NFTs have been polarizing from the start.
When these pricey, crypto-backed JPEG files wormed their way into the mainstream early last year, there were far more questions than answers about what these things could actually do, and what, if anything, they meant for the future of digital art.
The biggest and most pressing of those questions had to do with non-fungible tokens’ relative energy consumption: Bitcoin (BTC) and other cryptocurrencies have long been known to require huge amounts of energy, much of which comes from cheap fossil fuels. How exactly do NFTs fit into that existing framework, and what’s the responsibility of artists looking to put their work on the blockchain?
This article is part of CoinDesk’s Mining Week series.
Kyle McDonald is an independent researcher and artist looking to answer some of these questions. Inspired by an online challenge from cantankerous bitcoin advocate [and CoinDesk columnist] Nic Carter, McDonald created a dashboard for determining the percentage of Ethereum transactions attributable to NFTs. Late last year, he shared another dashboard, this time looking at the emissions of the entire Ethereum network.
His work is informed by lessons learned from the early NFT backlash. Even if we understand in some vague way that crypto is bad for the environment, it’s important to have hard, current data to back it up.
Last spring, Memo Akten, an artist and assistant professor at the University of California at San Diego, devised his own methodology for calculating the carbon cost of a single NFT transaction on the Ethereum blockchain, and publicized it in a series of viral posts on social media; he took them down soon afterward, citing their role in directing “abuse and harassment” toward artists engaging with NFTs. (McDonald points to several errors in his work, which may have over-indexed those emissions numbers.) Another viral anti-crypto post from last spring called the ecological cost of NFTs a “crime against humanity,” but mostly cited studies that were woefully out of date.
Blockchains are essentially giant public lists of transactions; in order to add a new transaction to the ledger, it’s verified trustlessly by a network of computers running special software. That verification process is at the heart of what makes blockchain networks “decentralized.” There’s no one central actor (say, a bank) signing off on every transaction and making sure it’s good to go.
It can also happen in one of a few different ways. These different processes are known as “consensus mechanisms” – the means by which this network of computers reaches agreement as to which transactions are legit.
The oldest consensus mechanism, known as “proof-of-work,” is structured as a kind of race between all the computers in the network: The first to crunch enough numbers and solve a complicated math problem is rewarded with a certain amount of the blockchain’s native token. It uses by far the most energy of any mainstream consensus mechanism, and is also the most popular. Bitcoin, Ethereum, Dogecoin, Monero,and Zcash all operate this way.
Proof-of-work’s big rival is proof-of-stake – a consensus mechanism that selects computers to verify transactions according to the amount of cryptocurrency held, as opposed to the amount of computing power expended on one of those math problems. It uses significantly less electricity than proof-of-work.
Scientists and data researchers tend to agree there’s already one sterling resource for quantifying the energy consumption of the bitcoin network, from Cambridge University’s Centre for Alternative Finance. Even Cambridge’s numbers are essentially guesses – its low estimates for energy consumption are 1/10th of the high estimates – but the analysis is more rigorous than it is for any comparable metrics for Ethereum. (The most commonly cited dashboard for Ethereum comes from Alex de Vries, the blogger and researcher behind the website Digiconomist.)
Looking past the hyperbole, McDonald’s work targets the numbers themselves, incorporating far-reaching data about chip performance and energy mixes for international power grids.
In a recent conversation with CoinDesk, McDonald explained how he put his Ethereum energy dashboards together, and what he thinks about the responsibility of artists making their way into the fraught crypto space.
The following Q&A has been edited and condensed.
How did your dashboards end up growing out of the debate around NFTs and carbon emissions?
I made the NFT activity one first, directly in response to Nic [Carter’s] bounty. Nic’s bounty came about because there was a lot of discussion around that time about the emissions of art NFTs. This all started with Memo Akten releasing his crypto.wtf NFT emissions calculator, and then people started using it not just to check on their own emissions but to shame other artists for their emissions. And there are some problems with Memo’s calculator. He had a bug in there for a little bit, where I think it pooled the entire emissions for the entire Foundation smart contract, and if you put one artwork in, it would end up pulling the entire platform. [Foundation is a popular NFT marketplace – it launched on Ethereum, switched to a low-emissions layer 2, or companion, system called xDai in late 2020 and then eventually switched back to Ethereum.] It just made it look like anybody who put in from that platform had a ton of emissions that was super-disproportionate to anybody else. And eventually he took it down, because he felt like his main point had been made.
