Recognitium
FR Create account

One Word in the First Paragraph

Bitcoin begins with a belief. This essay is about one word in it.


Bitcoin begins with a diagnosis. In its first paragraph, before any block or any proof of work, the whitepaper says: “Completely non-reversible transactions are not really possible, since financial institutions cannot avoid mediating disputes.” Everything that followed was built to answer that sentence. This essay is about one word in it: since.

The sentence draws a conclusion. Because banks must mediate disputes, payments cannot be final. The conclusion does not follow.

A dispute exists only because a payment cannot be undone. If an act could truly be made never to have happened, there would be nothing to dispute; it would simply be erased. No one has ever done that. A chargeback does not erase a payment; it presupposes it, and adds a second payment that answers the first. A refund, a recall, a court order: each is a new act that cites the old one. Accounting has known this for five centuries: an error is corrected by a reversing entry, never by an eraser.

Transactions were never reversible. They were always answered.

So banks cannot avoid mediating disputes, and transactions are irreversible, and the two facts do not contradict each other. The second causes the first. Because what is done cannot be undone, someone must decide what is done next.

What the first paragraph actually describes is something else: a record that its keeper can rewrite. A bank’s ledger can be edited by the bank. That is the defect, and it is why the bank becomes judge, defendant and archive at once.

The problem was never mediation. It was the pen.

What decentralization was for

The whitepaper says plainly why the bank had to go. In the old model, “the mint was aware of all transactions and decided which arrived first.” Without it, participants need “to agree on a single history of the order in which they were received.” Decentralization was the means. The goals were three: one order that no one can rewrite, one history that is the same for everyone, and no one able to refuse your transaction.

Bitcoin reached them by making the pen too expensive to lift. It worked, and nothing here diminishes it: Bitcoin solved double-spending without a bank, and Recognitium writes its own history into Bitcoin every day. But Bitcoin paid for permanence with energy and priced it in its own token, one confirmation at a time, too slowly and too dearly for a payment of one cent. Every chain built since has refined that answer: faster blocks, cheaper fees, more programmable ledgers, all standing on the same wall.

The same three goals, beneath the chain

Recognitium takes the pen away from everyone instead, and meets the three goals without a vote.

One order that no one can rewrite. Every receipt takes its place in a single public order. Each party keeps its own records and data; when two parties transact, each commits the other’s position into its own receipt, so their histories interlock. The state of the whole is summarized in one root, the chain is anchored in Bitcoin every day, and anyone can verify any receipt offline, for free, without asking us. Sealing costs one cent and takes milliseconds.

One history for everyone. There is one public chain, and the receipts that prove it are in the hands of everyone who was ever served. To show two parties two different pasts, the operator would have to make one of them disappear from hands it does not control.

No one able to refuse you. The engine receives a fingerprint and nothing else. It does not know whose it is or what it stands for, so it cannot single out a person or a content; anyone can seal anyone’s fingerprint. Its only power of refusal is to stop serving everyone at once, and that power disappears as operators multiply and the instrument comes into the client’s own hands.

Strangers need nothing more. Two people who will never meet do not have to know each other to settle. If someone spends the same thing twice, both spends take their place in the same public order, and the later one loses. No identity is required and no vote is held.

The operator is left with two verbs: on and off.

That is why this work sits beneath Bitcoin, not above it. Blockchains and tokens refine one answer to double-spending, a wall paid for in energy. Recognitium goes beneath that answer, to the problem it was built for.

What remains

Ask what a forger would have to do. To change a receipt that someone else already holds, he must either find a second preimage of SHA-256 at that exact position, which is the hardest form of breaking it, or reach every hand that holds the chain and replace every copy. The first is beyond any computer we know how to build. The second is beyond computation altogether. And the instrument itself is built to have only two behaviours: an honest receipt, or silence.

In that instrument, the secret behind each receipt is drawn from a physical event that no one could predict, is larger than the fingerprint it leaves, and is erased before the receipt can be read. It has no copy before, no trace in the output, and no existence after. Nothing rests anywhere to be stolen, and no one, the operator included, ever knew it.

That instrument is where Recognitium is going, and it is being built now. Today’s receipts are sealed in software. The physical guarantees become claims when they are measured, and their result will be published, whatever it is.

What changes

Because a transaction cannot be undone, every dispute must end in a new act, and that act is sealed like the first: final, ordered, visible to anyone who holds the payment it answers. Mediation does not disappear. It becomes mandatory, because it is the only way left to answer a payment, and irrevocable, because its answer enters the same order as everything else.

The mediator no longer needs to hold the ledger. Anyone with standing can decide; no one can rewrite.

The ledger that holds the asset

Someone must still hold what is exchanged: the securities, the money, the title. Recognitium does not hold it, and does not need to. It needs one rule: the ledger that holds an asset may move it only according to the receipts. Each time its state changes, the ledger seals the Merkle root of its new state. Anyone can replay the receipts in their public order from the previous root and must arrive at the new one. A ledger that moves an asset without a receipt seals a root no one can reproduce, and everyone sees it.

The rule holds at two strengths. Today, in software, the ledger can still move on its own, but never unseen. When the instrument sits in the ledger’s write path, the ledger cannot commit a state that no receipt binds: it follows the receipts, or it falls silent. The custodian keeps the asset and loses the power to move it alone.

An exchange of an asset against payment follows from the same rule. Both legs cite the same receipt, and the next root of the asset ledger and the next root of the cash ledger must each contain its leg. If one moves without the other, two public roots disagree with one public order. Each institution keeps its books private; only the order is shared.

The ledger may hold the asset. It can no longer move it unseen.

What came before

None of this starts from nothing. Stuart Haber and Scott Stornetta chained the hashes of documents in 1991, and Bitcoin is built on their idea. Certificate Transparency has run public, verifiable logs kept by named operators since 2013, so that anyone can catch a log that shows two histories. Eric Budish showed in 2018 that the security of proof of work must keep pace with the value it protects, which is the cost of tying permanence to a token. Rollups already prove that a state follows an ordered list of transactions, and exchanges publish their reserves as Merkle sum trees. What is new here is narrower and deeper: the first paragraph’s premise removed, an engine too blind to refuse anyone, an order built to bind every position to a physical event no one could predict, and ledgers that may move an asset only along that order.


Satoshi asked how to pay without trusting a bank. The question beneath it is older: how can people who do not trust each other share a past that no one can rewrite? Bitcoin answered with energy. Recognitium answers with order.

Signature of José Niño José Niño Founder of Recognitium October 2026 · Paris, France

References

  1. Satoshi Nakamoto, “Bitcoin: A Peer-to-Peer Electronic Cash System”, 2008, sections 1 and 2.
  2. Stuart Haber and W. Scott Stornetta, “How to Time-Stamp a Digital Document”, Journal of Cryptology, vol. 3, no. 2, 1991.
  3. Ben Laurie, Adam Langley and Emilia Kasper, “Certificate Transparency”, RFC 6962, IETF, June 2013.
  4. Eric Budish, “The Economic Limits of Bitcoin and the Blockchain”, NBER Working Paper 24717, 2018.