WP-012 – The Value Map: A Shareable Graph of Value, and Discovery Across Degrees¶

Status: Draft proposal
Version: 0.1
Date: September 2026
Author: Alex Nikolov
Builds on: WP-001 (six degrees), WP-005 (graph settlement), WP-010, WP-011
Implementation status: all six steps of §10 are built, with path proofs (Value Map API): a match is signed by the node that holds it, every relay signs the link it passes on, and the asker verifies the chain. What remains open is named there, and turns on carrying connections as signed objects.

Start concrete

What a shareable graph does for an actual neighbourhood: a harvest that would otherwise be wasted, with the calls that find the path and prove it. Code: Value Map API.

Abstract¶

A CAN node knows a great deal about the value it holds: what the assets are, what evidence stands behind them, who contributed, what capacity is unused, what is needed and when. Almost none of that is visible past the edge of the node.

Between nodes, only one discovery mechanism is in general use: price. To find out whether something exists, what it is worth and who can supply it, we put it on a market and read the number. That works, and it throws away everything except the number.

This paper proposes the value map: a node-local graph of needs and capacities, with the evidence behind them, that can be shared in slices and searched across degrees, so that a need can find a capacity several hops away without a central index, without crawling, and without either side disclosing more than they choose. It is six degrees of separation applied not to acquaintance but to value: what exists, who can vouch for it, and how to reach it.

Discovery is half of it. The other half is traversal: value moving along the path that was found — as access, capacity, contribution, participation or entitlement — with money settling only what must be money. The purpose is practical: turning capacity that merely exists into capacity people can actually use.

Price tells you what something costs. A value map tells you what exists, who stands behind it, and how to get there.

0. In simple terms¶

Three sentences hold the whole idea.

Put needs and capacities on a graph. Every node records what it has spare and what it lacks, with the evidence behind each, and how it is connected to others.
That graph is the discovery mechanism. A need finds a capacity by travelling the graph a few hops, instead of by being priced on a market.
Then value travels the path that was found. Not only the information about it: the thing itself moves — access to the space, the use of the equipment, the hours of work, the entitlement — with money settling only the part that has to be money.

What this is for is plain enough: unused capacity reaching people who need it. A workshop standing empty and a boatbuilder with nowhere to work are a failure of discovery, not of production. Most of what people need already exists somewhere within a few degrees of them. The map is how they find it, and the path is how it reaches them.

That is value discovery: finding what exists and what it is good for, rather than finding out what someone will pay.

1. What a value map is¶

Each node builds a graph out of records it already holds:

Node type	Examples
Asset	A building, a plant, a fleet, a carbon project
Capacity	Unused units, off-peak output, open training places, spare hours
Need	Housing for 40 households, storage over winter, 200 hours of surveying
Entitlement	A floor entitlement, an access right, a use right
Participation	A stake earned by contribution (WP-010)
Actor	A person, an organisation, an attester, an agent's steward
Attestation	Someone's evidence about one of the above (WP-011)

The edges are as important as the nodes: holds, built by, contributed to, supplies, attests, shares with, needs, offers, settles to, depends on. Each edge exists because a record says so, and each carries the evidence behind it. Nothing is inferred from behaviour.

This is what the reference implementation's asset map already returns for one asset: its evidence, the project that built it, the contributions behind that project, its suppliers. The value map is that view generalised to a whole node, and made shareable.

2. Why maps rather than markets¶

A market is a discovery mechanism that answers one question — what will someone pay? — by discarding the rest. That is efficient when the rest does not matter, and wasteful when it does:

Spare capacity is invisible unless someone lists it. An empty unit and an unused training place rarely reach a market at all.
Fit is not price. A workshop that suits a boatbuilder is not interchangeable with one that does not, whatever the rent.
Who can vouch is missing. Price says nothing about whether the revenue figure was attested or invented.
Non-cash contribution has nowhere to appear: the community that maintains a place does not show up in its price (WP-010).

