The Obligation Ontology

Referrals, paybacks, debts, and claims are one model

by N. P. Clarke

Every reciprocal system is the same graph

Strip away the domain and the same shape is always left standing. Value moves from one party to another. The movement creates a claim of return. The claim is carried somewhere — a memory, a ledger, a contract. Something backs it. It is honored, or it is broken. And either way, the outcome travels outward through the network as trust.

A warm introduction. A twenty-dollar loan. A kindness on a bad day. A mortgage-backed security. A favor between nations. A parent's sacrifice. These are not analogies for one another. They are the same object with different values in its fields, differing only in which currency, at what scale, and under whose norms settlement gets judged.

State that as strongly as it deserves: obligation is not a metaphor borrowed from finance and applied loosely to human life. Finance is one specialization of obligation, and a late one. The debt of gratitude is older than the bond market, and it carries the more complete instruction set.

The objects are few

The whole model runs on a small vocabulary, and the smallness is the point.

A party is any node that can give, owe, or be owed — a person, a couple, a household, a firm, a nation, or the network itself. A currency is the denomination of what flows: money, trust, reputation, favor, attention, care, time, information, access, status. They convert into each other, but never cleanly, and the friction between them is where most human trouble lives. A transfer is a directed flow — an intro, a loan, a kindness, a payment, a vouch, a sacrifice.

Each transfer creates an obligation: the expectation of return. That is the generalized debt, and it does not require a contract to exist. A remembered kindness has the same structure as a promissory note; it simply has no signature and no court. Obligations accumulate on a ledger — the emotional bank account, to borrow Covey's name for it, the loan balance, the reputation, the trade balance, the running tally two people keep without ever admitting they keep it.

Every obligation carries terms: when it comes due, on what condition, whether it can be forgiven. It is backed by collateral. It ends in settlement or in default, and default is never local — a broken promise withdraws trust, and trust is what the whole graph is transmitting.

The rest follows. A rating is the probability that a party honors its obligations — credit score at one scale, reliability at another, a track record of honored reciprocity at a third. One concept, three costumes. Norms decide what kind of claim you are in at all: gift economy, market, family, treaty.

Trust is the universal collateral

Follow the backing of any obligation down far enough and it terminates in the same place. Money is collateralized by assets, but assets are only worth what someone will honor them at. A favor is collateralized by future reciprocity, which is a bet on character. A sovereign obligation is collateralized by an economy, which is a bet on a population's continued willingness to cooperate. Underneath every currency, the thing actually pledged is a belief that the other party will do what they said.

That is why trust behaves like an asset class and not like a mood. It can be accumulated, spent, borrowed against, leveraged, and destroyed. And it has an asymmetry no other collateral has: it builds slowly and breaks fast.

Obligations can be operated on

Obligations are not static records, and the operations on them are the same everywhere. You can assign one — sell a debt, or pass a favor along to a third party. You can pool them, as a bank's loan book does and as mutual aid does. You can tranche them, deciding who gets satisfied first, which in money is a capital structure and in a family is who gets helped when there isn't enough to go around. You can net two claims against each other, which is what "we're even" means. You can forgive one, which finance calls a write-off and the rest of life calls grace. And you can insure one — a co-signer, a guarantee, a mutual friend who says I've got your back.

The whole of structured finance is these operations applied to money-denominated obligations at scale, with lawyers. The same operations applied to trust and favor are friendship, family, and community. Nobody writes them down, but everyone runs them.

Same ontology, different dials

Between a couple's emotional bank account and a sovereign debt crisis, the difference is not one of kind. It is four settings: the scale of the parties, the currency, the governing norms, and whether the claim is enforceable by anyone but the participants.

A pair trading in trust, collateralized by relationship history, rated by reliability, defaulting into a betrayal that poisons the wider circle. Firms trading in money, collateralized by assets, rated by agencies, defaulting into write-downs that cascade through a market. Same columns, different values. That is the whole distance between a hug and a bond market, and it is smaller than it looks.

Every obligation runs on a clock

Scale-invariance alone would give a flat map. Two more dimensions make it a living thing, and the first is time.

Every obligation carries a horizon — instant, short, generational, perpetual — and a dynamic. Favors and grudges compound like interest. Care accrues across a life. A thank-you owed tomorrow and a debt of gratitude owed forever are the same object at different horizons. Time-value applies to every currency, not only to money.

Obligations nest

The second dimension is the one that makes it a universe rather than a map: every obligation is a whole that is also a part.

A daily kindness sits inside a marriage, inside a family, inside a community. A single introduction sits inside a chain of introductions, inside a network of them. A household loan sits inside a bank's book, inside a financial system. The same objects and the same operations apply at every level, which is exactly why the model generalizes.

Two consequences fall out of that. Timescales nest with the obligations: a lifelong bond is millions of daily deposits, a thirty-year mortgage is three hundred and sixty monthly ones. And contagion runs vertically, not only sideways — a personal default stresses a household, then a community; a sovereign default reaches banks, then firms, then households, then marriages. Trust ruptures travel up and down the nesting, which is why distant financial events show up in kitchens.