But around that time, and I think even before he took it down, Nic posted this bounty saying, “I don’t think that NFTs are the majority of the activity on the Ethereum – maybe it’s misplaced to be blaming artists for the emissions of Ethereum when it’s not even the main thing that Ethereum’s doing.” And this was true. It’s always been less than 20% of Ethereum activity, with a few brief spikes. I think, though, that he was also engaging with his other question, which was how do we assign responsibility for emissions within this network where emissions do not go up or down depending on how many people are using the network? And then there was this third thing, which was like, “Are these emissions numbers that we’re working from in the first place even right?”
There’s this paradigm in carbon accounting where if you can’t really find a direct correlation – like a marginal change in emissions – based on your use of a service, then one way that you can compute emissions is by looking at how much you pay for that service versus how much that service is raking in from everybody. This is an idea called value-based carbon accounting, and that’s sort of the approach I’ve been taking when thinking about NFTs.
It comes down to transaction fees. What people are paying to use the system should be a kind of proportional marker of what their emissions responsibility is. So instead of looking at the number of transactions, I look at the fees to make an estimate of what percentage of the Ethereum network’s emissions belong to NFTs. And that’s been hovering somewhere between 15% and 20% for the last few months. And it was also around 15% to 20%, towards the end of last year, with a little lull around the New Year.
What about the overall emissions dashboard?
I started working on that because when the first NFT debate was happening, with Memo’s stuff, there were people like Sterling Crispin and other artists who were pushing back and saying, “All this stuff is based on the work from Alex de Vries – Digiconomist – and who even trusts his stuff? Is this really good science?” I think Alex actually has done a lot of great work on this topic, but he’s also very loud and opinionated. And he has, I think, created some misdirection. For example, he talks about, “per-transaction emissions.” And I understand why he does that, it gives you a useful understanding of scale. But it also misleads people by making them think that every transaction has a marginal [cost in] energy or emissions. And so he’s gotten a lot of blowback from the crypto community for those kinds of things.
I saw that and I thought, to have this discussion we need to have some kind of trustworthy source. Where are we going to find that?
I need to know, what is the data center overhead, the PUE [power usage effectiveness]? What is the grid loss for the different places where Ethereum is running? What is the hashing efficiency of all the different GPUs [graphics processing units], or the PSU [power supply unit] efficiencies? I tried to do my best to get real answers to each of these questions, and then piece them together like a big puzzle. I looked at a ton of GPU benchmarks and really tried to understand how hashing efficiency has changed over time with the release of different GPUs.
How do you even begin figuring that out? How many of these data farms and mining centers actually report what equipment they’re using, and what sources their power grids are drawing energy from?
There’s a few good tricks here. Once you know that 30% or 40% of the mining is happening in China around 2020, then you can look at China and say, “Well, what do we know about China?” We know that miners are moving from South China, like, Sichuan, to Xinjiang in the Northwest, and they’re going back and forth with the wet season. So they can take advantage of the hydropower. And then you can find basically government documents from China, as well as external estimates of the energy mix from companies that are interested in getting good prices for their electricity, and use that as a proxy for understanding what the energy mix of the miners is. There’s always the possibility that, you know, a bunch of the mining in a region is actually off-grid, and it’s using electricity sources that are not connected to the grid at all, but I haven’t found an indication for Ethereum that that’s a major thing. I think for Bitcoin it’s much more true that there’s a lot of off-grid mining. Like, there’s something like 1% to 2% of bitcoin which is mined using flared gas, which is not connected to any grid at all, in Texas, mainly.