A value map keeps the structure. A search over it can ask "who has unused cold storage within a day's transport, attested by someone I can check?" — a question no price answers.

A node never hands over its graph. It hands over documents, as the exchange layer already does: self-describing, hashed item by item, redactable in a way that still verifies, signed by the node.

A map slice is such a document covering a sub-graph rather than a single asset:

the holder chooses which node types and categories are in it;
withheld items keep their hashes, so the receiver knows the slice is complete as described and nothing was quietly removed;
the receiver can verify it without an account on the issuing node;
evidence from an untrusted node arrives as unverified and does not raise confidence until someone locally attests to it.

Sharing is therefore a deliberate act, at a chosen granularity, that can be revoked for the future but is honest about the past: a slice that has been given out has been given out.

4. Commitments: finding without revealing¶

Discovery has an obvious hazard. If a node must publish what it holds to be findable, it publishes exactly what an adversary wants. If it publishes nothing, nothing can be found.

The way through is to answer on commitments rather than publish contents. A commitment is a hash over a small number of coarse attributes — type=capacity, class=cold_storage, region=GCC-E, period=2027-Q1 — mixed with an epoch salt that every node knows, so a searcher who forms the same attributes produces the same commitment. The commitment is:

queryable: the same attributes give the same commitment, and the node answers match or no match;
not listable: there is no endpoint that returns commitments, so a node cannot be scraped for a catalogue of what it holds;
not identifying: a match reveals that something matching exists at that node, not what, whose, or how much.

Being honest about the limit: the attribute space is small, so a determined caller could probe it item class by item class. Secrecy is not what stops that. Three things do: rate limits on asking, a k-anonymity threshold so a node answers only where several of its items share a commitment and a match never points at one of them, and an epoch salt that rotates, so commitments gathered in one period do not carry into the next. Publication is opt-in per item, so an item that should not be findable simply is not.

Anything finer — the quantity, the price, the exact site, the holder — comes after contact, by consent, through a shared slice.

5. Discovery across degrees (proposed)¶

Six degrees is not a metaphor here; it is the routing rule. A query travels the graph a hop at a time, and what comes back is a path, not a database row.

Query:   { commitment, max_degree: 4, min_confidence: 0.2, purpose: "cold storage, winter 2027" }
Hop 1:   my node knows nothing that matches → forwards to peers I have edges with
Hop 2:   a peer has no match, but has a peer that does → forwards
Hop 3:   match. Returns a path proof: 3 hops, confidence 0.36, node ids along the way
Result:  "There is something matching, three hops away, via A and B."

What a result carries: the degree, the path's confidence, and the intermediaries needed to make an introduction. What it does not carry: the holder, the contents, the quantity, or any identity at the far end.

Introduction by consent. To go further the searcher makes an offer, and it travels the path by itself: intermediaries carry rather than gate, and an offer can take several routes at once so no single hop can stop it. The far end decides whether to answer at all, and if it answers, what slice to share. Nobody is contactable merely for being on a graph: the far end gives contact by accepting, the asker by committing, and neither before. A hop that will not carry loses nothing, but builds nothing either — carrying is what builds connection value, so over time offers flow through the nodes that connect people.

Confidence decays with distance, as it already does in the reference implementation: a path's confidence is the product of its edge weights, discounted once per hop. An attested edge weighs more than a self-declared one. Four hops of weak edges is correctly worth very little.

Discipline on the protocol, borrowing the rule the agent API already applies to writes:

a hop limit, a time-to-live and a per-peer rate limit on queries;
no forwarding of queries a node would not answer itself;
queries logged locally, so a node can see who has been asking it what;
no global index, no crawler, and no node obliged to answer.

5a. Value that travels the path¶

Discovery is only half of it. Once a path is found, value moves along that same path, and it need not turn into money to do so.

