Altcoins with
pictures
“It’s extremely simple. They are altcoins with pictures. Anything suggesting otherwise is larp and cope.”
He was right, and the people who argued back mostly proved his point by how hard they argued. Punk6529 gave the better answer: the combination is a feature rather than a bug.
What the sentence leaves out is the part that matters here. A token can be made liquid by making more of it. A picture cannot, and there is exactly one of most of the things we care about. Everything on this page is somebody trying to get around that, and most of it is the wreckage.
What we built
Seven things the trade already does, running at three in the morning
Every one of these does something in the opposite order to the internet. A launchpad prices the work and then discovers demand; a gallery discovers demand and then prices the work. That inversion is the whole series.
Nothing here is a new idea about art. Every mechanism below is something a good dealer, a good auction house or a good gallery has been doing for decades, and in one case for a century. They work. They are also done by hand, which means they are done slowly, unevenly, for people already in an address book, and only during office hours.
What we did was write seven of them down in a way a pool can execute. The edition is priced by the bidding rather than guessed at beforehand. The guarantee is funded before the sale instead of promised during it. The edition raises its own price. The market pays more for a work that was kept. A museum can exercise a right of pre-emption inside the same transaction that sold the lot. The person who made the introduction is paid on the sale rather than by telephone six weeks later.
None of it replaces judgement, and none of it is trying to. Choosing the work is still the whole job. This is the paperwork, and the paperwork now answers on a Sunday.
The gallery waitlist, forever
The waitlist
A dealer shows the work to twelve collectors, sees who is serious, and only then decides how large the edition is and what it costs. Here the bids are funded and public, and everyone who clears pays the same price.
Open the room
Supporting an artist at auction, for a century
The house bid
A dealer who believes in an artist buys their work at auction to stop the market breaking. Here the money sits in the pool at a published price and cannot be withdrawn by anyone, including us, and how much is left is on the wall.
Open the room
Auction guarantees, since the 1970s
The guarantee
A third party agrees in advance to buy the work at a set price if nobody else does, and is paid for standing there. The money is posted before the sale rather than promised during it.
Open the room
Gallery price ladders, since the 1960s
The edition
An edition sells well and the price goes up. Every gallery does this by hand, slowly, and only while somebody is paying attention. Here each copy sold moves the next one.
Open the room
The saleroom premium, as old as the saleroom
Fresh to market
A work held thirty years carries a premium; one that came back too soon takes a discount. The trade has always priced this in a whisper. This prices it at the moment of sale.
Open the room
Droit de préemption, France, 1921
The right of pre-emption
The hammer falls and the State takes the work at the price the room just agreed. Here a museum can, in the same transaction, and the buyer is made whole to the wei.
Open the room
The introducing dealer, forever
The curator’s share
Somebody sends a collector to a gallery and is paid out of the margin, by telephone, weeks later, at a number nobody else sees. Here it is one per cent, on the sale, in public.
Open the room
Every one of them is a real contract with a real test suite, and each room says plainly what it costs and where it fails. None is deployed to a public network yet.
The problem
A floor is not a bid
A floor price is the cheapest thing somebody is asking for. It is not money and it disappears the moment they change their mind. For a work that exists once there is nothing identical to quote against, so on most days, for most pieces, the honest bid is nothing at any price.
The art world answered this a century ago by paying a person to carry the risk. A dealer takes the work on consignment, holds it for months and finds a buyer. At auction, a guarantor agrees in advance to buy at a set price if nobody else does, and is paid whether or not it happens. Both work. Neither answers on a Sunday, and neither scales past an address book.
What was tried
Six ways to build a bid, and what each one became
Lend against it
Borrow against your picture and keep it. Liquidity without selling.
You keep the price risk, and every loan in the book is written against the same floor. Over one weekend in August 2022 BendDAO's depositors withdrew almost everything, its reserves fell to a handful of ETH, the collateral it had already seized drew no bidders at all, and it voted mid-crisis to drop its liquidation threshold from 85% to 70%.
Break it into shares
A thousand people can own a piece of one picture, so the picture becomes tradeable.
What trades is the shares, and they trade as an index of the floor, so the premium that made the work worth owning evaporates. Worse, the way out often cannot be reached at all: buying the picture back needs a majority of the shares to vote at once, and spreading them across a crowd is exactly what makes that impossible. The vault holding the original Doge photograph needs half its tokens and has 0.67% participating. Nobody who owns a piece of that picture is ever going to be paid for it.
Bid on the whole collection
Real money, standing, funded. A floor you can actually hit.
The offer is for any piece, so it is for no piece in particular, and it gets filled with the worst one every single time. The market maker absorbs that until they stop, and then the floor was never there.
Make a futures market on the floor
Trade the direction without touching the art.
The thing underneath is an index, and nobody can want an index. There is no settlement in the object, so the market has no natural counterparty and almost never found users.
Pay people to make a market
Bootstrap the bids with rewards until the market stands on its own.
The bids were for the rewards. When a programme changed, the bids left with it. Nobody has yet shown liquidity of this kind outliving the subsidy that created it.
Have people stake money behind a valuation
A price with real capital behind it, so appraisal stops being an opinion.
The capital was there for the reward rather than for the work. A number backed by someone who does not want the object at that price is a number, not a bid.
What they have in common
Money that does not want the object cannot hold a bid on the object.
Look at who was standing on the buying side of each one. Lenders who wanted a return and not the picture, so when the loans went bad nobody wanted the collateral. Funds holding a share of an object rather than the object, with nobody to sell it to. Offers that applied to any piece, which is to say to none. Traders who wanted an index. People who wanted the rewards.
