Ten

Not the Thing

The proudest fact in the history of Curio Cards is true. Curio Cards is named in ERC-721, the standard that defines what a non-fungible token is. It is named there twice.

What it says is that Curio Cards are not one.

What Hunt thought it said

Start with the belief, because it was held in good faith for years and it is the reason this chapter exists.

On air, 24 July 2018:

“It’s super funny because in the ERC 31 spec, it actually mentions CurioCards. There’s some certain flags or something that you can do because they saw the way that Rhett wrote our Curio cards and they liked what he did and they made it easier for you to run collectible type applications on the ERC 31 tokens and they mentioned our little project Curio cards.”

The claim has four parts: that the standard mentions Curio Cards, that its authors looked at Creighton’s code, that they liked what they saw, and that they added provisions to accommodate projects like his.

The first part is true. The other three are not supported by anything, and the document says something close to the opposite.

Three years later, describing the same discovery:

“And one of the things they said in their little paragraphs, they said, this is a collectible card project similar to Rare Pepe’s or Curio cards.”

“And if they say Curio cards, they must have looked at us. And, yeah, and for science, that’s all we want.”

“And that right there was enough for me, like being in the spec. I was like, we did it. You know, we’re in the spec.”

They must have looked at us. That inference is sound — somebody did look. And that right there was enough for me is the sentence that makes this chapter worth writing rather than merely correcting, because the thing that was enough for him was being seen at all. He did not ask to be credited with inventing the NFT. He asked whether anyone had noticed, found evidence that someone had, and stopped there.

What it actually says

ERC-721’s front matter records status: Final and created: 2018-01-24, with four authors: William Entriken, Dieter Shirley, Jacob Evans and Nastassia Sachs. In the Backwards Compatibility section sits this, verbatim:

Note: “Limited edition, collectible tokens” like Curio Cards and Rare Pepe are not distinguishable assets. They’re actually a collection of individual fungible tokens, each of which is tracked by its own smart contract with its own total supply (which may be 1 in extreme cases).

Read the second sentence, because it is a precise technical description of the design in chapter four. One contract per card. Each card an ordinary fungible ERC-20 token. Supply tracked per contract. That is exactly what Rhett Creighton built, and the standard’s authors understood it perfectly — they are not describing it inaccurately or dismissively. They are describing it correctly and then drawing a line with Curio Cards on the far side of it.

And it sits in a specific place. Immediately above it is a three-item list headed “Example NFT implementations as of February 2018”: CryptoKitties, CryptoPunks, and the Auctionhouse Asset Interface. The note is the paragraph that follows — the “and here is what does not count” paragraph.

Two sections earlier, under Rationale, the standard gives its examples of the real thing:

“Examples of existing or planned NFTs are LAND in Decentraland, the eponymous punks in CryptoPunks, and in-game items using systems like DMarket or EnjinCoin.”

So, in one document: CryptoPunks appears in the Rationale as an example of what an NFT is. Curio Cards appears in Backwards Compatibility as an example of what an NFT is not. Two sections apart, written by people who had looked at both. That is the competitive history of the two projects, settled in February 2018, by third parties, in the founding document of the entire category — four years before either project’s market reckoning.

The second mention of Curio Cards is in the References, under “NFT Implementations and Other Projects”:

Curio Cards. https://mycuriocards.com

Which produces the small absurdity that the canonical NFT standard links to mycuriocards.com in its bibliography while its body text says Curio Cards do not qualify.

Who wrote the sentence, and when

The sentence is not Entriken’s and it does not originate in the EIP file. It is Dieter Shirley’s, and it predates the standard.

Shirley — then and now of the team behind CryptoKitties — posted what he called his first complete draft as a comment on the original GitHub issue. The comment was created 9 November 2017 at 18:28:25 UTC, and GitHub records its updated_at as identical to its created_at, meaning it was never edited. Verbatim, in that draft:

(It should be noted that “limited edition, collectable tokens” like Curio Cards and Rare Pepe are not non-fungible tokens. They’re actually a collection of individual fungible tokens, each of which is tracked by its own smart contract with its own total supply (which may be 1 in extreme cases).)

