ONTOLOGY PROTOCOLS
09. Protocol for the Generative Architecture of Concepts
Revision 1.7 · consolidated version
1. Purpose
The Protocol describes the minimal generative construction S–O–M–W: Self, Other, Between, Whole. It establishes the conditions under which a candidate W may be accepted as a new independent semantic determination and separates correctness of generative origin from truth, knowledge, existence of a referent, and canonization in CCO.
(S,O,M) → W_candidate
This notation does not denote a deterministic function. One configuration may have no W, one W, or several candidates.
2. Four Functional Positions
S — Self: the initial independent determination of the current generative act.
O — Other: a determination distinct from S and relevant to the declared context.
M — Between: substantive mediation of the joint consideration of S and O.
W — Whole: a candidate for a new independent determination not reducible to a simple repetition of S, O, or M.
A functional position is not identical to a permanent entity type. One Semantic Unit may occupy different functional positions in different acts.
3. Combinatorial Structure of One Act
For four functional positions there are six unordered pairs: SO, SM, SW, OM, OW, MW. Under the adjacency rule “two pairs are adjacent if and only if they have exactly one common source element,” they form the line graph L(K4).
6 vertices; degree of each vertex 4; 12 edges; 8 triangular cycles.
SO↔MW; SM↔OW; SW↔OM.
The combinatorics 4–6–4–1 describes four positions, six pairs, four triples of source positions, and one whole. The abbreviation 4–6–8 refers to four positions, six pairwise mediations, and eight triangular configurations of L(K4).
Identification of this combinatorics with the complete geometry of the stella octangula is not a normative consequence of the Protocol and remains a separate hypothesis.
4. Conditional Minimality of the Quaternity
The Protocol does not assert that every recursion or every act of thinking requires four elements. Quaternity applies to a generative act in which two source determinations, their mediation, and a new independent result are functionally distinguished.
If these four functions are genuinely independent in the model under consideration, reducing them to three positions requires demonstrating that one of the functions is not independent.
5. Versioned Context
Cᵛ := (Q, Constraints, ContextVersion)
Q fixes the research question; Constraints fix the declared restrictions. The context must be defined before the relevance of O is assessed. A change in the essential content of C creates a new version.
6. Selection of the Other O
O need not be generated from S. It is selected from the available semantic space and must be both distinct and relevant.
O*(S,Cᵛ) := {Oᵢ ∈ U_CCO | Oᵢ ≠ S ∧ Distinguishable(S,Oᵢ) ∧ Relevant(Oᵢ|S,Cᵛ)}
A common Domain is neither a necessary nor a sufficient condition of relevance.
Relevant means a substantive contribution of O to consideration of Q. In research and machine implementations, such a contribution may be recorded through Impact_Q; however, this Protocol does not establish a universal Impact_Q algorithm.
A noisy textual change, the need to refute an accidental concept, or a mere increase in the number of associations do not by themselves constitute relevance.
7. Admissibility of Mediation M
PreAcceptM(M|S,O,Cᵛ) := Joint ∧ Constraining ∧ Nontrivial
Joint: the declared content of M depends on joint consideration of S and O.
Constraining: M establishes a substantive condition of joint consideration that excludes at least one alternative interpretation of the pair and preserves at least one.
Nontrivial: M is not a simple enumeration of independent characteristics of S and O.
If a space of interpretations is formally defined for a particular model, Constraining may have a formal implementation. Without such a model, mathematical notation does not replace substantive justification. A universal computable test of Nontrivial remains OPEN.
8. Proposal of W and Generation Mode
ProposedΓ_G(Sᵛ,Oᵛ,Mᵛ,Wᵛ)
G ∈ {FormalDerivation, ModelDerivation, InterpretiveConstruction, OtherDeclaredGeneration}
G records the mode by which W is proposed as a result of the S–O–M configuration. ProposedΓ_G records the declared generative origin of W, but does not by itself prove the truth of W, the existence of its referent, or the sufficiency of grounds for the corresponding statements.
Generation mode and support basis are distinguished:
GenerationMode ≠ SupportBasis
SupportBasis applies not to the semantic identity of W as such, but to the statements through which the origin, properties, reference, or other status of W are justified.
Depending on the type of statement, SupportBasis may include FormalProof, EmpiricalEvidence, SourceStatement, ModelSupport, InterpretiveArgument, or another explicitly declared basis. The presence of SupportBasis does not automatically raise the epistemic status of a statement: status is established under Protocol 03.
Empirical support may support a statement about W or its origin, but is not in itself a mode of generating the semantic identity of W.
9. Identity of W
If W already exists canonically in CCO, its canonical identity is used:
CanonicalExisting(W) ⇒ IdentitySpec(W)=CanonicalIdentity(W)
The author of a new generative act may not narrow an existing CanonicalIdentity to fit the available trace.
Technical or research presence of W in the system does not imply its canonicity:
ExistingRecord(W) ⇏ CanonicalExisting(W)
If W is represented in CCO but not canonized, its explicitly declared current identity is used while preserving its noncanonical status:
NonCanonicalExisting(W) ⇒ IdentitySpec(W)=CurrentDeclaredIdentity(W)
For a new W, ProvisionalIdentity is recorded together with the proposal. Before origin is verified, it must pass IdentityAdequate.
IdentityAdequate(Wᵛ,Cᵛ) ⇒ BoundarySpecified(Wᵛ,Cᵛ)
BoundarySpecified requires explicit boundaries sufficient to distinguish W from non-W and from relevant nearest alternatives within the declared context. This is a preliminary specification requirement, not the final Bounded test.
10. Essential Content
Ess_Cᵛ(Wᵛ) := content necessary to preserve the recorded IdentitySpec(Wᵛ) in context Cᵛ
For canonically existing W, Ess is consistent with CanonicalIdentity. For a noncanonically represented W, Ess is consistent with CurrentDeclaredIdentity and does not thereby acquire canonical status. For a new W, Ess is derived from ProvisionalIdentity that has already passed IdentityAdequate and BoundarySpecified.
Essential content is not reselected to fit the available trace.
11. Traceability of Origin
GroundedΓ_G(Sᵛ,Oᵛ,Mᵛ,Wᵛ,Cᵛ) :⇔ ∀e∈Ess_Cᵛ(Wᵛ), ∃Trace_G(e | Sᵛ,Oᵛ,Mᵛ,Cᵛ)
Trace_G shows the origin of essential content e from S, O, M, or from an explicitly admissible transformation belonging to the declared GenerationMode G.
New content need not be literally present in the inputs. A grounded path of its derivation is required.
Novelty is admissible; an unsupported semantic leap is not.
FormalDerivation requires formal derivation; ModelDerivation requires derivation within the declared model; InterpretiveConstruction permits an argued semantic trace but does not automatically acquire the status of formal proof or empirical confirmation.
If the trace is supported by separate proofs or data, they are recorded as SupportBasis of the corresponding Statements, not as a replacement for GenerationMode.
12. Acceptance of a New Whole
AcceptAsNewWhole(Wᵛ) := Irred(Wᵛ) ∧ Bounded(Wᵛ) ∧ Operable(Wᵛ)
Irred: W is not reducible to a simple repetition of S, O, or M.
Bounded: W has its own stable boundary of meaning or applicability sufficient to distinguish W and non-W.
Operable: W is capable of independently participating in subsequent generative acts.
BoundarySpecified and Bounded are procedurally related but are not identical and are not defined through one another.
Acceptance of W as a new Whole means only that it passes the internal criterion of generative independence:
AcceptAsNewWhole(W) ⇏ Canonical(W)
AcceptAsNewWhole does not automatically create a canonical Concept ID, include W in the canonical CCO layer, or establish truth or existence of a referent.
13. Versioned Commitment
Each verification applies to specific versions of its inputs. Iterative research is permitted: C, O, M, W, or IdentitySpec may be refined. However, an essential change creates a new version and invalidates dependent results of the previous version.