The auction guarantee is the only version of this instrument that has worked, and it has worked for generations. The only structural difference is who stands behind the price. A guarantor is almost always a serious collector of that artist, so being wrong means owning something they wanted anyway.
That is the whole thing. Not a mechanism problem, a counterparty problem, and no amount of code has ever produced somebody who wants the picture.
What we built
Small enough to explain in a sentence
Uniswap v4 lets you attach rules to the room where a trade happens. We spent a long time designing rules for an edition: a price that comes down on a published schedule, a premium for choosing a low number, protection for whoever keeps an offer open. Then we checked each one against the code that would have to run it.
Almost none of them can be built. The descending price is decoration, because our asking price can never sit below what the room is already trading at. The protection for patient money is not something the plumbing permits anyone to do. And a price we published as the honest appraisal turned out to be movable for a fraction of what it valued.
What survives is a vault and one rule: you cannot buy half a picture. Two thousand bytes. It sets no prices, keeps no list, takes no cut, and runs no clock.
The one thing worth keeping from all of it: you can read what that contract is allowed to do from the last character of its address. It is built so that it cannot be paid by a trade, and anyone can check that before trusting a word we say.
Where we were wrong
Three of ours, in order of how much they cost
Everything above is other people. It would be a poor page that only listed those.
01
A room that was a lottery
Cost: the whole design.
We built a room where you bought a ticket and won a piece from the wall at random. It was finished, tested and never opened.
Paid entry, plus chance, plus a prize is a lottery in most legal frameworks, and a verifiable random draw does not change that. It was counsel's call and not the code's, and no amount of engineering was going to answer it.
02
A fix that paid you to lie
Cost: the idea underneath it.
In that room, the money you put behind your own work also set how often it came up. Promise a lot and your piece was drawn almost never. Promise little and you were giving it away. So anything genuinely good sat at the top, unreachable by design, which is the flaw we most wanted to fix.
We tried setting the odds from the gap between what a work was worth and what its owner had promised, so a masterpiece and a cheap print would come up equally often. Then we did the arithmetic properly.
Saying your work was worth less than it is saved you about a ninth of its value, every single time it sold. Not on average and not at certain prices. Always, by the same amount, with nothing else about your position getting worse. A room where understating your own work is the winning move is not a room a gallery should open, so we did not.
03
A contract you could walk off with
Cost: nearly a great deal more.
In the version that replaced all that, a buyer can change their mind: hand the work back and take the money that was set aside for exactly that. Handing it back needs the buyer to grant permission, because by then the picture is theirs.
We wrapped that step in a catch, so that a seller whose own wallet could not accept the work back would not be able to trap the buyer's money. A reasonable worry, and a real one.
But the buyer is the one who grants the permission. Simply withhold it, the handover fails, the catch quietly absorbs the failure, and the payment goes out anyway. Keep the picture and take the money. It is fixed, there is a test named after the attack, and it was found by someone writing the attack rather than reading the code.
None of the three was found by thinking harder about it. The first was a lawyer's answer. The second came out of running the numbers rather than admiring them. The third was found by writing the attack and watching it work against our own contract.
And twice more, in smaller ways, the careful choice turned out to be the hole. A flag we set to handle failures gracefully turned a total breakdown into a page that said there was nothing here. A rule written to protect the artwork blocked a legitimate way of buying it. Every one of those was added on purpose, by someone being careful, which is the part worth remembering.
What we think it means
A decade spent trying to replace judgement with machinery
Every mechanism on this page was built to work without anybody deciding anything. That was the point of them: a formula that would produce a bid whether or not a human being wanted the picture. None of them managed it, and the reason is the sentence in the middle of this page. Money that does not want the object cannot hold a bid on the object, and wanting is a judgement. You cannot assemble it out of a curve.
The old answer has judgement at the centre of it. A dealer takes a work because they believe they can place it. A guarantor stands behind a price because they would be content owning the thing. Both are a person deciding, in public, with their own money at stake. That is why they work.
And the last clause is why they are not enough. They reach exactly as far as an address book. A Basquiat gets a guarantor. A four thousand euro photograph does not and never will, because a person's time does not scale down to it.
So the job was never to make pictures liquid. It was to carry the thing that already works to the works and the people it was never going to reach. A contract does not care whether the number is four thousand or four million, and that is the only real advantage on offer here.
It also puts the order back the right way round. When liquidity is the product you have to accept any picture, because you need volume, so you quote for a collection instead of a work, so you are handed the worst one every time, so you stop. Every failure above starts there. When the art is the product, standing behind a price is a service you offer on the pieces you believe in, and you are allowed to say no.
That is what a gallery has always been. It is the one thing none of the protocols on this page could do, because not one of them was permitted to have taste.
Make good work. Choose it carefully. Then let anyone stand behind it. In that order, and the last part is the only one that needed inventing.
What is still open
Whether anybody wants the other side
Everything above serves one question, and no contract answers it. Will somebody put real money behind a named work they do not own, and be paid for holding that offer open?
That is the auction guarantee, opened to anyone, at a price they choose themselves. We do not know whether it has a buyer. We are asking ten collectors before we deploy anything, because a testnet cannot answer it and neither can we.
The reasoning behind all of it, with the holes left in, is at the problem, the arithmetic we published about somebody else's pool is at the pulse page, and the room itself, running on a local chain with real contracts and no invented numbers, is at consign.