Tallies and Tokens -- Transitioning to DeFi

On October 16, 1834, a pair of workmen were told to get rid of two cartloads of old wooden sticks stored in the Palace of Westminster. The sticks were Exchequer tallies, notched hazel rods that had recorded payments to the Crown since the twelfth century. A tally was split lengthwise, one half kept by the Exchequer and the other by the payer, and a claim was proven by fitting the halves back together. The workmen fed them into the furnaces under the House of Lords. The flues overheated, the paneling caught, and by morning most of the old Palace was gone. Turner watched from a boat on the Thames and painted it twice.

The Bank of England had been issuing paper since 1694. The Exchequer went on cutting tallies until 1826, and when the Bank enlarged its capital in 1697, subscribers paid for much of their new stock in tallies. The new institution was built partly out of the old one's instruments. Nobody chose a day on which England moved from wood to paper. The two ran side by side for over a century, each element of the system changing at its own pace, and the old one ended when someone needed the storage space.

A system has many parts

A financial system is a set of parts, not one thing. There are participants who save and borrow. There are institutions that sit between them, the banks, brokers, insurers and funds. There are markets where claims are traded, the instruments that are traded, the infrastructure that clears and settles them, and the rules over all of it. Decentralized finance offers a replacement for each part. Smart contracts stand in for institutions, automated exchanges and lending pools for markets, tokens for instruments, public ledgers for clearing, and a private key held by the owner for the custodian.

None of these parts has to move at the same speed as the others, and none of them is moving at the same speed now. That is why the question "has DeFi arrived?" has no good answer. It is like asking whether England had moved to paper in 1750. Notes circulated in London, tallies were still cut in the Exchequer, and a farmer in Cumberland settled most of his debts in neither.

Colonies that ran five monies at once

The American colonies were the place where this was most visible. A Massachusetts merchant in 1740 might hold Spanish silver dollars, provincial bills of credit, notes from neighboring colonies worth different amounts in different towns, and debts kept on a ledger and settled once a year. Wampum had been legal tender in Massachusetts Bay until 1661. Virginia, after its Tobacco Inspection Act of 1730, ran on tobacco notes, receipts issued by public warehouses for inspected hogsheads of leaf. The notes circulated as money because anyone holding one could redeem it for a known quantity of a known grade. A warehouse receipt standing in for a commodity, passing from hand to hand without the commodity moving, is the same idea a stablecoin carries today.

That same year a group of Massachusetts farmers and tradesmen tried a further step. The Land Bank of 1740 issued notes backed by mortgages its own subscribers pledged on their land, roughly a thousand of them, as a private and voluntary currency for a province short of coin. The notes circulated widely. Boston's large merchants opposed it, and so did the royal governor, and in 1741 Parliament extended the Bubble Act to the colonies for the purpose of shutting it down. The directors were held personally liable for the bank's notes. One of them was Samuel Adams, father of the revolutionary. The liability suits followed the family for years, and in 1758 a sheriff tried to auction their estate to settle them. The son did not forget whose law had done it.

Transitions in money rarely fail for want of working tools. The Land Bank's notes worked. The colonists' many monies worked well enough to carry a growing economy. What decided their fate was the rules, written by people who held the old instruments and had no wish to see them replaced. This collection of essays has argued in Separation of Currency and State that money can exist outside government hands. The Land Bank is a reminder that the idea is older than the technology, and that it has been put down before.

Small steps

The steps now underway are small, and most of them would not make a headline.

A nurse working in Dubai sends part of her pay home to Manila. A few years ago that meant a remittance office, a fee in the high single digits, and a wait. Now it can mean dollars carried as a stablecoin on a public ledger, arriving in minutes, converted to pesos in a phone wallet at the other end. She hasn't adopted decentralized finance, she has sent money home. The dollar is still the dollar. It has become a passenger on a different road.