Note the differences from the final text. In November 2017 it is in parentheses, spelled “collectable”, and says not non-fungible tokens. By the time it reaches the EIP it has been promoted out of its brackets, respelled, and softened to not distinguishable assets. The substance is unchanged.

William Entriken carried it into the file. The EIP’s first commit — SHA 31ed49ea42bca57407395dd6ddaadbaaf38ac461, dated 24 January 2018, commit message EIP-721 Deed Standard — already contained it. The sentence has been unchanged since 6 March 2018, and ERC-721 went Final on 22 June 2018.

One honest gap, flagged because it is the kind of thing that gets guessed at. Shirley opened the original issue on 22 September 2017, and that post was subsequently edited. GitHub does not expose body-edit history, and the earliest archived capture of the page is 7 December 2017. So it cannot be established whether the Curio Cards sentence was in the September issue body. The firm floor is the 9 November 2017 draft comment, and that is the date this book uses.

Nobody argued about it

The last piece is the one that recasts the whole thing, and it required counting.

All 398 comments on the ERC-721 issue were retrieved and searched. Nine match “curio” case-insensitively. Eight of those nine are the words “curious” and “Curiously”. The single genuine mention of Curio Cards in the entire standards process is Dieter Shirley’s November 2017 draft.

Nobody argued about it. Nobody from Curio Cards participated. No one asked for the project to be included, and no one asked for it to be removed. There was no debate about whether one-contract-per-card should qualify. A sentence was written once, by a man drafting a specification, as an aside in parentheses, and it stayed.

Which means Hunt’s inference — they must have looked at us — is right, and the scale of the looking was one person, once, for long enough to write a parenthetical. Not a committee weighing Curio Cards’ claim. Not a tribute. One line of housekeeping in a draft, by the author of a competing project, distinguishing his design from an earlier one.

One correction the project needs to make

While the standards record is open, there is a claim on the project’s own documentation that does not survive contact with it. docs.curio.cards states:

“The two most popular NFT standards later developed (ERC721 and ERC1155) reference and share many design elements with Curio Cards.”

ERC-721 does reference Curio Cards, in the two places above. ERC-1155 does not mention Curio Cards at all — a search of the final text returns zero occurrences. The claim is half right, and the half that is right is the exclusionary footnote.

This matters more than a pedantic correction, because ERC-1155 is the standard the 2021 Curio wrapper uses. The project’s own current infrastructure runs on a standard it incorrectly believes cites it.

Why this is the better story

It would be easy to write this chapter as a debunking, and that would be a misreading of what the document does.

The exclusionary footnote is not an insult. It is a description of something genuinely unusual, written by someone competent, at the moment the category was being defined. Curio Cards did not fit the definition of a non-fungible token because Curio Cards was built before that definition existed — thirty-one separate fungible contracts with zero decimals, which is what you do when there is no standard for unique things and you need unique things by Tuesday. The standard’s authors looked at that solution, understood it, concluded it was not the architecture they were specifying, and said so in a footnote.

And then the footnote preserved the thing it excluded. ERC-721 is one of the most-read technical documents in the history of the field. Because of a parenthetical in a November 2017 draft, the name “Curio Cards” and a link to mycuriocards.com sit permanently inside it — which is how a project that had already failed, whose founders had dispersed, whose website had its shop links deliberately removed, remained findable at all. The sentence that said not the thing is a substantial part of why anyone in 2021 knew there was a thing to look for.

Hunt was right about the shape of it if not the content. He said, in July 2018, to a caller who told him it did not matter:

“So it’s pretty cool where we’re a little footnote in the code of Ethereum forever now. Pretty cool for a little startup type project.”

A little footnote in the code of Ethereum forever. That is exactly what it is. It just does not say what he thought it said, and what it actually says is more interesting: that in February 2018, when somebody sat down to write the definition of the thing Curio Cards is now famous for being first at, they named Curio Cards as the counter-example.