The referral is the first currency

This is where the ontology meets the method. SparkTorch's central claim has always been that the hero is the Handoff — the seam where one person's trust becomes another's. Read it through this model and the Handoff is a transfer of collateral. Nothing else is actually moving. The Spark has accumulated trust inside a relationship and pledges it on someone else's behalf; the introduction is the instrument.

That is what a referral is: an obligation deliberately created between people, denominated in reputation, backed by the relationship history of whoever made it, and settled only when the person vouched for honors the vouch. It is why a referral that goes badly costs the introducer more than anyone else — they posted the collateral. And it is why seam integrity is measurable at all. You are watching whether the pledged trust survived the transfer.

Referrals are this model's first currency because they are the one everyone already understands. They are not its boundary. The same objects run underneath a friendship, a career, a balance sheet, and a treaty.

What the machines are for

None of this changes where the flame lives.

A machine can carry the ledger — remember what was promised, who is owed, what is coming due, where trust is quietly eroding — and that is worth a great deal, because most obligations fail from forgetting rather than from bad faith. But a machine cannot post the collateral. It has no relationship history to pledge. When it vouches, nothing is at stake.

So the crew keeps the books and the humans keep the promises. Empower, never replace — in this frame not a slogan but a structural fact. The one thing that cannot be delegated is the thing being pledged.

Reciprocity, made legible.

Where this stands, and where it is standing on someone else

None of the above is offered as a discovery. The claim is one of arrangement — that a set of things already studied separately are one thing — and the honest way to make that claim is to say plainly whose shoulders it is made from, and where the existing evidence pushes back.

The lineage. Mauss established in Essai sur le don (1925) that gift exchange is obligatory rather than free, and that the threefold obligation — to give, to receive, to reciprocate — is the structure of the thing [8]. Gouldner made the norm of reciprocity a general sociological principle rather than a curiosity of archaic societies [9]. Sahlins sorted reciprocity into generalized, balanced, and negative, and — crucially for the model above — keyed which one applies to social distance rather than to the goods involved [10]. That is the norms dial, named half a century early. Fiske proposed that people run a small number of distinct relational grammars and switch between them by context, one of which, equality matching, is recognisably reciprocity [11]. Graeber argued the historical case that debt precedes money rather than descending from it [12]. Covey supplied the emotional bank account [13]. This essay is a synthesis across that lineage, not a replacement for it.

The strongest empirical support is for the cross-domain claim. Among the Tsimane' of Bolivia, cooperation across five distinct domains — meat sharing, produce sharing, field labour, childcare, and sick care — was best explained by models that included exchange in kind and across domains, with reciprocal exchange having a strong effect independent of kinship and proximity [1]. That is the essay's central assertion observed directly: people run one exchange system across incommensurable currencies, and it behaves market-like without being a market. More recent work on multilayered support networks describes the same phenomenon as multilayered reciprocity — different resource types exchanged across domains — and shows how asymmetries in it produce durable status structure [7].

The claim that trust is the collateral has direct support. Molm's experimental program finds that the reciprocal form of exchange, independent of close ties or personal association, is what produces strong trust and affective bonds [4]. Trust is not a by-product of the relationship; the form of the exchange generates it.

And the referral claim is the best-supported part of the essay. A vouch is generalized indirect exchange — you help someone who is not the person who helped you. Comparing the three forms head to head, generalized indirect exchange produced greater solidarity than reciprocal direct exchange, which in turn beat negotiated direct exchange [5]. That ordering is the case for the Handoff, made by someone else, before anyone was selling a method: the least contractual form of exchange builds the most trust. It also explains the asymmetry the practitioners already feel — that a bad referral costs the introducer most.

Where the literature pushes back, and it should. Molm's core finding is that reciprocity is structured and variable across forms of exchange, and that those variations have profound and different effects on trust and solidarity [2]; she argues explicitly that theories built on any single form of exchange are limited by it [3]. Read carelessly, that is an objection to this essay. Read properly, it is the essay's own point turned into a warning. The claim here is that obligation shares a common structure — the same objects, the same operations. It is emphatically not that the forms are interchangeable or that their consequences are the same. They are not, and the differences are exactly what the four dials are for. Anyone who takes "it's all the same graph" to mean a favour can be settled like an invoice has read the model as a licence when it was meant as a map. Reciprocity also varies by culture in ways this essay does not attempt to capture — whether people escalate reciprocity or return it in kind differs systematically across societies [6].

What remains unproven. The nesting claim — that the same operations hold at every scale from a daily kindness to a sovereign default, and that contagion runs vertically through those levels — is the most ambitious thing here and the least evidenced. It is offered as a conjecture worth testing, not a finding.

References

Empirical work, retrieved and checked:

The lineage: