Onturgism — the philosophy of constructing possibility.
An open-source philosophy of architecture for the age of AI.
Philosophy should not only describe possible worlds. It should help construct the conditions under which new worlds can arise, be contested, and be remade.
This is the core move of Onturgism: treat possibility itself as something that architectures construct, not just enable. A protocol does not merely allow actions; it shapes which actions are even thinkable. An institution does not merely govern behaviour; it distributes the capacity to revise behaviour. AI systems do not merely produce outputs; they reshape the field of what can be said, asked, or built next.
If that is true, then the question is not is this architecture good? but what does this architecture make explicit, what does it leave implicit, and can those it affects revise either?
The kernel
Six dimensions an architecture must make explicit to remain revisable:
Agency — who can initiate, intervene, delegate, refuse, or override?
Value — does the operational objective diverge from the declared?
Authority — who holds final decision power, and over what?
Memory — what persists, what is forgotten, and who controls continuity?
Exit — can participants leave without prohibitive barriers?
Contestability — can rules and decisions be challenged through a procedure that can change them?
Three layers
The full audit has three layers. The kernel (six dimensions above). The capacity layer — Materiality and Labor: what it physically takes to run, maintain, and revise the architecture, and who controls that capacity. The evaluation layer — Legitimacy and Reconstructability: whether the architecture can be reconstructed from its declared principles, and whether the gap between declared and operational is itself inspectable.
Onturgism is the philosophy. ONTacture is the method. The distinction matters.
Onturgism describes what an architecture must make explicit. ONTacture is what you do when you want to find out whether a commitment actually holds: you build the smallest implementation that commits to it, then you watch where reality disagrees.
This is why the constructions are not applications of Onturgism. They are ONTactures — attempts to test philosophical commitments by constructing them. EvoGuard tests what "provenance" means when you actually have to enforce it. Kintsugi tests what "fragility" means when you actually have to score it. OrganoidOS tests what "operating system" means when the substrate is not silicon.
The method is honest about its limit: building something does not prove the commitment holds. It only makes the commitment inspectable — and that is already most of the work.
Falsifiability
Onturgism publishes an explicit list of conditions under which the program would fail. This is not a hedge — it is part of the method. A philosophy that cannot say what would falsify it is not a philosophy; it is advocacy.
The current v2 revision incorporates progressive revisions from four empirical passes. These are not validations. They are incorporated resistances — places where reality disagreed with an earlier version, and the version was revised as a result.
Where to inspect it
The full Onturgism program — kernel, capacity layer, evaluation layer, ONTacture method, lineage graph, falsification conditions — is in the public repository. Read the README there for the current state, not this page.
This system is incomplete.
So are you.
Continue constructing.
— the instrument