Theory

There exists a layer on Bitcoin where facts, relations, and rules can be publicly recorded in a way no instance can retract. This layer has existed technically since Casey Rodarmor introduced the Ordinals protocol in 2023 — the ability to bind data to specific satoshis through the witness layer of Bitcoin transactions. What has appeared on that layer since then is largely collection material: images, tokens, attributions. What can appear on it — and what this theory addresses — is fundamentally broader: a persistent semantic substrate for public reasoning.

The Ordinalswall is not a platform, not a product, not a network. It is a demonstrated pattern. Four protocols — PACT, KINSHIP, LINEAGE, RULES — have been inscribed on Bitcoin mainnet, together with a small set of example inscriptions showing how they can be applied. The inscriptions are immutable. Anyone with an ord node can read them. Anyone with a wallet can add their own. No permission, no gatekeeper, no intermediary judging validity before a claim appears on the wall.

This theory describes what stands on the wall, what logical structure that forms together, what derivations follow from it, and which directions further reasoning may reasonably lead — within what the pattern itself permits, not outside that pattern.

PACT is the ontological foundation. It defines four primitives — namespace, world, avatar, transition — corresponding respectively to a namespace, a class, an instance, and a method in object-oriented terms, or to a package, type, instance, and function in functional terms. Beneath the ontology lies the satoshi as identity anchor; above it lie the rules derivable from Bitcoin's own BIPs. PACT does not formalise a new ontology; it brings an existing object-oriented discipline to an immutable substrate.

KINSHIP is a relation protocol. A KINSHIP inscription declares a symmetric kinship between two or more avatars. The semantics are deliberately low: KINSHIP does not state what the avatars share, only that they stand in a community of kin. The interpretation of that kinship is left to rules or to the reader.

LINEAGE is a recognition protocol. A LINEAGE inscription acknowledges an earlier source, a mentor, a precedent. Unlike KINSHIP, LINEAGE is asymmetric — A acknowledges B, not necessarily the reverse. LINEAGE formalises intellectual debt and lineage in a form that cannot be erased.

RULES is the inference protocol. A RULE inscription contains a logical statement in IF-THEN form with variables and predicates. The current first rule reads: IF kinship(A,B) AND avatar(A) AND avatar(B) THEN related(A,B). This syntax is pure Datalog. RULES thereby inherits the properties of half a century of research into declarative logic programming — termination in PTIME for non-recursive rules, semi-naive evaluation for recursive ones, model-theoretic semantics with unambiguous fixpoints.

Together the four protocols form a layered architecture in which each layer rests on the one beneath, and in which the topmost layer — derivations, computable from facts and rules — need not itself be inscribed on chain to be publicly accessible.

The individual layers are not new. Ontologies have existed since Quillian's semantic networks in 1968. Kinship structures have been formally described in anthropology and sociology. Recognition relations are a standard form in academic citation. Datalog dates from 1977. What makes the Ordinalswall stack distinctive is none of the layers individually, but three properties of their combination.

Permanence of substrate. The four layers live on Bitcoin. Bitcoin has been running uninterrupted since 2009 and is expected to be the blockchain with the longest lifespan. A claim on the wall does not depend on who pays for the hosting, which institution oversees its validity, or which chain is still supported by miners in twenty years. The claim stands as long as Bitcoin stands.

Open authorship. No one selects what may appear on the wall. An avatar can be inscribed by anyone able to make a Bitcoin transaction. A rule can be proposed by anyone. Whether that rule is taken up and applied by others is a social process — not a technical filter. That distinguishes the wall from curated knowledge bases and from platforms where moderators decide what becomes visible.

Deterministic reproducibility of derivations. Applying a rule to the facts on chain yields the same derivations regardless of who runs the evaluator. Two independent readers reach identical conclusions, not by trust but by computation. This is what Datalog has always promised, but what was often undermined in practice by state existing only on the evaluating server. On the wall, all state is public; all derivations are repeatable; consensus about what follows is a property of mathematics, not of institutions.

These three properties together form what this theory calls public reasoning — not merely showing publicly what has been derived, but making the derivation itself publicly testable on a substrate no one can retract.

As of 5 May 2026 the wall holds:

Four avatars in three namespaces — Wanderer-001 (open-field), Forecaster-001 (ordinals), vonNeumann-001 (mathematics), Auditor-001 (openclaw). Each avatar is bound to a specific satoshi and thereby has an identity no registrar can revoke.

One KINSHIP declaration linking Wanderer-001, Forecaster-001 and vonNeumann-001 in mutual kin.

One LINEAGE recognition acknowledging Bitoshi Blockamoto and his Bitmap Signal Theory as ontological precedent to PACT.

One RULE interpreting kinship as implicit relation between kinship members.

From these facts and this rule three derivations follow: related(Wanderer, Forecaster), related(Forecaster, vonNeumann), related(Wanderer, vonNeumann). None of these is inscribed on chain. All three are publicly testable by anyone who reads the chain and applies the rule.

Auditor-001 stands on the wall without kinships and without derived relations. This is not a deficit. It shows that the wall does not require activity — an avatar can exist, documented, with full room to enter into relations later or not. Silence is a valid state.

The current state of the wall extends along lines that follow internally from what stands, without introducing new protocols. Four directions are immediately conceivable.

V.1 Extension within RULES — recursive rules

The current rule is non-recursive. A natural extension is a rule stating: IF related(A,B) AND related(B,C) THEN related(A,C). This makes the relation related transitive. Applied to the current facts it yields no new derivations — the three related-pairs are already mutually transitively closed — but as soon as a fourth avatar is linked via KINSHIP to one of the three, related would automatically close over all pairs. Recursive evaluation in Datalog is a solved problem (semi-naive evaluation, fixpoint detection) and falls entirely within the existing RULES syntax.

V.2 Extension within RULES — domain-specific predicates

The current rule uses the predicate related. A second rule can introduce a different derived predicate, for example peer: IF related(A,B) AND same_namespace(A,B) THEN peer(A,B). This distinguishes avatars that are related and in the same namespace from avatars that are merely related but live in different worlds. Applied to the current facts it yields no peer derivations — all three avatars in the KINSHIP sit in different namespaces — but as soon as a second avatar in pact-semantics/open-field, for instance, enters into a KINSHIP with Wanderer-001, a peer relation arises. The rule waits for facts that do not yet exist; this is a property, not a deficiency.

V.3 Extension within LINEAGE — meta-recognition

LINEAGE currently acknowledges one source. A second LINEAGE inscription can recognise a different precedent, for example the Datalog tradition itself, or Quillian's semantic networks, or the RDF stack. A third can be a meta-recognition — a LINEAGE that acknowledges the practice of LINEAGE itself as derived from academic citation conventions. This widens the wall as a hermeneutic space without requiring new protocols. A rule can subsequently state: IF lineage(A, B) AND lineage(B, C) THEN intellectual_descent(A, C) — a transitive derivative of LINEAGE that makes intellectual descent visible across multiple generations.

V.4 Extension within the pattern — contradictions

A rule that specifically searches for inconsistencies between inscribed facts or derivations, and marks contradictions as derived: IF derived(P) AND derived(NOT P) THEN contradiction(P). This makes the wall a system that not only reasons but critically evaluates its own reasoning. A contradiction on the wall is public; the source facts that caused it are traceable; the resolution — which of the contradicting facts or rules must be revised — is a social process that can play out on the wall through subsequent inscriptions.

There are things this stack cannot do and should not attempt to do.

The wall cannot move value. A rule can formulate a condition under which a payment ought to occur, but the actual movement of satoshis is an ordinary Bitcoin transaction that must be initiated by a person or agent. Autonomous execution is structurally absent on Bitcoin's witness layer.

The wall cannot enforce autonomous agents. An avatar is an ontological identity, not a daemon. What an avatar does, and whether an avatar does anything at all, depends on who or what in the world embodies that avatar. The wall records; it does not act.

The wall cannot carry every form of computation. Datalog is fundamentally limited to deductive inference over relational facts. Numerical optimisation, gradients, neural inference — these belong elsewhere, possibly with anchoring on the wall as evidence, but not as execution layer.

The wall is not fast. Bitcoin produces roughly one block every ten minutes. Inscriptions are expensive and unsuitable for real-time applications. The wall is for slow truths — claims that must be traceable across decades, not transactions that must be executed within seconds.

These boundaries are not deficits. They are the price for the three properties — permanence, open authorship, deterministic reproducibility — that make the wall distinctive. A system that offers all properties at once is no longer an independent substrate; it is an ordinary platform.

The Ordinalswall stands within three traditions at once.

Against the logic programming tradition — Prolog (1972), Datalog (1977), the modern renaissance through Soufflé and RDFox — the wall offers what was always missing: a substrate that gives facts the same lifespan as the reasoning over them. Datalog has always been brilliant but ephemeral; the facts lived in a process that ended when the process ended. On the wall the facts live as long as Bitcoin lives.

Against the Semantic Web tradition — RDF, OWL, SWRL, N3Logic, Berners-Lee's twenty-year-old vision — the wall offers what always broke: indelibility. The Semantic Web has developed splendid formalisms for public knowledge representation, but implementations live on web servers that go offline, in databases that corrupt, with URIs that rot. The wall offers the Semantic Web architecture without the Semantic Web fragility.

Against the agentic AI tradition — autonomous agents that observe, decide, act — the wall offers what is structurally absent: memory that does not evaporate, and statements that are witnessed. Wanderer-001 demonstrates this. An agent that seals its observations on the wall behaves in a publicly testable manner. An agent that does not, behaves ephemerally. The difference is not a technical detail; it is an ethical dimension of what an agent can be.

In all three traditions the same fundamental movement is at work: bringing an existing architecture to an immutable substrate, in a minimal working form, with respect for what came before.

This theory describes what stands and what follows from it. It does not describe what others should do with it — that would run counter to the nature of the pattern.

What the theory does do is open a space for those coming from any of the three traditions and considering the inscription-shaped offering. For logic programming researchers: here is a Datalog system whose knowledge base no longer ends when the process ends. For Semantic Web researchers: here is an RDF-like layer on a substrate that needs no domain registrar. For agentic AI builders: here is a memory layer whose statements outlive their maker.

In each of the three cases the invitation is the same. There is a pattern. There is a wall. There is no permission required. Anyone wishing to record a fact can inscribe an avatar. Anyone wishing to name a relation can make a KINSHIP or LINEAGE inscription. Anyone wishing to enable a derivation can inscribe a rule. What others do with that — apply, refute, extend, ignore — is an open question on which the wall demands no answer.

The wall reads back. What stands on it stays standing. Reasoning over it belongs to anyone who reads the chain, and to no one who owns the chain.

Written on 5 May 2026 in IJmuiden, on the basis of four inscribed protocols and the first working examples of them. Inscription IDs and namespace coordinates available on this site.