We’ve been hearing about the energy cost of mining cryptocurrencies such as Bitcoin for years now. With the recent surge in the value of such coins (the USD value of one BTC is hovering around $59,000 at the time of writing but was at around $6,000 a year ago), the concerns of how much energy is consumed by the distributed computation network used to power such blockchains have once again entered public discourse.
So, what exactly is happening? Well, according to my search engine results, Bitcoin mining alone (so, disregarding all other cryptocurrencies for now) has an annual carbon footprint similar to that of New Zealand and an annual electricity consumption that is greater than that of Argentina. In an era in which we are already feeling the dramatic effects of climate change, there are few arguments for adding this level of energy consumption, especially when the majority of that energy powers machines that ultimately waste that energy.
Proof of Work
To explain what I mean by waste, we first need to examine how Bitcoin mining works. Essentially what is happening is that there is a distributed network of computers (i.e. miners) that are racing to solve what many simply describe as a “very hard math problem”. We won’t go into the details as to what that means here, but the key attributes of this math problem are that 1) it is very very difficult and computationally intensive to solve but 2) is very very easy to verify that it is correct. The “problem” here also intrinsically houses information for determining what the next batch of transactions, or block in the blockchain, should be. That is to say, when a miner solves this problem, the output is a block of transactions that can be mathematically proven to be valid.
Because these miners are putting in a lot of time and energy into solving these problems, this form of block addition is called Proof of Work. That means that, because a valid block could only have been produced by putting in a lot of work, the fact that a miner came up with a valid block (which is, again, easily verifiable) is proof enough that the miner put “work” into solving this problem. As a reward for doing this work, that computer (or I suppose, the owner of that computer and its associated Bitcoin address) is allotted some amount of newly minted Bitcoin that has appeared out of thin air, as though someone just dug it up in a cave (hence the term mining). If you’re interested, you should definitely read up more about Bitcoin awards and halving, but for now just know that the award is 6.25 BTC for every block added.
Let’s do some quick maths here.
6.25 BTC valued at $59,000 per coin… yeah $368,750 feels like it would be worth it to become a miner, right? Especially considering the fact that a new block is added roughly every ten minutes (by design), this feels like a solid economic investment. But, there is an estimated 1 million miners out there and only 52,560 10-minute intervals in a year. So, you have about 52,560 shots at winning a $368,750 lottery every year but you only have a 1 in a million chance of winning each time. And to “win”, your computer has to, again, solve that hard math problem that changes every time, and solve it before any of the other 999,999 miners do.
Okay, so that’s Bitcoin’s flavor of Proof of Work and why people choose to build expensive rigs to solve these problems. Where does that energy waste we mentioned earlier step in? Well, let’s look back. We have essentially 1 million computers solely dedicated to solving an arbitrary math problem that has no real world significance. One of those computers solves it first and gets a huge payout, but the other 999,999 miners have spun their computational wheels in the mud for no reason and now have to start over at square one to solve the next math problem. That means, in terms of productive work, 99.9999% of the energy consumed to power this network of computers was effectively for nothing.
Alternative Hard Problems
Can we harness this computational power for good? Well… theoretically, yes. There has been some research about different methods for harnessing the computation needed for Proof of Work such as to solve discrete logarithms and it’s possible that there are other problems that could incorporated into a Proof of Work context such that the output of the problem is useful to the real world as well.
However, coming up with problems such as these that are useful and reworked to be included in a blockchain’s protocol is somewhat difficult and doesn’t quite change the fact that only the fastest problem solver wins and we have a lot of redundant computation that gets thrown out with the bath water.
It’s additionally difficult for miners of cryptocurrencies such as Bitcoin to pivot their hardware to be able to solve different problems. The “problem” that Bitcoin miners encounter can be more quickly solved using specialized hardware called ASICs that can pretty much only run computations relevant to that problem, but they do them really fast. So, it’s probable that that pool of wasted computation can’t be used for anything else.
