier than most people would have thought, at least mining outsiders, I should say. Now, the fact that the US, Kazakhstan and Russia have gained are not too surprising. just based on essentially anecdotes and personal relationships that we have with some of the miners.

So it was a trend that was already visible, but now actually having empirical data to at least corroborate some of those personal experiences is a good thing. And this is also for me personally, the most fascinating thing was to finally actually document with empirical data, the seasonal migration between Pramik Xinjiang and Sichuan in China, which really, you know, it was not a secret in the industry, like everybody knew about it, but you couldn't actually really prove it to an extent. And so what the data here clearly shows, again, within the limitations of the methodology and the samples, is that those migrations definitely took place, and they're actually quite substantial.

So we really talk about massive amounts of hash rate moving there. So just seeing that with actual data, that was quite fascinating to me. And unfortunately, it's probably, well, it's the first time that it's been documented this way and probably also the last time now. Right. Yeah, the hash rate migration in me is one of the most fascinating things I've ever discovered, really, in terms of Bitcoin. I mean, talking about an emergent phenomenon based on fluctuating costs of electricity,

causes this enormous infrastructure, you know, like migration on a seasonal basis, like birds flying south for the winter, except it's ASICs getting loaded onto trains and going south for the summer, I think that's when the wet season is in China, right? If I'm not mistaken. And so you've like hundreds of thousands of ASICs

going, well, historically, going from Sing Chang and Inner Mongolia down to Sichuan, Yunnan. And so you have this great visualization of this. And the swings are just so big. I mean, like in January 2021, you're saying that this data shows that Sing Chang and Inner Mongolia collectively

at 65% of Chinese hash rate. And then in the peak of the wet season in kind of September 2020, we're looking at 61% of Chinese hash rate in Sichuan alone, which is mostly hydro. So huge, huge swings. And I mean, that paired with the fact that a lot of this hash rate is now migrating out of China, it just drives home the fact that Bitcoin is this pretty unique thing where the industrial base is just highly, highly portable. I can't think of any other industry that's like that.

Yeah, it's really fascinating to see that. But also, I think it illustrates, again, very well the point that it's easy to make, you know, or to make certain, how should that phrase this? Yeah, I can cut this out. But essentially what I want to say is that it illustrates quite well the fact really that there's so much underlying complexity and that, you know, we tend to always think in like clear, simple categories and so on. but it just doesn't apply to Bitcoin at all. Like every time you think you finally nailed something down, and I'm sure you and pretty much all your listeners have that on a continuous basis, I guess still, always when you think you grasp something,

then there's like this other thing that brings you down, not a rabbit hole, and you realize, oh, well, it's actually a lot more complicated and nuanced than I used to think. And this is just one of those examples as well. And there's so many in Bitcoin and crypto more generally. It's, that makes it for me at least, personally such an interesting domain to work in. Yeah, I don't know about you, but I got dragged down this rabbit hole of the Chinese energy grid dynamics,

which I'm sure you're quite familiar with at this point too. And I mean, I found it very intellectually stimulating to learn about the supply demand imbalances on the Chinese grid and how they're trying to refix those. I just couldn't believe that the Chinese grid was so unbalanced. Yeah, well, just energy markets in general, right? So if you go to the US and you have all these different grids, they're managed by different board, they're called again RTOs. And it's just fascinating how different, essentially, policies and economic models lead to completely different outcomes.

It's, yeah, it's just an hugely interesting subject on its own. So here's a question for you. the Chinese grid, you know, it's kind of interesting because my characterization, my initial gut reaction when I heard about the Chinese ban was, well, this is probably going to be a net decarbonization, you know, a net move towards decarbonizing mining because I was obviously aware of the heavily coal-based provinces where mining was pretty big. But of course, you know, there's also the significant hydro element. So as hash rate leaves China or is simply turned off in China and new miners got

delivered elsewhere, whether it's Russia, Canada, USA, et cetera, I mean, what is your interpretation of the change in the carbon intensity of the Bitcoin network? Because it's not, you know, entirely clear what it was in China in the first place. So what's your, what's your interpretation of that transition? Oh, it's just way too early to tell, to be frank. So if you look at China as a whole, I think it's fair to say that it's on the rather higher end in terms of carbon intensity when you compare it between different countries. But as we all know, there's this huge internal disparities.