The question about the GPU mix is actually a little more direct, in some ways. There’s two services I looked at for that. When you register your workers with Nanopool, they actually publish your “worker” IDs publicly. So, if you go into your Ethereum farm and you name your workers “3070-row one-column three” or something, then I can go in and see they’re using 3070 GPUs to do this mining. And I can also see what your hashrate is on that collection of GPUs and do that for a ton of GPUs and then use that to extrapolate to the ones that I don’t know about.
And then I can also kind of cross-check it against one other resource, which is [HiveOS]. It’s a tool that people use on their big farms, and they use it to keep track of how the GPUs are running.
There have been a lot of arguments, mostly from companies in the industry, that bitcoin mined with flared gas is actually good for the environment because it mops up extra gas that would otherwise be burned on-site at refineries. On the other hand, it’s a process that piggybacks on existing gas businesses. What’s your take on that argument?
There’s two main things that come up, which are the kind of pro-energy argument for proof-of-work, which basically is [that it uses up] flared gas and stranded energy sources, and proof-of-work as a controllable load resource – something that is constantly providing this base load of demand that can be turned off to provide some additional capacity.
I think it’s complicated because bitcoin is genuinely reducing emissions through flared gas mining. However, the portion of emissions that it’s reducing is so minuscule compared to the portion of emissions that it’s incentivizing. I’m not sure that’s a great argument. The other thing with flared gas mining is a lot of the places where that’s happening right now, [local governments are] considering just telling oil well operators to just stop flaring it – just bottle it up and sell it.
It turns out the reason they don’t do that is because it’s expensive. But if you get the government to make people do things that are slightly more expensive, it turns out it doesn’t necessarily end their whole operation. It can be a net benefit for everybody.
I also have to admit, like, yes, bitcoin is a controllable load resource [CLR]. Yes, it has actually added capacity to the West Texas grid in the winter, when they turned it off. At the same time, I’m still not convinced that it’s the best way to do things. I think a lot of these pro-bitcoin arguments are right – bitcoin does some good things. I don’t know if it’s the best way to do those things.
With this situation of bitcoin being a CLR – so are car batteries for electric vehicles [EVs], and I think there’s a lot of potential that has yet to be unlocked in terms of recharging EVs at the right time of day in a way that can take advantage of more renewables. I think EVs, even now, have more potential than bitcoin as a resource on the grid to provide a little bit of that backup capacity.
I know bitcoiners would like to see bitcoin as something with everyday utility, but it’s really hard for me to accept that argument when there’s proof-of-stake coins that seem to be operating in a similar way to proof-of-work coins.
Is there a kind of cynicism baked into the idea that governments shouldn’t be playing a role in this stuff?
I think the crypto project in general is about being free from regulation and oversight. And I understand that, I understand that feeling – I don’t want to have a boot on my neck either. But the extreme vision of that gets into this zone where no one is accountable to anybody anymore, and we’re not really listening to each other about what we all want the future to be, or crafting it together. It’s just sort of giving up to market forces. And I’m not resigned to that. I think the only future that is workable is a future that we figure out together.
We’re in the middle of climate change right now, and there’s not space for “trying stuff out” at this point. We know what we need to do, and it’s massive reductions [in emissions]. And proof-of-work is not helping with that right now. There is a way to get rid of proof-of-work, and it’s not by banning mining, because people will just move to places where it’s still allowed. But the way to get rid of proof-of-work is by banning crypto exchanges that trade proof-of-work coins, and to not allow them to exchange between fiat and proof-of-work coins. I think that will severely dent the volume and market cap of proof-of-work coins, if you’re not allowed to hold them and you’re not allowed to trade them at the exchange level.
Would you advocate for that?
I would advocate for government-imposed restrictions on exchanges trading proof-of-work coins. I don’t know yet that I would advocate for restrictions on holding proof-of-work coins because I think the kind of oversight that’s required to make that happen is getting into a very surveillance-y society, beyond what we have now. And I’m not really interested in regulations that can’t be implemented.