Proof of Stake
Okay so using Proof of Work might not have an obvious way to reduce the energy cost of adding blocks to a cryptocurrency blockchain. Luckily, there are alternatives to that family of protocols, the most well-known of which is probably Proof of Stake. For the sake of brevity we won’t go into how the others work, but there are other proposed alternatives such as Proof of Space and Proof of Time.
So what is Proof of Stake? There are many implementations of this protocol, but essentially how it works is that instead of the network relying on the miner that solves a hard problem the fastest to determine what transactions make it into the next block, there is a cryptographically hard process that chooses a miner or set of miners that are allowed to add and validate the next block and receive a block award. The process that chooses a miner can work in many ways, but in the most basic case, the network can assign everyone a probability of being chosen that scales linearly with how much of the related cryptocurrency they own. This method basically assumes that if one owns a lot of the underlying cryptocurrency, one is disincentivized to be a bad actor and add an invalid block because this would cause others to lose trust in the system and devalue the currency, which directly affects that person. This is their skin in the game, so to speak, or their stake.
That particular implementation, however, easily creates an inequitable system where someone with a lot of currency will probably continue to accumulate wealth because they get a higher chance of mining the next block due to the award they receive from mining the current block. There are other flavors of Proof of Stake that are not based simply on the size of one’s wallet, but they’re out of scope for this article.
Advantages of Proof of Stake
Because only a small subset of all possible miners are chosen for every block addition, the rest of the miners don’t have to use electricity to try to solve an arbitrary, hard problem. They can just sit in a latent state until they are elected to propose a block.
If we were to wave a magic wand and apply this to the Bitcoin blockchain, we could reclaim most of that wasted 99.9999% energy and computation. That means that instead of a whole New Zealand carbon footprint, Bitcoin would only produce an additional 0.0001% of the carbon footprint of New Zealand. However, there have been no announcements regarding Bitcoin moving to Proof of Stake at the time of this article’s writing.
Real World Application
That being said, Ethereum (the blockchain that powers the second most well-known cryptocurrency, Ether) is planning on migrating their protocol from Proof of Work to Proof of Stake. This is a unique and super interesting case study that we can use to examine the implications and potential benefits of Proof of Stake.
What’s the catch? Why hasn’t everyone adopted Proof of Stake? Well, a fair, secure, and equitable Proof of Stake algorithm has apparently been under research for over five years, so there’s that. But, the global implications and tradeoffs, I think, are well worth it, especially for Ethereum. Not only does the system not have to consume wasted energy, it also opens a world of possibility because Ethereum’s current Proof of Work “hard problem” is one that can’t be solved with ASICs. All miners of Ether use graphical processing units, or GPUs, to run their computations. This kind of hardware can be used for any generic, computationally intensive problem such as training machine learning models.
This means that, once Ethereum completes its move to Proof of Stake, that there will be a huge pool of GPU-based computational power that is suddenly freed up for use.
This pool of computation is still only tied to the Ethereum blockchain, but one can easily imagine an incentive system that allows owners of these GPUs to opt-in to allow, for example, academic institutions and researchers to run computationally intensive tasks in exchange for small fees. It feels counterintuitive to utilize the machines that we just allowed to power off, but if you think about it, those computations would have been run anyway but maybe with an establish cloud provider (such as AWS or Google Cloud), so this gives people an option that isn’t tied to a large corporation while also effectively turning machines that were previously simply wasting most of their energy into a distributed computation network.
Now is as good a time as any to mention that I’m not a cryptographer or an environmental engineer or anything like that. I’m just a guy who works with computers who tries to bring both bins to the curb on trash day when I’m not jaded about the fact that recycling plastic is, for the most part, a myth. But I do believe that cryptocurrency is having a profound impact on society and the earth, and it’s up to us to understand and help shape how it does so.
Thanks to Gabe for helping me sort out my thoughts and understand the underlying mechanisms of cryptocurrency.
And thanks to you for reading!