So just, you know, the episode with Xinjiang and Sichuan, those migrations showed it very clearly. And so at least before that, before the recent crackdown, you knew at the particular, you knew at the particular point in time doing the season, at least the peak of the season, pretty much how, you know, what share was powered by hydro or what share was powered by coal. So now what actually happens, all of that is going away and miners are probably becoming more, or those machines are probably becoming more dispersed and distributed across the globe. And that just makes it a lot more complicated to follow them. So if anything, we actually probably will know even less about

that's Bitcoin's energy footprint than we used before. Because we finally had a handle on the Chinese data on a province by province basis, and then it went away. Exactly. And you don't know whether those similar migrations might actually happen in the future between countries where we're in different regions. So you never know. So I think it's just way too early to tell.

Yeah, it's a funny problem to grapple with. It's constantly changing. And I guess we're just going to have to look forward to the next release of the index and the mining map. Right. But even that doesn't tell you the power mix that those miners are using, right? And it could be more granular in a sense that right now we're just looking at countries and Chinese provinces. So even like within a given kanshi, no matter how small it there is, do you just like those huge discrepancies potentially as well? you still need to have additional data sets and information.

And so we've been looking for quite some time into essentially complementing the index and the mining map with a carbon emissions model that actually brings those two together. But so far, we haven't found a robust or a methodology that we think is sufficiently robust and that we're comfortable with publishing. And so until we don't have found such a robust methodology, we don't publish anything. Yeah, that's very sensible. I mean, you've read the academia. I've read it to, I've been very critical of some of the academic efforts to devise carbon emissions figures for the Bitcoin network. Some of them just use global generic energy mixes, which, you know, didn't really make sense. But in the absence of data, that was the simplifying assumption that was made. Some used country level generic energy mixes, which I also felt, you know, in some cases, is over-simplification because the organizing unit of miners or energy is, you know, the specific grid you're on.

And then in some cases, miners are kind of not even grid-connected. They're just drawing power directly from power plants. So, I mean, it's endlessly, endlessly complicated. But I suppose you could just make an estimate at the country level. and call it a day. Oh, absolutely. It would be quite crude. Yeah.

Right. And we could do that. Well, actually, people can do it right away with our data, right? So that's really something that we think is very important that all the data that we collect, at least in an aggregate form, we also make it publicly available so people can actually play around with the data, use it for whatever purposes they want. And so essentially you can just download that data and then apply average country emission factors to that and do the analysis for yourself.

But I think given now the attention and scrutiny, really, that our tool has attracted, we feel that it is not responsible, I would say, reasonable to actually put out a crude model, which might then be essentially taken as, you know, just a given fact, which it definitely isn't. And so until we have found a better methodology that we feel more comfortable with, we prefer and not publishing anything and just waiting for additional data sets and just doing more research on that. This is kind of like publishing a book which has the recipe for gunpowder or explosives in it or

something. It's like here's how you can actually cook it up, but we don't recommend that you do it. It's like the ingredients are there. You need the energy electricity estimate, the country by country analysis, and then you can just pull carbon intensity factors from the IEA database. And you can derive, I think if you did it today, you'd probably get something in the 40 to 50 megatons of carbon emissions per year for Bitcoin. Right.

And there's actually a website that does there. I forgot the name, but somebody actually built a website just to do that. Really? So perhaps, yeah, if you can find the link later, you can put it in the show notes. So then moving to the more general energy. debate, it's kind of changed a lot, I would say, in the last six months. We've got, you know, these new initiatives, like there's the crypto climate accord, there's the Bitcoin Mining Council. Have you followed any of their work and their influence on the debate?

Yeah, so of course we do follow any sort of initiative that have to do with Bitcoin's energy and electricity consumption and the environmental effects. So I think of all, it's really a positive development. So we do welcome any initiative that creates new insights, or also new data points, because we really see that as complementary. And for us, it is an actually good thing to compare or being able to compare different methodologies and just see the results. And if the discrepancy is actually not that high, that is actually a good confirmation for your methodology too, right? So this is also why Dig Economist Index has been quite useful to us as well because it allowed this to essentially test whether the assumptions

hold or not. And actually for most of the time, the assumptions are pretty closely following each other to some extent, of course. And so overall, I think that's great, though I don't want to comment on specific initiatives, as there are quite significant differences, I would say, between these, though. And they also, to some extent, cover different things. So it's quite hard to put them in one single bucket. But yeah, just to reiterate what I said in a sense like we are not, you know, competing or seeing ourselves is competing against anyone.

So any initiative that helps bringing more clarity and transparency there is a net plus. So from our perspective, we see this very positively. So, I mean, don't feel free to not answer this if you feel that you can't. But the Bitcoin Mining Council, for instance, they published some bottom up data, based on a sample of, I think, 32% of hash rate at the time, which was end of Q2, 2021. And they found 67% what they called sustainability in terms of energy mix, which meant renewables plus nuclear plus a small amount of carbon offsets.

Is that data that you could potentially employ in a model, or is it in a form that's difficult for you to sort of engage with? Um, so I definitely don't want to rule out that we might be using, um, data sets or data points like these in the future, no matter where they come from, but of course we first need to validate the methodology, the way that the data has been collected and making sure that it fits within our own essentially, you know, quality standards and everything. In terms of, um, actually contextualizing the data, you've done some really good work on that. I think you've made a very significant.

effort to both derive comparisons. I mean, the revamped version of the site is really excellent. Drive comparisons. Talk about constraints to the comparisons. Compare Bitcoin to the energy consumption of various countries, to various industries, to sources of energy loss. Tell me about that. that work and what it's been like to try and devise these comparisons and how people have reacted

to it? Yeah. So the reason we actually did this comparison in the first place is just that many people are not, you know, energy experts or electricity experts. And so they do not really have an idea of the scale of those numbers. So like even if you asked me a few years ago, what does a hundred tell what I always mean? I would have absolutely no idea. So it's quite a lot of natural and I would say necessary evil in the sense that you need to put it into context to compare it to something. Now, the problem with Bitcoin really is that there is nothing like it, right? Like even other crypto is in a sense. And I don't want to get into any sort of, you know, tribal issues here.

But like, I think it's fair to say that even Bitcoin among crypto is quite unique. And outside of crypto, it is literally nothing like it that exists. You can, of course, do partial comparisons compared to gold, compared to payment or something. settlement systems to, you know, no trees and things like that, but you only still get a partial view of all of this. Then there's also really the subjectivity. So no matter how well-intentioned you are, you always make a sort of, you know, statement or implicit statement by deciding what you compare it to. So, for example, if you compare Bitcoin to a country, well, then it just looks incredibly large, right? And people just get very emotional about it. Now, if you compare

the same electricity amount to, say, a metropolitan area or a city, then the picture looks completely different, right? And so depending really on what you compare it to, you can really influence the perception of whether this is a good or bad thing in a sense. And whether you like it or not, you just can't escape it. And so I think really the challenge is, How can we provide comparisons that essentially make it look both small and large, depending on what you compare it to? So essentially showing both sides. So sure, it consumes as much as a country of, let's say, 20 million inhabitants.

But hey, in the US, you're just sitting here with like five million inhabitants that consumes the exact same amount. So it puts, you know, things in perspective. Now, the big problem that we faced and where we had will spend endless hours, really, with unfortunately not a lot of success, is to really try to find that comparable data. So, for example, find metropolitan area data with one single data set that is consistent rather than having tons of different sources.

And the thing that we found is, and maybe we were looking in the wrong places, so any direction on that would definitely be appreciated. But what we found at least during our research is that there is just a lack of useful, robust data on the energy footprint of most industries or activities. And we just spent so much time trying to find those relevant comparisons, but the data was just not there or was not, you know,

valid enough, I think, to really be used to make a case here. And that just creates this huge issue that you want to do comparisons that, you know, provide a fair and balanced perspective, but you don't have the underlying data for that. So what do you do? And so the kind of compromise that we did is to at least add some explanatory tax to those comparisons, just to kind of highlight the limitations of these to at least, you know, make users and visitors aware of the fact that, you know, those compares only provide partial views and insights.

