Fifteen
What the Chain Remembers
On 26 September 2017 — after the Tuesdays had stopped being reliable, after the last card had gone up, before anyone had used the word archaeology about any of this — Thomas Hunt posted four words and a handle.
“Collecting is intensely human. @MyCurioCards”
Nobody replied in any way the archive records. It is the last substantive thing he says about Curio Cards for four months, and then for most of four years.
It is also, as it turns out, the correct answer. Everything in this book that worked, worked because of that sentence, and everything that failed, failed for reasons that had nothing to do with whether it was true.
The gap was the product
The central strange fact of Curio Cards is that its value was created by neglect, and the people it happened to can barely bring themselves to describe it as a mechanism.
Hunt, trying:
“It’s that four-year gap. It’s that time that’s made them presumably valuable.”
“it’s the strangest thing to be in a project four years ago and to kind of fail at it to not have great success and then four years later everyone’s like we love what you did”
Note presumably. He is hedging about the thing that made him money, which is characteristic, and he is also right to hedge, because the mechanism is not obvious. It is worth laying out, because this book has now assembled all the pieces of it and they only make sense together.
Had Curio Cards succeeded in 2017, it would have been a small trading-card business, and it would have been superseded within a year by CryptoKitties, and the cards would have been early-but-not-first examples of a form that moved on. Chapter nine has Hunt’s own bitterness about CryptoKitties arriving in the same year, and chapter nine also has the Beanie Baby comparison he reached for in January 2018, which is the comparison a man makes about a product category he thinks is a bubble.
Had Travis Uhrig left the vending-machine links on the website, the remaining cards would have trickled out to whoever wandered past over four years, and there would have been no discovery event — just a slow leak. Chapter nine has him taking the links down because selling worthless cards to people felt wrong.
Had the contracts had an owner function or a version number, somebody would have switched them off or superseded them. They did not, and none of the founders can explain why — chapter seven has Hunt shrugging at it.
Had the first person to find the vending machines in March 2021 bought the lot quietly, there would have been no market. Chapter eleven has him publishing instructions instead.
And had the project moved off Ethereum — which it nearly did twice, in chapters three and eight — there would have been nothing at a findable address at all. Chapter three has the reasoning that saved it, and it is the single best decision in the project’s history:
“what we’ve been interesting about that is that the cards would have been orphaned on their own chain and no one would have been mining it and no one would been looking for them.”
He says it twice, four years apart, in near-identical words — once in June 2021 and again on the sixth anniversary. It is the most stable thought he has about his own project.
Five pieces of luck and restraint, none of them coordinated, four of them indistinguishable at the time from failure. That is not a business strategy. It is a sequence of things not happening, and a project that was sealed rather than killed.
Hunt’s own name for it is the right one:
“And again, just looking at all the happy accidents that had to happen”
The thing he was right about
So strip away the priority claim that chapter fourteen declined to defend, and the footnote that chapter ten showed says the opposite of what he thought, and the million dollars that arrived four years too late to be a business. What is left that Curio Cards actually got right?
Two things, and he identified both of them early, in plain language, while the project was failing.
The first, from January 2018, in the middle of the quiet years, explaining the category to a caller who had asked about something else:
“But what’s so great about Curio cards or crypto kitties is that they can prove that they’re cryptographically rare. And they can say we printed one of these cards or we printed a hundred of these cards. And that’s all they’re ever going to be.”
And that’s all they’re ever going to be. No art-historical claim, no first, no standard. A print run you can prove and cannot extend. Everything this book has recorded about supply figures — the burns of chapter four, the 111 of chapter five, the locked cards of chapter twelve, the thirty-three complete sets of 2024 — is downstream of that one property working exactly as designed, for nine years, without anyone maintaining it.
The second is the payoff, and it is the sentence this book has been walking towards:
“People collected sets of the things that we made to be collected into sets. So that’s the whole thing all right there. I don’t need anything else.”
Said in 2021, about strangers. He had made things to be collected into sets; four years later, people he had never met collected them into sets. I don’t need anything else.
And immediately before it, the smaller version, which is funnier and somehow more complete:
“And then they did exactly what I thought they do. They took a picture of it and they posted it on the internet.”
That is the entire thesis of digital collecting, delivered as an anticlimax. He expected people to take pictures of their cards and post them. They did. That was the hypothesis, it was tested, and it came back positive after a four-year delay.
Where it went
The project outlived its founders’ involvement and did something with the proceeds, which deserves recording because it is the one part of the Curio Cards story where the original promise of chapter five got extended rather than abandoned.
The project’s documentation, on the arrangement the community voted for:
“All of the Card vending machines from 2017 are sold out, where one hundred percent of the sales revenue went to the original artists. The community voted for a 1% royalty fee on all OpenSea trades to be equally allocated to the artists. No OpenSea cut is taken by the team or community. Curio Cards has always been about supporting artists and promoting digital artwork as a medium.”
Two things to keep straight, because they get conflated constantly: 1% is the fee rate; 100% is the artists’ share of it. And the wording is “equally allocated to the artists” — an equal split across the seven, not pro-rata by how many cards each made or what they fetch. Phneep made sixteen cards and Thoros of Myr made one, and the royalty does not care.
The DAO that administers it set out its own purpose in January 2022 in terms that are, by the standards of that year, startlingly modest:
“It was envisioned that the purpose of guild presence was to assist community-driven decisions, support the artists, and protect the historic value of the Curio Cards project; in short, to be a coordination tool, not a speculative investment tool.“
And there is one documented downstream use of the royalty money, which is the most satisfying fact in the book:
“Cryptopop has continued to collaborate, building the Cryptopop Art Guild in the Philippines, where some of his Curio Cards royalties are going toward mentorship and community establishment for young underprivileged artists.”