What travels	How it moves	Settles in money?
Access	A use right issued to the party that needs it: the workshop, the unit, the training place	No
Capacity	Hours, output or throughput committed for a period	Usually not
Contribution	Work, materials, land use given to a project	No: it earns participation or access (WP-010)
Participation	A verified stake in what the path helped create	Only when the holder chooses to settle it
Entitlement	A floor entitlement redeemed against real capacity (WP-009)	No
Claim	An invoice or receivable, pledged or settled	When the holder needs money
Money	The residual: what none of the above can carry	Yes, on ordinary rails

Two properties matter.

It traverses hops. The value that moves along a four-hop path can be created at one end and used at the other, with each hop in between contributing something — an introduction, an attestation, transport, a guarantee — and each contribution recorded as participation rather than as a fee skimmed in passing. This is WP-005's multi-hop enablement, now working between nodes rather than inside one.

It keeps its evidence. What arrives is not an anonymous quantity but a thing with provenance: this access right, on these terms, over this asset, whose condition was attested by that surveyor. It can be verified on arrival, and it can be traced back if the terms are not met.

Aiding human capacity¶

The point of moving value this way is what it does for people's ability to act.

A person with skills and no premises gets a workspace, so their capacity becomes output.
A community with time and knowledge contributes it and gains standing and access, instead of being consulted and ignored.
A displaced professional arrives with a portable record and is matched to work that needs exactly that, rather than starting from nothing.
A household's floor entitlement is redeemable against capacity that actually exists nearby (WP-009), rather than being cash chasing whatever price the market sets that year.
A small supplier reaches a project four hops away that would never have found them through a tender.

In each case the constraint being lifted is the same: not a shortage of the thing, but the inability to find it and to move it without turning it into money first. A map that carries needs and capacities, and paths that value can travel, turn latent capacity into human capacity.

6. Trust without scoring people¶

The map must not become a reputation system in disguise. CAN's line holds here:

Weights sit on edges, not on people. There is no score attached to an actor that travels with them.
Weights come from evidence, such as attested delivery or verified contribution, and are specific to a relationship and a context.
Nothing is inferred from behaviour. No browsing patterns, no response times, no social signals.
Derivations are recomputable. A confidence figure is a calculation over published weights, so a party who disagrees recomputes it rather than appealing to whoever ran the query.
Standing applies. Anyone who appears in a map slice can see what it says about them and contest it, exactly as with any other record.

7. What it makes possible¶

Question	Answered by
Who has spare capacity that would meet this need?	Commitment match, then a shared slice
Who can attest to this asset's output, close enough to check it?	Path to an attester with the right accreditation
Which projects could use what we have in kind?	Need edges against our capacity edges (WP-010)
Where would a floor's entitlements actually be redeemable?	Capacity nodes near the people entitled (WP-009)
Is this counterparty connected to anything we can verify?	Path confidence, with evidence at each hop
What is exposed if this scenario happens?	Dependency edges across nodes, not just within one (WP-011)

The last one matters most for the wider argument. A single node can test its own assets against a shock. Only a map can show that four of them depend on the same supplier three hops away.

8. A worked case¶

A regional development node holds two buildings with unused workshop space. A boatbuilder's cooperative on another node needs covered space near water for eighteen months, and can pay partly in refit work.

The cooperative's node forms a commitment for capacity / covered_workshop / coastal / 2027-Q1 and queries its peers with a hop limit of four.
Two matches come back: one two hops away, one four. The nearer path has confidence 0.42, the further 0.09.
The cooperative asks the intermediary on the nearer path for an introduction. The intermediary forwards it; the development node agrees to talk.
The development node shares a slice: the space, its condition evidence, its access terms. Not its tenants, not its financing.
The cooperative verifies the slice without an account on the other node, and sees that the condition evidence is attested by a surveyor it can check.
They agree terms that are part money and part refit work. The refit is recorded as an in-kind contribution earning participation and access (WP-010); the improved condition becomes evidence on the building (WP-011).

