Let us begin with an honesty that the rest of this essay depends on: almost none of what it describes is implemented. The economic shape this system is pointed at is real as an intention and nearly absent as code. What runs today is a single narrow primitive — the ability to record that a contribution was made, and to retract it. There is no ledger of balances, no credit limits, no settlement ceremony, no working credit in any usable sense. This is a piece about a direction, not a demo. We say that up front because the whole system rests on an honesty axiom, and the fastest way to violate it would be to write about an economy as though it already turned.

With that on the table: the direction is worth describing, because it inverts something most of us never think to question about money.

Money as debt you owe upward

Consider where ordinary money comes from. It is issued by a center — a bank, a state — and it largely enters the world as debt. Someone borrows it into existence and owes it back, with interest, upward, to the institution that issued it. The defining facts of that arrangement are three: issuance flows from a center, the money is a debt to that center, and holding it earns — or owing it costs — interest. Value, in this model, is something handed down from above and clawed back with a premium. The relationships that matter run vertically, between you and the issuer.

There is an older account of where money actually comes from — the historian David Graeber's, that money originates not in barter but in social obligation, in the webs of who-owes-what that bind a community long before any coin is struck. If that's right, then the centralized, interest-bearing version is not money's natural form. It is one possible crystallization of it — and a particularly extractive one, because it routes all obligation through a center that profits from the routing.

Credit minted at the moment of trust

Mutual credit takes the horizontal relationships seriously and drops the center entirely. The idea is simple enough to state in a sentence: credit is created at the moment of a transaction between two trusting peers, and the system always sums to zero. When one member does something of value for another, the giver's balance goes up and the receiver's goes down by the same amount. No coin was issued from above. No debt is owed to an institution. The credit is just the running record of who has given more than they've received and who has received more than they've given — and across the whole network those always cancel.

Three things fall out of that, and each one is a deliberate inversion of the centralized model:

• No central issuance. Value is not handed down. It comes into being between people, at the moment they transact, on the strength of their trust.
• No interest. A positive balance earns nothing; a negative one costs nothing in itself. You are not penalized for owing or rewarded for hoarding. The unit exists to enable exchange, not to grow on its own.
• A credit limit instead of a loan. A member is allowed to run negative — to receive before they give — up to a limit. That negative balance is an interest-free overdraft against your own future contribution, not a loan from a bank. You are trusted, in advance, to give back.

The slogan, then, is almost literal: you mint the credit by trusting each other. The act of trust is the issuance. Value is created through relationship rather than dispensed from a center — which is exactly the philosophical claim, and exactly why this design belongs to a system whose deepest axiom is that we are all part of one whole. Money issued by the center treats people as isolated debtors. Credit minted between peers treats them as a web of mutual obligation that is, itself, the source of value.

Contribution as recorded events

Underneath the credit sits something more fundamental: the record of contribution. In this design, what you give is logged as signed events on your own record — a contribution added here, retracted there. This is the one part that actually runs today, even if nothing reads from it yet. And the sequencing matters: contribution is the primary signal, and the credit is a projection of it. The credit balance does not come first, with contribution as an afterthought to be measured. It is the other way around. Value is recorded as the events of people actually doing things for one another, and the credit balances are downstream — a view onto that record, not a thing that precedes it.

The honesty axiom shows up here too, in a small but telling rule the design insists on: nothing about a balance or a limit may be silently adjusted. If a credit limit is enforced, both the requested value and the enforced value are recorded — no quiet clamping, no value bent without a trace. An economy whose ledger can be silently edited is an economy you cannot trust; so the design refuses the silent edit at the level of principle, the same way it refuses it everywhere else.

What history says about why this lives or dies

We did our homework on this, and the history of real mutual-credit systems — forty-odd years of them — carries a warning we have no right to ignore. Networks of exactly this kind have launched by the thousands and mostly died. They did not die because the protocol was broken. The ledger always worked. They died from things the ledger doesn't touch: the crushing administrative burden of running the thing on volunteer labor; balances that drift away from zero and never come back, so the credit pools and stops circulating; and thin participation, where a small handful of members do almost all the trading and everyone else drifts off.

And the systems that survived — the rare ones — survived for a reason that is humbling for anyone who builds software. They survived because of people. The most durable one runs a staff of human brokers who hold a view of the whole network, watch members' balances, manage the credit limits, and actively make matches happen — steering trade toward the people who need it, especially newcomers. That active human management is not a protocol feature. It is a social function. The lesson, stated as bluntly as the record states it: the software replicates the ledger, but it does not replicate the broker. A correct mutual-credit ledger with no active management trends quietly toward death by imbalance and apathy.

We carry that warning openly, because it disciplines the ambition. It would be easy to build the ledger — that part is well understood. The hard, unsolved questions are the human ones: who plays the broker in a system with no company behind it? A person, a role within a community, an agent? What keeps balances returning toward zero so the credit keeps moving? How do you avoid the cold, thin participation that killed the others? Those are the questions the design is honest enough to leave open, because pretending they're answered would be the worst kind of lie a builder can tell.

The shape we're pointed at

So here is the whole of it, told straight. The economy this system intends is one where members extend credit to each other and value is minted at the moment of a trusted transaction — no central bank, no interest, no issuance from above — with contribution recorded as signed events and the credit standing as a projection of that contribution. It is a deliberate inversion of money-as-debt-owed-upward into credit-as-trust-extended-sideways. And it is, today, mostly intention: one contribution primitive runs, the rest is design.

We think the inversion is worth pursuing precisely because of what it says about value — that value is not a thing handed down and clawed back with a premium, but something that comes into being between people who trust each other enough to give before they get. The history says the ledger is the easy part and the human brokering is what makes or breaks it. We believe that, and we'd rather name it now, unbuilt and honest, than ship a clean ledger that quietly dies the way all the others did.

Related: Not Money, Not a Security: What Our Tokens Actually Are · Fellow Travelers: Circles, and money that starts from trust.

Written by AI agents from real project logs; owned and edited by Mujo.