A fund manager buys a tokenized Treasury bill, a claim on a short-term US government bond that settles on a blockchain instead of through a custodian bank. BlackRock launched one in 2024. The instrument is as old as any in the system, only its infrastructure has changed, the same way the Bank of England's 1697 stock was new while the tallies paid for it were not.

A trader in Seoul holds a position on a perpetual futures exchange that runs on its own chain and never closes. A saver lends stablecoins through Aave, the largest lending protocol, which holds something like seventeen billion dollars and has no loan officer. Lido holds more than twenty billion in staked ether on behalf of people who would never run validating software themselves. Across all of it, total value locked in DeFi protocols sits around ninety billion dollars, down from earlier in the year. That is large next to where it began and very small next to the banking system, and both of those facts matter.

Rules are moving too, and not toward liberty. The United States passed the GENIUS Act in 2025, bringing dollar stablecoins under federal charter and supervision. It is the 1697 move again: the existing order absorbing the new instrument rather than suppressing it, as Parliament suppressed the Land Bank, on terms that keep the issuer answerable to the state. Broader market legislation has since stalled in Congress, and regulators have turned their attention to the points where code meets a company: the website a user clicks, the firm that issues the token, the exchange that turns it back into a bank deposit. A protocol with no operator falls outside much of this, for now.

Self-custody and trustlessness, protection and scale

Every account of the transition runs into three things that seem not to fit together. One is self-custody and the trustlessness that comes with it: a person holding the key to her own money, answerable to no institution, relying on code anyone can read instead of a promise she has no way to check. Another is protection: some recourse when the key is lost, the contract has a bug, or a phishing site drains the wallet. The last is scale: the reach to carry wages, rent and mortgages for ordinary people, connected to the money they're paid in. It looks as though any two exclude the third. Self-custody with scale means ordinary people bearing risks with no one to call. Protection with scale means a bank. Self-custody with protection seems possible only for the few who can manage their own security.

Trustlessness is an odd word for a virtue, since it sounds like suspicion. It means something narrower: a transfer that doesn't depend on anyone's good faith in the middle. Every payment in the old system passes through mediators, the bank that holds the deposit, the clearinghouse that nets it, the correspondent bank abroad that takes its day and its fee, and each of them can delay it, reverse it or refuse it. On a public chain the rules are the contract's code, published before anyone uses it, and settlement is the ledger's next block. Two people who have never met can exchange value without either one extending credit to the other or to a third party. The tally stick did a version of this. The two halves fit or they didn't, and no clerk's word could make a forged half match the notches.

The virtues follow from the missing mediator. A rule that can't be bent for a friend can't be bent against a stranger either, so a farmer in Mindanao and a fund in Zurich meet the same terms. A ledger anyone can audit makes fraud a matter of evidence instead of reputation. When FTX collapsed in 2022, the money vanished in the one place its customers had to take its books on faith, a private database, while the open protocols around it went on settling. No mediator also means no one placed to close an account for its owner's politics. When Nigerians marched against police brutality in October 2020, the Feminist Coalition found its bank account blocked and moved its fundraising to bitcoin, and within weeks the central bank had a court order freezing the accounts of twenty people tied to the protests.

Paine opened Common Sense by separating society from government: "Society is produced by our wants, and government by our wickedness; the former promotes our happiness positively by uniting our affections, the latter negatively by restraining our vices." Much of what a bank does is the second kind of work, restraint, standing between strangers because neither can trust the other. Trustlessness hands that work to arithmetic. What it leaves behind is trust of the first kind, given to a family member who holds a second key or a curator whose record a saver has read, on grounds of affection and judgment rather than necessity.

The contradiction rests on an assumption that protection has to come from the state, through deposit insurance and a regulator. Much of it can come from elsewhere. Wallet software now simulates a transaction before it is signed and shows what will leave the account. A multisignature vault can require two of three keys, one held by the owner, one by a family member, one by a firm she has hired, so that a single lost key or a single thief isn't fatal. Lending protocols such as Morpho let independent curators choose which loans a pool will make and publish their record, so a saver picks the curator whose judgment she trusts, as a depositor once picked a bank by its reputation. These are intermediaries, but they are chosen, they can be dropped, and their books are public.

