How Harmful Are NFTs Really? A Meta-ish Survey
I am a nature photographer, and love to share my work online. It’s a good idea in terms of science communication, and I occasionally make some money on the side licensing images to magazines and blogs, which supports my nature, and especially underwater photography. Hearing about non-fungible tokens I recently put an NFT of a fish online:
Following this I had a good discussion with a friend on how energy-wasteful NFTs are; thank you very much to my friend, as usually you get smarter by talking to smart people. So, I looked at the information I could find on the topic.
The Scholarly Studies
Let’s first list all the peer-reviewed, quantitative studies by computer scientists which deal with the topic. The energy use per NFT will of course depend on a number of factors, such as the type of hardware used, the number of sales, ect ect. Here are all the scientific papers on the topic:
Other Information
The fact that there is seemingly nothing in the scientific literature yet (Did I miss it? Please let me know) doesn’t mean that there is no information on the topic available. I found a few interesting articles and blog posts.
The claim about the enormous NFT energy cost seemingly mostly goes back to one blog post by this artist/AI engineer. I believe he acted in good faith when compiling the information, and he put a lot of thought and energy into it. But did he get his numbers right?
This article in Quarz explains the numbers from above blog post with some very nice infographics. Note the several times they stated “carbon emissions: unclear”, and the numbers they have are (I believe) derived from one person’s estimate, the blog post mentioned above. The conclusion is that:
Over its lifecycle, the average NFT will accrue a stunning footprint of 211 kg of CO2, equivalent to driving 513 mi in a typical US gasoline-powered car.
That’s a lot of energy/CO2! But are the numbers correct?
This article in an ARTnews paints a more nuanced picture, and interviews a few blockchain experts. In conclusion, it’s complicated. Are NFTs hitchhikers into an already existing blockchain or do they use energy on their own? Do NFTs drive the demand for cryptocoins (likely not)?
They also state (as of last April, I think it’s still true):
Yet no peer-reviewed research has been done by academics and the methodology being used by different parties varies widely.
And, finally, the libertarian view of things: In recent years, especially as an immigrant or would-be-immigrant in a few different countries, my views have drifted towards the libertarian side of things. Less meddling in people’s lives by the state would be a good thing in most cases. The libertarians like the cryptocoins because they are not controlled by any central bank. Bias is ok if one is open about that bias, and Reason certainly is. They make some good point about the types of energy used (which wouldn’t go to other uses otherwise), and the fact that multi-millionaire US politicians (probably with a carbon footprint 100+ time my own) lecturing everyone else on environmentalism is massive hypocrisy. The article is not specifically about NFTs, but about cryptocoins in general:
Also see this article on bitcoin energy consumption, which makes good points about the expected bitcoin mining growth (or lack thereof) and the types of energy used.
Given that there is so little concrete knowledge about NFT energy use, it’s surprising that so many people on social media are convinced (and outraged about) the high energy use of NFTs. But then, maybe it’s not surprising given the energy people like to spend on outrage.
The No-Free-Lunch Argument
One argument which I have yet to see debunked is that if NFTs are so outlandishly energy-consuming, who pays for this energy? Not me as the creator of the fish NFT above.
According to the US Energy Information Administration, the average US residential customer uses 893 kWh/month, which is about 8 times more than we use in the Philippines, with a household with 3 adults and a child, with a standard of living much above the average local. Our 100+ kWh/month cost us about 25 $. If an NFT, as sometimes claimed, uses the energy equivalent to weeks of US residential family use, the energy bill would be 100s of US$s, wouldn’t it?
The fees at opensea.io, which are 2.5 % of the sales price, and which must cover their profit and operating cost in addition to the energy cost, can’t pay for that, unless the sales are above ~ 10 000 of US$, which presumably few of them are?
So who pays for this “free lunch”? Either there is an element here which I don’t understand yet, and then I’d like to hear about it. Alternatively, the energy use of NFTs isn’t as high as claimed.
Finally, I managed to get some excellent thoughts on the topic from my computer science friend Chris Hiestand. This is what Chris had to say about the questions raised on the topic of NFT energy consumption:
I’m not sure about how to get that granular in determining energy usage of NFTs. But like [in our discussion above], there is incentive for mining to be done in places with very cheap, very dirty, energy. I’m not sure how many kWh 2.5% [of the sales price] buys in China or Kazahkstan, but it’s worth considering.
It also may be the case that miners get almost all of their profit from sales of the coins themselves (you get new coins by adding new blocks) and the “gas tax” you’re paying is just a cherry on top. If this is the case, then miners are going to be putting filler, or no useful data in any case, into the blocks just to get the coins. So it may be the case that the block is going to be added whether or not your NFT exists – so then you might as well piggyback on the new block anyway guilt free, right?
Well that’s still not the whole picture. Novel tools like NFTs and DAOs and DeFi, all built on etherium, are probably creating a bunch of the demand for etherium coins by investors and users in the first place. So it may be the case that NFTs have a terrible environmental impact because of their secondary effects on increasing the demand and therefore price of ETH, therefore incentive to mine. I don’t know how you’d ever figure out how much this is the case. You’d have to guess at how much eth price has gone up because of NFTs and divide your share of that by how many NFTs you have traded.
I may be wrong or missing something, but I’m thinking the answer of how much energy per NFT, with precision, is unknowable.
I Conclude
Yes, obviously blockchain operations are computationally intensive, by design, and that costs energy, via the laws of physics. But how much an NFT costs in energy use, and if a concrete number can even be assigned to an individual NFT or even to a system of NFTs, is far from clear.
The way I understand it, the value of crypto-coins is derived from their sparsity, which is a consequence of the algorithms which create the coins, not of the planetary/geological events which made gold sparse. These coins might be wasteful and a fad, but they are not a Ponzi scheme. NFTs are a part of this system, but it’s quite unclear how much they contribute to the energy use.
So, I think you get the idea that I don’t claim to have all the answers, but I think I have some good questions. And, seriously, if I missed information or arguments, let me know about them.
