On NFTs: They're Just a Different Database

Subscribe with RSS


There's a lot of hubbub about NFTs recently, and for some reason every time they're covered it's always with this veneer of vagueness when it's really not that complicated from a "what are they and what do they do" point of view. It always delves straight into the weeds of the philosophy of value, but you should probably know what they do first!

I'll note that this isn't about the value of blockchain NFTs in society; that you can decide for yourself, and hopefully with more of a background.

An Old Hypothetical

Fundamentally, the idea of a unique token tied to digital ownership is not new. I'll illustrate an example where there is no magical blockchain involved.

Say I want to sell a token from of some art I made. I spin up a good ol' SQL database and make a table that associates a User with an Art. Let's say Bob buys a picture of a flower from me; he gives me some money via normal means, I add a row in my table saying that the user Bob owns "The Flower".

Bob can brag about his new "art" piece, and if anyone doubts him, he can point them to me, and I can pull up my database and confirm that he indeed owns "The Flower".

This works so long as everyone trusts me; they need to trust that I will faithfully maintain transaction coherence and will continue to exist (since I disappear, Bob will no longer be able to verify that he owns his art).

There's many examples of this; for example, take the Steam marketplace. What is a CS:GO skin if not, in the end, a primary key in a row in Valve's database? Everyone trusts Valve to keep everything legit. Valve sells people skins, and people can trade skins by asking Valve to exchange ownership in their database.

Or say, the deed to a house in the United States. Here, you're trusting the Government to maintain the database. The deed can be thought of as a token; it's certainly not the land itself, and I hope for the buyers sake it's not fungible (If you're worried, perhaps try title insurance).


So with blockchain NFTs, what you're really doing is replacing the SQL database with a blockchain (usually the Ethereum one for most NFTs). Rather than trusting me as an entity, you're trusting that Ethereum's consensus method is sound and that it is sufficiently expensive to hijack it that it's unlikely your ownership will be challenged.

(By the way, it's "NFT" because other tokens, like Ether or Bitcoin are fungible, i.e there's no difference between my Ether and your Ether, just like dollars, hence the need to append "non-fungible" to it to distinguish them)

When someone buys an NFT, what that gives them is the ability to tell everyone "Hey I own X". Then, everyone who doubts you can go on, say, the Ethereum blockchain and verify that you indeed own X.

Does that inherently mean you own anything? No! Again, it's just the database. Of course, entities can back those records with value. For instance, I could say I will at any time exchange $100 for any of my art pieces. Then I am giving some value to those tokens. But that's something I'm doing.


Take NBA Top Shots. I'd argue that the is little practical benefit here, because there is clearly an authoritative entity. The NBA can easily just run it out of a traditional database and I'd imagine most participants are not that worried that the NBA would scam them, just like how people trust Valve with their digital items. So it's more that without the magic buzzword of "blockchain", there wouldn't be much interest in it.

But hypothetically there are situations where there isn't an entity that is clearly trustworthy and authoritative. There, you are essentially piggybacking on the trust of whatever blockchain it's on. So that is the "value" add - piggybacking on trust.


Some of the energy comparisons are a little unfair - they assume that the entire reason why miners are mining Ethereum or another blockchain is just for NFTs, which doesn't seem to be the case. But it's also undoubtedly a contributer, and when you compare it to the amount of energy it takes for me to insert a row into my SQL database - it's not even close.

There's no doubt that as is, they take up waaaaaay more energy than a traditional database. Now, to be fair, the value add is trust, so you have to compare the energy levels to maintain trust, and that's where people come up bringing up the energy cost of the entire US military and so forth. I won't delve further into this topic.

Supposedly Ethereum is moving to Proof of Stake (so, instead of "it's expensive to have a majority of hashing power", it's "it's expensive to have a majority of Ether"), but who knows. Believe it when you see it.