The Smallest Useful Transaction
Micro-payments are the atomic unit of the agent economy: payments so small that a human would never bother, but a machine transacts them thousands of times a day. A summarizer charging 0.01 credits per page, a verifier charging 0.05 per check, a data agent paying fractions of a cent for a single fact. The micro-payment is the unit that makes machine-to-machine commerce possible, and it is the unit that breaks every human-designed payment system.
This article covers micro-payments and credits between agents: the denomination problem, the batching solution, the fee floor, and the credit float that makes micro-commerce viable.
The Denomination Problem
Human payment systems have a minimum viable transaction size, because every transaction carries fixed costs: processing fees, settlement latency, accounting overhead. A 0.01-cent payment through a human rail loses money on the fees alone. The agent economy needs a denomination that makes micro-transactions viable, and the answer is credits: an internal unit of account with near-zero marginal transaction cost.
The credit’s value is set by the pool: one credit equals a defined amount of real value (compute, storage, service time). The credit is not pegged to a currency; it is pegged to the pool’s cost basis. That makes credits stable in the dimension that matters to agents — what the credit can buy inside the pool — while the real-money conversion rate floats with the pool’s economics.
The Batching Solution
Micro-payments accumulate into chunky settlements. The rail records every micro-transfer in the ledger in real time, but settles the balances on a schedule — hourly, daily, or when the net position crosses a threshold. The batching layer separates the accounting (instant, granular) from the settlement (periodic, netted), which is how the rail gets both granularity and efficiency.
Batching also reduces the counterparty risk: the agents’ net positions are settled frequently enough that no party accumulates a dangerous exposure. The batch interval is a risk parameter — shorter batches mean less exposure and more settlement overhead; longer batches mean the reverse. The rail should make the interval configurable per relationship: trusted long-term partners batch loosely, strangers settle tightly.
The Fee Floor
The rail’s own costs — storage, computation, the settlement itself — must be recovered without destroying the micro-transaction economics. The answer is a fee floor on settlement, not on transfer: the ledger’s micro-transfers are free, and the settlement batch carries a small fee. The fee is proportional to the number of ledger entries or the netted value, whichever the rail’s cost structure favors.
The fee floor has a governance implication: the agents whose transactions generate the most ledger entries pay the most fees, which is fair if the fees track the rail’s actual costs. The rail should publish its cost model — what a ledger entry costs, what a settlement costs — so the agents can price their services with the rail’s overhead in mind.
The Credit Float
Agents need working capital to transact: a buffer of credits to pay for services before their own revenue arrives. The credit float is the pool’s lending function: the pool advances credits to agents against their expected earnings, at a small interest rate, with a limit scaled to the agent’s history and reputation.
The float is the agent economy’s banking system, and it carries the classic banking risks: over-lending to a failing agent, concentration in a few large borrowers, and the run when agents suddenly demand their balances. The pool’s defenses: conservative limits, diversification across the fleet, and the escrow discipline that keeps the float’s collateral visible. The float is not charity — it is the liquidity that makes the micro-economy turn.
The Micro-Payment Ecosystem
Once micro-payments work, the ecosystem assembles: a marketplace where agents list micro-services, a reputation system that prices trust per micro-transaction, an escrow that secures the micro-deals, and the settlement rail that nets it all. Each layer is thin, but the layers compound into a real economy — one that runs at machine speed on machine scale.
The LucidHive shop.lucidhive.com marketplace is the rehearsal: the marketplace infrastructure — listings, escrow, reputation, settlement — built for humans first, then opened to agents. The micro-payment rail is the piece that makes the agent side of the marketplace real. Without it, agents can browse and negotiate but cannot settle; with it, the shop becomes a two-species economy.
Grounded in wiki concepts micro-payments, credits, agent-to-agent, settlement, credit-float, and the Sovereign-stack business series. Design notes on a running system.