V₁ → V₂ ⇒ Validation(V₁) ⇏ Validation(V₂)
The change delta and the reason for the new version must be preserved in the history of the generative passport. A retroactive change of an input does not transfer old justification to the new configuration.
14. Valid Generative Act
ValidGenerativeAct_G(Sᵛ,Oᵛ,Mᵛ,Wᵛ,Cᵛ) := Relevant ∧ PreAcceptM ∧ ProposedΓ_G ∧ IdentityAdequate ∧ GroundedΓ_G ∧ AcceptAsNewWhole ∧ VersionConsistency
VersionConsistency means that all results used refer to compatible versions of the inputs.
A valid generative act creates a valid generative candidate, not a canonical unit:
ValidGenerativeAct(W) ⇒ AcceptedGenerativeCandidate(W)
AcceptedGenerativeCandidate(W) ⇏ Canonical(W)
15. Transition from Generative Candidate to Canonization
The question of including W in the canonical CCO layer is governed by Protocol 08 and is not decided by Protocol 09.
Normative sequence:
ValidGenerativeAct → AcceptedGenerativeCandidate → Protocol 08 verification → Canonization or Rejection or Open Identity/Status
Canonization requires applicable Protocol 08 procedures: verification of semantic continuity or novelty, impact analysis, proposal status, dependencies, and other necessary checks.
Proposed ≠ AcceptedGenerativeCandidate ≠ Canonical ≠ True
Generative validity is an input to the CCO development procedure, not its replacement.
16. Transfer of Functional Role
After W is accepted as a new Whole, it may act as the Self of the next generative act:
Roleₙ(Wₙ)=W ⇒ Roleₙ₊₁(Wₙ)=S
This is a functional transfer of role, not a complete law of recursion. It does not automatically determine Oₙ₊₁, Mₙ₊₁, or Wₙ₊₁.
Participation of an accepted but not yet canonized W in a subsequent research act is admissible only with explicit preservation of its noncanonical status and version. It does not turn W into a canonical unit.
17. Semantic, Epistemic, and Ontological Boundaries
Validity of a generative act means correctness of generative origin and representation of a semantic unit within that act. It does not establish truth, knowledge, existence of an external referent, or canonicity.
Semantic validity ≠ Epistemic validity ≠ Ontological actuality ≠ Canonical status
A Concept/Semantic Unit does not automatically receive T/F or K_T/K_F/U. These statuses apply to corresponding Statements under Protocol 02 and Protocol 03. The presence of a Referential Anchor likewise does not prove the existence of a referent.
Absence of a conclusion about existence is not a negative ontological status.
18. Branching and Global Structure
For one S, several O are admissible; for one pair S,O, several M; for one configuration, several W candidates. One Semantic Unit may participate in multiple acts.
Therefore, generative history need not be a tree. This Protocol does not canonize it as a hypergraph, hierarchical graph, or another specific mathematical object before separate definition of the types of nodes, acts, versions, and relations.
19. Machine Representation and Architecture 1.7
The Protocol does not introduce a sixth fundamental identifier space and does not declare Mediation, Generative Act, GenerationMode, or SupportBasis to be a new fundamental entity type.
Until insufficiency of Architecture 1.7 is separately demonstrated, the generative passport must be represented by means of the existing architecture: Semantic Unit, Statement, Semantic Structure, Relation Type, Technical Projection, and version/provenance metadata.
GenerationMode describes the generative path. SupportBasis applies to justification of the corresponding Statements. Their technical storage must not erase this distinction.
If such expressiveness proves insufficient, an architectural change is admissible only through a separate Architecture Change Proposal (ACP).
20. Spatial Model Projection of the Generative Structure
Spatial model projection is a subsequent technical and research representation of an already derived semantic and combinatorial structure. It is not a source of semantic identity, does not create a new Semantic Unit, and does not form a new fundamental CCO identity space.
semantics → combinatorics → geometry → Spatial Projection
Projection is admissible only after the essential semantic positions, relations, boundaries, and combinatorial structure of the object under consideration have been obtained and fixed independently of the presumed geometry. A geometric model may use these results as input but may not retroactively determine their content.
