How It Works
From surplus to exchange in four steps
Describe what you have. Describe what you need. Let the algorithm find the connections.
Overview
The Surplus Exchange Protocol has four core components:
Here's how each component works.
Subjective Value
There's no shared currency or exchange rate in this system. Each participant values what they give and receive on their own terms — your ledger is yours, theirs is theirs. The system doesn't need to know the "true" value of anything; it just needs to find chains where everyone ends up better off than they started. See the Philosophy for why this matters.
Describe
Tell the system what you have and what you need, in your own words. AI translates your descriptions into matchable terms.
What you describe
Every participant describes two things:
- What they have: surplus capacity, spare time, excess inventory
- What they need: services, goods, help they'd value
These descriptions form a network. Each "has" is a potential connection to someone's "need."
How the system understands you
AI interprets natural language — no jargon, no categories, no forms to fill. "Pitch deck design" and "presentation help" are understood as related capabilities.
This is what makes cross-industry matching possible. The system bridges vocabulary differences between industries and professions. A vocabulary of terms grows from actual usage, getting more efficient over time.
Why this matters
Previous exchange systems required manual categorisation — someone had to decide that a law firm's contract review and a restaurant's catering were comparable units of exchange.
AI handles this interpretation, which is why multi-party matching across industries is feasible now when it wasn't before.
Matching
The algorithm searches the network for closed loops — chains where everyone gives something and receives something.
How the algorithm finds chains
The descriptions from the Describe step form a network. Each surplus is a potential connection to someone's need.
The matching algorithm searches this network for cycles, closed loops where:
- A helps B
- B helps C
- C helps D
- D helps A
When a cycle is found, everyone in the chain can give something and receive something. The loop closes without anyone losing out.
What makes a good chain
The algorithm ranks chains by:
- Match quality: How well do offerings fit needs?
- Trust levels: Have these participants exchanged before?
- Chain length: Shorter chains are easier to coordinate
- Timing alignment: Can deliveries happen in a workable sequence?
- Surplus urgency: Time-sensitive surplus is prioritised — a free Tuesday afternoon this week matters more than one three months from now
- Relationship diversity: New partner connections are preferred — the algorithm avoids routing too many exchanges through the same participants
Top-ranked chains are proposed to participants for confirmation. The algorithm also ensures no single participant becomes a bottleneck — the more chains someone is already in, the less their inclusion improves a new chain's score.
What the algorithm can find that humans can't
A network of 100 businesses might have thousands of potential connections. A network of 1,000 has millions.
No human broker could evaluate all possibilities. The algorithm checks them in seconds.
More importantly, it finds connections across industries and contexts that no one would think to look for. The theatre and the accountant have no obvious relationship. But the algorithm spots that both connect to a designer who connects to a cafe.
You can always see your own scores and understand why you were or weren't matched for a particular chain. The algorithm's decisions are transparent to participants — though you only see information about your own matching, not other participants' profiles.
Trust
Trust is built gradually through completed exchanges and verifiable track records, not ratings or reputation scores.
The cold-start problem
New participants have no track record. How do you trust them?
The answer: you don't. Not at first. Trust is built gradually.
Trust tiers
- Bilateral exchanges only
- Single concurrent exchange
- Track record begins
- Chain participation
- Multiple concurrent exchanges
- Track record accumulation
- Full exchange participation
- Can vouch for new members
- Visible history to partners
- Network stabilising role
- Large, complex chains
- Extensive verified history
How trust is earned
Every completed exchange generates signals:
- Did both parties deliver?
- Were they satisfied?
- Was timing as promised?
These signals accumulate into a track record. Good track records open access to larger exchanges and more complex chains. Trust reflects ongoing activity — a business that stops participating gradually loses network position. This prevents early participants from accumulating permanent advantages.
Vouching
The default entry path is the Newcomer tier. You verify your identity and start with small bilateral exchanges. No introduction needed. Everyone can participate.
Vouching is an accelerator. An established member can vouch for you, letting you skip the Newcomer tier and start at Probationary with access to chains:
"I know this business. I'm willing to stake some of my reputation on them."
This is about speed, not access. Vouching lets you move faster, but you don't need it to get started. If the newcomer performs well, both gain. If they don't, the voucher's reputation takes a small hit too.
Exchange
Every exchange follows a clear lifecycle from proposal through satisfaction, with human confirmation at every step.
The lifecycle of an exchange chain
Proposal: The algorithm finds a viable chain and proposes it to all participants. Each participant sees what they would provide, what they would receive, who else is in the chain, and proposed timing.
Confirmation: Every participant must confirm before anything happens. Review the proposal, check track records, then confirm, decline, or make a counter-proposal — suggesting different timing, modified scope, or adjusted terms. Counter-proposals circulate to the other chain participants, and the chain can be renegotiated until everyone agrees or the algorithm looks for alternatives.
Execution: Once confirmed, participants deliver what they promised. Delivery happens in whatever order the chain requires, often in parallel.
Satisfaction: After delivery, both sides signal whether they're satisfied (fully, partially, or not at all). These signals feed back into trust calculations.
What happens if something goes wrong
Exchanges can fail. A participant might not deliver, or delivery might not meet expectations.
When this happens:
- Any participant can flag an exchange as stuck — signalling that something isn't working as agreed
- The first step is always a conversation between the participants involved
- If the conversation resolves things, both sides signal the outcome and move on
- If it doesn't, either participant can escalate to governance for review
- Persistent patterns of non-delivery affect trust and future matching opportunities
The system tracks commitment delivery, not balance. The question is "did they do what they said?" not "did they give enough?" You choose what to commit to, and trust reflects whether you followed through.
Summary
| Component | What it does |
|---|---|
| Describe | Tell the system what you have and need, in your own words |
| Matching | Finds chains of exchange across the network |
| Trust | Builds confidence through track record and vouching |
| Exchange | Coordinates confirmation, delivery, and satisfaction |
| Foundation | Subjective value — each participant tracks their own sense of balance (no shared currency) |
Continue exploring