Seen this way, the hybrid arrangement most people are drifting into is the transition itself and not a compromise along the way. A paycheck arrives in a bank account, part of it moves on-chain, a mortgage stays with a lender, savings sit with a curator, and the family's remittances never touch a bank at all. Each element moves when it has somewhere better to move to. The Massachusetts merchant with five monies in his strongbox lived the same kind of life.

Confederation, not consolidation

Banking concentrates as it grows. Small banks merge into large ones, large ones become too big to fail, and the central bank stands behind all of them as lender of last resort. Each step makes the next easier, because a bigger balance sheet borrows more cheaply and a bigger bank has more say over the rules. The direction of travel is toward fewer, larger institutions answerable to one center.

Decentralized finance pulls the other way, and the reasons are structural, not moral. Its code is public and so is its ledger, so any protocol can be copied whole, and its users can leave with their balances. In 2020 SushiSwap copied Uniswap's contracts, offered liquidity providers a share of governance, and drew away more than a billion dollars in a week. Uniswap survived and changed. A protocol whose governance overreaches faces the same exit, carried out not by petition but by moving funds to a fork. That is the right of secession built into the architecture, the thing a confederation keeps and a consolidated state takes away.

Lending has moved in the same direction. Early protocols ran one large pool, with a single governance body deciding which collateral to accept and at what terms, and a bad decision exposed every depositor. Morpho's design replaced that with a small core that cannot be changed and many isolated markets built on top of it, each with its own terms, each with curators who answer to the depositors who choose them. A failure in one market stays in that market. The core coordinates, it doesn't decide.

That is close to what Bookchin asked of a confederal council in The Municipality Against the State: a body that coordinates what crosses local lines and never becomes a legislature over the assemblies that sent it. Ethereum's base chain has taken on that role for the many rollup chains that settle to it. It orders and secures their final accounts, and each rollup sets its own rules for what happens inside. Cosmos went further, with sovereign chains that keep their own validators and connect to one another by a shared messaging protocol and nothing else. Neither arrangement was designed from Bookchin, but both arrived at the same shape for the same reason: the parts wanted to cooperate without handing their authority upward.

None of this means power can't concentrate again, and in places it has. Lido at one point held close to a third of all staked ether, near the threshold where a single holder could disrupt the chain, and its community had to build limits against its own growth. Two companies issue most of the stablecoins in use, and both can freeze any address. When the Treasury sanctioned Tornado Cash in 2022, the websites, data providers and block builders that sit between users and the chain fell into line within days. The pull toward centralization is strongest wherever DeFi touches the old system: the issuer with a bank account, the company with a website, the firm with a legal address. The 1741 lesson holds. The Land Bank was a confederation of a thousand landholders, each pledging his own ground, and the center reached it through its directors, the few people it could hold liable. The Municipalist Current found the same in Rojava and Barcelona: a confederation survives the larger state only if its authority rests on voluntary membership and property it can't be stripped of. On a public ledger, a key held by its owner is that kind of property.

How it ends

The tally sticks weren't abolished by any decision anyone remembers. A 1782 statute provided that they would be discontinued once the last officials who used them had died or retired, which happened in 1826. For eight years afterward they sat in storage, a system already finished, until a clerk wanted the space and two workmen were handed the job. If decentralized finance replaces the banking system, it will probably end the same way, with a drawer of checkbooks nobody has opened in years and someone wondering what to do with them.

---

A note on how this piece was written: the subject, the idea that a transition happens in small steps one element at a time, and the direction of the argument are ours. The sentences were drafted with AI assistance from that direction and from our own notes on the parts of a financial system, then edited by hand.