I think a lot of bitcoiners, Nic Carter among them, see proof-of-work as the real innovation, and proof-of-stake as essentially traditional centralized computing with a crypto sheen – to them, saying you’d ban it is probably like saying you’d ban what’s worthwhile about the project of crypto. How do you balance that with the question of preserving the environment?
He wouldn’t even see it as a thing to balance because he looks at the idea of proof-of-work as a good, as a CLR. He says this is incentivizing renewables, which in some weird way is a little true. It’s just not the thing that I want to see incentivizing renewables. When you get really down to it, I don’t know that it’s about the environmental question at root level, because you can make some environmental arguments for proof-of-work.
Is it tied up with a broader culture war around crypto?
It’s about this war between whether we should all be able to figure this out independently, through market forces, or whether it’s something that we need to figure out together, interdependent instead of independent. And, yeah, I’ve got a perspective on that, and a lot of bitcoiners feel exactly the opposite.
The battle is currently being lost by my side, and I don’t know yet that it’s going to change around, but we’ll see. Canada has put restrictions on how much proof-of-work mining can happen there, Kazakhstan has restrictions on how much proof-of-work mining can happen there. When we continue to see more restrictions like that, I think we’re going to have to start asking, how much do we really want to be doing this?
As far as I can tell from anything we have up to this point, [crypto] is the acceleration of all of the deregulation. It’s an acceleration of the “capital always wins” situation.
What responsibility do artists have, at this point, to mitigate their own emissions through NFTs?
This idea of personal responsibility in the context of emissions was kind of co-opted in 2005, when [advertising firm] Ogilvy made the carbon footprint campaign for BP. This huge oil company was realizing that they had an impending crisis on their hands, because people were going to start realizing very soon that climate change was due to their company. And what they decided to do was to say, “Okay, well, look away from us for a second and look at yourself. Maybe it’s your problem.” This idea of the personal carbon footprint – that was an ad campaign. That wasn’t something that scientists came up with or economists came up with.
That ad campaign was so successful that we still have this idea of personal responsibility being the most important thing. And we kind of lost, culturally, our ability to hold large polluters accountable and people with power accountable. So what I feel about the personal responsibility of artists is, yes, you need to follow your conscience. And there’s a lot that you can do as an individual to make a difference. But ultimately, if that is distracting you from holding power accountable, it’s a problem.
I don’t want to focus on whether any individual artists should be or should not be working with proof-of-work. I want to focus on the exchanges, the marketplaces, places like the Ethereum Foundation, that still [have] some ability to guide whether proof-of-work is used or not, and make plans around it and allocate resources to it.
Every dollar that’s taken away from supporting proof-of-work is well spent.
The other thing with “decentralized” systems is that it can be hard to know who to hold accountable.
Theoretically, yes. But practically, there’s still Infura [the Ethereum APIs developed by the software giant ConsenSys]. There’s still the Ethereum Foundation. There’s still all of these centralized points where there are individual people that are keeping things running, and individual organizations keeping things running. Web 3 people haven’t solved this goal of having a truly decentralized web yet. You still have servers, you still have organizations that are providing API endpoints.
In a way there’s also an opportunity for accountability, in crypto, because so much of the data is transparent.
One of the reasons this has been such a big topic is because it is legible. When you look at Facebook or Google, all you’ve got is their ESG [environment, social, governance] reports or sustainability reports. And then you go through there and you’re like, “Okay, theoretically this is from a third party, but also Google hired them to make this, they probably want to make it look good somehow.” And then you’ve got things that are totally impenetrable, like the U.S. military. It is incredibly intangible, there is no blockchain for the military that we can analyze to figure out how many emissions there are.
We need to keep the pressure on here, because this is something new, and it’s probably something like one whole percent of all global emissions. But we shouldn’t let it distract us from some of the bigger issues too, like transportation and electricity more broadly.
Further Reading from CoinDesk’s Mining Week
Our reporters visited crypto mining farms around the world, interviewed key players and crunched network data to shed light on a little-understood industry.
Even with the surge in popularity, home bitcoin mining only accounts for a small slice of the industry’s overall pie.
Despite favorable business conditions, a country’s political environment can deter international capital.