Cryptopop is Luis Buenaventura, who flew from the Philippines to a San Francisco meetup where somebody introduced him to somebody, and who made cards 17, 18 and 19 — including UASF, the card that was accidentally deployed twice and became the most valuable single object in the collection. Money from that accident now funds art mentorship in the country he came from.
One correction while the DAO record is open: Curio+ and cards 31 onward are a separate project. No official Curio source — not the documentation, not the FAQ, not the DAO’s own posts — mentions them. The documentation is explicit about why anyone can appear to extend the collection and cannot actually do so: “Anyone can use the original 2017 token factory to make a new NFT with the (now wildly outdated) Curio NFT standard. However, any tokens made this way are not associated with the original collection.“ The factory is still there. It still works. It confers nothing.
What the chain cannot do
It would be a tidy ending to say the blockchain remembered what people forgot, and chapter fourteen has already shown why that is too tidy.
The chain remembered the deployment order, the supplies, the IPFS hashes, the vending machines, the duplicate at seven in the morning on 20 June 2017, the 97,846 cards burned a week after launch, and the exact second — 17:22:51 UTC — at which a hundred thousand pictures of apples became available for a quarter.
It did not remember who the whale was. Hunt, still:
“we still don’t know this is still a mystery to us a true true mystery”
Every transaction is public. The addresses have been there since July 2017. Nobody has matched them to a person, and the founders spent years suspecting the wrong one.
It did not remember what the first ten cards meant:
“So they say that the first 10 cards tell a story, but it’s been lost.”
It did not record who built the wrapper, or who audited it, or why five cards are missing from card 30’s count, or whether Hunt knew about 17b in 2017. It cannot tell you whether a billboard of user-painted tiles is art. Fifteen things in the back of this book could not be established from it, and fourteen more are places where the record and the memory disagree, and the chain adjudicates only some of them.
A thousand years of the wrong sentence
There is one more preservation in this story and it is the strangest, because it is the only one anybody did deliberately.
In a decommissioned coal mine in Svalbard sits the Arctic World Archive, and in it — according to TopDog Studios, who put it there, and according to Curio Cards, who say they are in it — is a reel of silver-halide film carrying the Curio Cards artwork and a register of who held which card on a particular day. Chapter thirteen has the documentation and the gap in it. The claim is the project’s own; the archive’s own announcement does not name them.
What is worth noticing here is what else went onto the film.
TopDog’s partner listing describes Curio Cards, in the words of the people doing the archiving, as “First NFTs on Ethereum that helped define the ERC721 and ERC1155 standards we know today.”
Hold that against chapter ten and chapter fourteen. ERC-721 names Curio Cards as an example of what a non-fungible token is not. ERC-1155 does not mention Curio Cards at all. And being first on Ethereum is a claim this book spent a chapter declining to defend, because Terra Nullius, Etheria and PixelMap are all earlier and nobody on any side has reasoned through the last of them.
So the sentence etched onto thousand-year film alongside the pictures is a sentence that is wrong in all three of its parts.
This is not a complaint about TopDog, who were doing a generous thing and were repeating what the press and the auction house had told them — chapter fourteen traces that hedge hardening outlet by outlet until Christie’s dropped the qualifier entirely. It is simply the clearest demonstration available of what archives actually do. They are indifferent. They preserve the artefact and the mistake at the same resolution, for the same thousand years, because a medium cannot tell the difference between a picture and a claim about a picture.
The chain, for its part, recorded only what happened: thirty-one contracts, in an order, at timestamps. It made no claim about being first, because it cannot. Everything in this story that turned out to be wrong — the supply figures, the eclipse date, the footnote’s meaning, the 826, the archive’s own partner description — was added by people afterwards, trying to say what the thing meant.
And one more time, for the symmetry: the footnote in ERC-721 that said Curio Cards were not the thing is a substantial part of why the project remained findable at all. The sentence that got it wrong preserved it. The sentence on the film gets it wrong in the other direction, and will preserve it too.
What the chain did is cheaper and more useful than memory: it kept a receipt. Travis Uhrig’s line about the forum thread is the right instruction for anyone reading this:
“And they went into the comments on that forum are rich by the way you can go back and read the skepticism when we announce the project and compare it with the current situation.”
The skepticism is still there, dated May 2017, under the announcement, next to a project that opened a Christie’s sale. Nobody has to be believed.
17:22:51
The book began with a timestamp and should end with what it is a timestamp of.
At 17:18:04 UTC on 9 May 2017, a wallet that had never transacted before deployed a factory contract. Four minutes and forty-seven seconds later it called that factory to create card 1, Apples, by an artist called Phneep, with the IPFS hash sitting in the constructor arguments where anyone can still read it. The transaction cost about four dollars in gas. The cards it created went on sale for twenty-five cents.
Within four months the project was effectively over. Within fourteen months one of its founders was describing it in the past tense as an idea they nearly had. Within nine months a sentence in a draft specification would name it as an example of what a non-fungible token is not, and that sentence would outlive the company, the website’s shop links, the founders’ involvement, and the entire market the project had been built for.
Four years and five months after the factory went up, the same thirty-one contracts — none of them maintained, none of them upgradeable, all of them still answering — were lot one at Christie’s, and the auctioneer called bids in ether for the first time in the firm’s history.
Hunt did not say the thing about footnotes at the auction. He said it in July 2018, on his own show, to an audience of nobody, about a project everyone including him had given up on, and the caller he was talking to 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.”
He was wrong about what the footnote says. He was right that it was forever, and right that it was a footnote, and right about the only thing that turned out to matter, which he had already posted on 26 September 2017 to no one at all.
Collecting is intensely human.