So take them with a grain of salt. Yeah, I mean, it is so challenging because you're right, Bitcoin is quite unique. I mean, on the one hand, it's this sort of gold-like commodity that you can store wealth in outside the state, outside the banking system. On the other hand, it's also a payments network, which settles $10 to $20 billion a day. So, you know, what do you compare it to? Do you compare it to the financial system, to Fedwar, Swift, to bank branches? Or do you compare it to literal gold extraction? There's no perfect mapping.

Exactly. But as far as gold is concerned, I'm looking at your site right now. It's about two times as much as Bitcoin in terms of its energy impact. And certainly gold has enormous ecological costs. If you're aware of how gold mining works, it can be very destructive environmentally. But we just don't hear the same debate about gold mining at all. I mean, I don't think I've ever heard anyone really complain, aside from Bitcoiners, about the energy impact of gold mining, which, by the way, is a lot of it is diesel fuels,

so it can never be decarbonized. Yeah, absolutely. I think this is also an observation that is quite interesting in the sense that Bitcoin really seems to be held to a somewhat different standard, at least in the public debate. A weird thing is because Bitcoin is just this really quirky thing that still exists after more than 10 years and all these obituaries. right? And people are just like, okay, how the hell does this thing, you know, still exist and consume so much electricity?

It must be a total waste. And so the debate gets very emotional. Now, if you actually look at the environmental effects and as we say, you can't really directly do that, but again, you can do those theoretical, you know, thought experiments, for example. So even if you assume that all of Bitcoin mining, with assuming currently is 70, 10, what hours, was exclusively powered by the really dirtiest coal you could find and that being processed in the least efficient or most inefficient power plant, which actually used to be in Australia but got decommissioned a few years ago.

So even then, with all the worst case assumptions, they're so far away from reality, even then you get to a total figure of, I think it was about 110 megatons of carbon dioxide emissions, which is still only, and I say only, you know, on the quotation marks here, but essentially only 0.35% of the words totally emissions in 2020. So even like in this absolute worst case scenario, Bitcoin is still just a blip really on the environmental map in a sense,

which is not to say that we should downplay any of this, not at all. But it's just, it's quite interesting that it has occupied so much place in the public debate. given the relative scale, frankly. And yeah, it's just an interesting observation. But I think... No, so please go ahead. Yeah, no, no, I agree. I mean, what is 30 basis points of global emissions, as you say, worst case?

It's kind of an interesting thing that I think the other thing is that it's so transparent. It's so easy to get this data. well, comparatively easy, as compared with getting data on gold extraction, for instance. Because with Bitcoin, you can do this analysis of taking hash rate and deriving the energy impact. And so that transparency is kind of a vice for Bitcoin. Michelle, you've spent more time talking to journalists about this exact question than possibly anyone else on Earth, I'll say. Over the last few years, have you noticed? that for it.

One of the top, I mean, certainly you're one of the top resources for the press. I mean, I don't want to give you undue work, but I should hope that they're reaching out to you in terms of doing their due diligence. Have you noticed their tone changing? Have you noticed any of these talking points or clarifications that tend to emanate from the industry, the Bitcoin industry, have you noticed these percolating out? to the press at all? It really depends.

So also, you know, just like Bitcoiner isn't really a generic thing, where there's no such thing as a typical Bitcoin. There's also no such thing as a typical journalist in a sense. So they all do have different interests. But what I could definitely see is that over the last few months, really there's just sort of like exciting interest in this really, you know, quirky, fascinating thing. they're still around. So it goes back to what I was just saying before that actually most journalists