Geometry ⇏ SemanticStructure
The presence of a vertex, edge, face, cell, free position, symmetry, or another geometric possibility is not sufficient grounds for introducing a concept, relation, functional position, or new Whole. A missing geometric element must not be filled with a semantic unit merely to complete the figure.
21. Spatial Object and Projection Type
Generalized projection form:
SemanticStructure ─Projection_G→ SpatialObject
where G is an explicitly specified and versioned geometric model. SpatialObject need not be a point. Depending on the independently established structure, it may be represented as Point, Edge, Face, Cell, Region, Path, or another formally defined type of spatial object.
The SpatialObject type is not assigned in advance. It must be compatible with the semantic role and combinatorial structure of the projected content. Coordinates (x,y,z) are a special case of spatial addressing of a Point and are not a mandatory property of every Semantic Unit.
22. Competing Geometric Models
The same semantic structure may have several parallel projections in different geometric models:
W → {P_G1, P_G2, …, P_Gn}
Differences among such projections do not imply differences in the semantic identity of W. Geometric models are treated as competing methods of spatial representation of one independently established semantic structure.
A model is assessed by its ability to preserve established semantic distinctions, relations, and combinatorial invariants with a minimum of additional assumptions.
No universal numerical metric for such comparison is established by this revision.
23. Separation of Validities
The Protocol requires distinguishing at least three independent checks:
SemanticValidation ≠ ProjectionValidation ≠ GeometryModelValidation
SemanticValidation concerns correctness of the semantic and generative result under this Protocol and applicable CCO protocols. ProjectionValidation means correctness of a particular projection within an explicitly specified version of a geometric model. GeometryModelValidation concerns verification of the geometric model itself and its ability to represent multiple independently obtained structures.
Successful ProjectionValidation does not prove the truth or ontological necessity of GeometryModel. Successful GeometryModelValidation does not establish the truth, canonicity, or existence of a referent of the projected Semantic Unit. Spatial proximity by itself does not prove semantic proximity, relation, or causal connection.
24. OPEN — Non-Canonized Computational and Spatial Questions
The following remain open: a universal computable criterion for Impact_Q/Relevant; a universal computable test of Nontrivial for M; a general machine model of IdentityAdequate, BoundarySpecified, and search for nearest alternatives; formal semantics of Trace_G for InterpretiveConstruction; exact version granularity and a matrix of dependent invalidation; a machine schema of the generative passport within Architecture 1.7; the mathematical class of the complete network of generative acts; the research direction Δ and its relation to synthesis; a formal criterion for selecting the SpatialObject type; a universal criterion for preservation of semantic topology under projection; a metric for comparing competing Geometry Models; rules for representing ambiguous or distributed projection; and conditions under which a spatial model is considered insufficient and must be expanded or rejected.
25. Normative Boundary
Protocol 09 establishes the rules of generative validity and admissible subsequent spatial model projection. It is not a universal algorithm for creating new meaning, a procedure of automatic canonization, or proof of the ontological necessity of the S–O–M–W model.
The S–O–M–W quaternity is a conditional minimal architecture for those generative acts in which two source determinations, their mediation, and a new independent result are independently distinguished.
Revision 1.7 does not canonize a particular geometry of the Concept Crystal and does not turn the stella octangula or any other geometric construction into a mandatory consequence of S–O–M–W.
Geometric correspondences do not become true or mandatory by virtue of acceptance of this Protocol.
The architecture is tested by concepts, not concepts by the architecture.
Consequently, semantic and combinatorial results are input to the testing of geometric models. Geometric success may support the fitness of a model as a representation, but cannot retroactively serve as grounds for forming semantic content.
This revision is consistent with Protocol 07 rev.1.4: Spatial Projection is a technical model projection and does not change the five-space CCO identity architecture; and with Protocol 08 rev.1.3: a change of Spatial Projection is not by itself a change of Semantic Unit.
← 08. Ontology Development and Verification Protocol · Ontology Protocols