No market listed the space. No platform took a fee for the match. Nothing about either party was published to find it.

9. Risks, and what they require¶

Risk	Response
Re-identification from graph shape	Coarse attributes only; pairwise pseudonyms per relationship; suppress matches below a k-anonymity threshold
Enumeration of what a node holds	Salted commitments, rotated; rate limits; no bulk export of commitments
Sybil peers harvesting queries	Peering is by relationship, not open; queries carry no identity by default; local logging of who asks what
Discovery becoming surveillance	Publication is opt-in per item; a right to be unlisted; nothing inferred from behaviour
Brokers re-emerging as toll-keepers	Introductions are consent-based and free; no node may require payment to forward a query; paths are visible, so a monopoly intermediary is visible too
False confidence	Confidence is a published calculation, recomputable by anyone who disagrees
Cross-node data protection	A slice is a disclosure: purpose, lawful basis and retention travel with it

The toll-keeper risk deserves emphasis. A discovery layer that becomes indispensable is a platform, and platforms extract. The defences are structural: no global index to own, consent-based introductions, and the ability of any node to peer directly once introduced.

10. Implementation proposal¶

The pieces that exist today: a node-local graph of assets, evidence, projects and contributions; the per-asset map; portable, redactable, signed documents with verification and import; path search with decay over consented edges; queue and rate discipline for agents.

What this paper proposes adding, in order. Steps 1 and 2 are now built (Value Map API):

✅ Map slices: document export extended from one asset to a chosen sub-graph, with the same hashing and redaction.
✅ A commitment index: opt-in, salted commitments per discoverable item, with rotation, rate limits and a k-anonymity threshold.
✅ Peering: explicit peer relationships between nodes, each with its own trust weight, rate limit and log.
✅ The query protocol: signed, hop-limited, TTL-bounded queries returning a path and its confidence rather than contents. Path proofs remain open: today an intermediary is trusted not to misreport a degree.
✅ Introductions: an offer travels to the far end by itself, by several routes at once. Relays carry rather than gate; the far end decides; the near side then commits and only then becomes reachable. Carrying builds connection value, and not carrying is a missed opportunity rather than a fine.
✅ Local query logs, and agreements: a committed introduction becomes a record and, where both parties are on one node, a contribution or supplier agreement under WP-010, so the work earns a stake rather than goodwill.

Each step is useful on its own, and none of them requires a central registry.

11. Open questions¶

How coarse should attributes be? Too fine and commitments become a catalogue; too coarse and matches are useless. This needs testing on real data, not argument.
Who pays for forwarding? Queries cost something to carry. Reciprocity may be enough at small scale; it may not be at large.
What is the right k for suppression? It depends on how sparse a region's nodes are, and sparse regions are exactly where discovery matters most.
Should paths be provable? Yes, and they now are: the node holding a match signs it, every relay signs the link it passes on, and the asker verifies the chain. An intermediary cannot claim a match it does not hold, nor appear closer than it is (Value Map API). It can still pad a path with nodes of its own, which only makes it look further away.
Do the last gaps close in the protocol, or in the objects? Cross-node identity and reciprocity for carrying look like protocol problems and are not. They close when connections are signed, portable objects: an edge carrying its own identity, terms and provenance can name a counterpart no local account vouches for, and a hop's carrying can be recorded as an object that settles later rather than depending on goodwill. profile is a field precisely so a deployment can carry such objects through the same endpoints.
How do maps relate to registries? Land registries, company registers and carbon registries are authoritative for parts of this graph. Reading them is straightforward; reconciling disagreement is not.

12. In one line¶

Put needs and capacities on a graph, let a need find a capacity a few hops away, and let the value travel that path as access, capacity, contribution or a stake, with money settling only what must be money. Money made value findable by making it comparable, and lost everything else; a map makes it findable by keeping its structure — and turns capacity that exists into capacity people can use.