I speak to are becoming more open really about hearing the other side of the coin to you or the other perspective as well. And I think this is really great development because it also means that at least generally the tone will change a little bit and become more balanced, which I think is just no matter what side of the debate you're on, it's just a very welcome change. So I do hope that over time they just would continue. Yeah, I've been impressed by some of the recent coverage. Not that it's glowing or, you know, endorsing Bitcoin or anything like that in the mainstream press, but I have seen more of a recognition for some of these points of clarification that Bitcoiners have been trying to make for a long time. Like, hey, Bitcoin

can buy energy that's off grid. It can buy. curtailed energy, whether it's hydro or something else. And so I've seen some developments there. We've had you on for about an hour and, you know, be respectful of your time. I think the last question I have for you is really, you know, there's all these potential levers to decarbonize your operations as a miner. So you could move to a grid, which is more sustainable. Or you could you know, pursue this curtailed hydropower directly, or you could engage in this flared gas mining, or you could simply just mine on the grid and buy offsets. Are there any levers that you think are particularly promising in terms of miners decarbonizing their operations?

Is there anything that you're sort of most excited about right now? I personally think there will be a combination of different measures that will yield the desired impact. So that includes more renewables, but also active decarbonization measures within the companies themselves, right, which are right now not counted as part of the Bitcoin network power demand. And also then offsetting VAR certificates and all the other things you mentioned. So it's really that combination. I don't think there's one particular option that that works better than others. I think of all really just being more transparent, I think, and disclosing more or you know more disclosures of power mixes what would I think also already do a

huge favor to pretty much everyone involved really because the fact that that you have this lack of transparency which by the way is changing right but I think in the end it is very important for everyone that includes a general public that includes minors themselves but also very importantly investors and regulators and policymakers that, you know, business and policy decisions may be taken on the basis of incomplete information or sort of inaccurate insights. And so in order to counter that, you know, just being more transparent definitely is a big first step there. And then working collaboratively on decarbonizing this, which, by the way, I think is really

in the very interest of the industry and the wider Bitcoin industry. in general because it risks really becoming an existential threat for them through the very simple relationship between the Bitcoin price and of course the industry itself. So if institutional investors aren't allowed by their strict ESG mandates or because of ESG concerns to invest in Bitcoin, well obviously that will depress the price, which will essentially impact the Bitcoin industry and particularly mining as well. So I think really that this creates a sort of natural incentive over the long term for the industry to find solutions to actually actively

decarbonize. How optimistic are you about the carbon intensity of the Bitcoin mining network decreasing over time and then B getting more disclosure and more transparency from these miners? So for B, I'm very confident because we already start seeing that. So at this point, I like to again thank the four mining pools that already contributing to this project. And as I say, I welcome all these other initiatives as well to bring more transparency in. For the first point, I can't give you an answer to that for all the complexities that we just discussed over the last hour. I don't think you can make any prediction. The reason being that, sure, this talking point of Bitcoin is among others that it will incentivize the creation of new renewable power capacity or generation capacity.

That might be true, but then the same argument also applies in the opposite direction, right? So we've seen all power plants that were previously decommissioned and now suddenly are becoming profitable again because of the extra demand that miners create for stable and low-cost power. Or at the very least, it extends the economic lifetime of some of the very old polluting power plants. So the thing is there's so much nuance on, you know, on all of these things that are, really do not feel comfortable to make any prediction or just any sort of, you know, essentially forward-looking comment there.

Well, an extremely fair and balanced take, I think, from you, Michelle. This has been an absolute wealth of knowledge. I'm so thankful for you coming on. And I look forward to supporting your work. Hopefully, I think it's among the most important projects on the informational side of the industry. So I want to thank you for joining us and for your work with the center. I mean, it really is needle-moving stuff. So thank you again.

Well, thank you very much for the kind of words. And also thank you for inviting me onto the podcast. So I can share our perspective as well. So just for your listeners, we are always, you know, open to collaborating or just, you know, in terms of whether it's insights, whether it's data, sets or whatever. And also, Widi, do encourage constructive criticisms. So if you don't agree with some of the data with the findings, we invite Witty to reach out.

So we can actually use and incorporate it into our processes to improve the methodology. And in the end, we did the validity of those results. So please do reach out. Awesome. Well, you heard the man. Michelle, thanks very much for the time today. Cheers.

Thank you very much.