FPF Reference

Ontological Parsimony (C-5)

Pattern A.11 · Stable Part A - Kernel Architecture Cluster

“Add only what you cannot subtract.”

The FPF kernel aspires to remain small enough to learn in a week yet broad enough to model engines, proofs and budgets alike. Unchecked growth of primitives—well‑known from earlier “enterprise ontologies”—bloats diagrams, stalls tooling and intimidates new adopters. C‑5 therefore demands minimal‑sufficiency: a new core concept enters the kernel only when all routes of composition, refinement or role‑projection fail to express it without semantic loss.

Keywords

  • minimalism
  • simplicity
  • Occam's razor
  • essential concepts.

Content

Context

The FPF kernel aspires to remain small enough to learn in a week yet broad enough to model engines, proofs and budgets alike. Unchecked growth of primitives—well‑known from earlier “enterprise ontologies”—bloats diagrams, stalls tooling and intimidates new adopters. C‑5 therefore demands minimal‑sufficiency: a new core concept enters the kernel only when all routes of composition, refinement or role‑projection fail to express it without semantic loss.

Problem

PathologyReal‑world symptom
Concept creepNear‑synonyms proliferate (U.Worker, U.Employee, U.Staff), breaking queries.
Zombie typesLegacy primitives linger unused yet block name space.
Tool churnEvery fresh primitive forces IDE, validator and dashboard updates.

Result: steep learning curves, fragile integrations, eroded trust in “first‑principles” promises.

Forces

ForceTension
Expressiveness vs SimplicityFine granularity helps static checks ↔ fewer nouns aid cognition.
Inclusivity vs PurityNew domains want vocabulary ↔ kernel must not be a dumping ground.
Evolution vs StabilityFramework grows ↔ users depend on a stable core.
Prestige vs UtilityAuthors enjoy naming things ↔ every name tcharacteristics everyone else.

Solution — Four‑Gate Minimal‑Sufficiency Protocol

A proposal to add a U.Type or core relation MUST clear all four gates before admission and survives under a Sunset Timer thereafter.

GateTest questionRationale
G‑1 CompositionCan existing primitives + roles/attributes express the concept without material loss?Follows “composition over creation.”
G‑2 Non‑RedundancyDoes the proposal overlap ≥ 80 % with anything already live?Blocks synonyms.
G‑3 Functional NamingDoes the chosen name state what the thing does, not what it is made of?Prevents vague catch‑alls; supports didactic clarity.
G‑4 Sharp BoundaryIs there a one‑sentence litmus test that unambiguously includes or excludes any candidate instance?Ensures crisp taxonomy edges.

Lifecycle — Sunset Timer A cleared type enters the kernel provisionally with a timer (default = 4 quarters). If usage count remains zero at expiry, the type faces Sunset Review: delete, demote to Extention Pattern, or renew with fresh evidence.

Manager’s mnemonic: “Compose, Unique, Functional, Crisp — or sunset.”

Archetypal Grounding

GateRejected candidate (why)Accepted approach
G‑1U.CoolantPump – expressible as U.System:Pump + CoolingCirculatorRole.Composition via Role.
G‑2U.Actuator vs existing U.Transformer (90 % overlap).Retain broader U.Transformer.
G‑3U.MiscellaneousObject – name signals no function.Reject; unclear purpose.
G‑4U.SmallPart – boundary depends on subjective size.Reject; fails crisp test.
—U.ProvenanceChain – required to record immutable evidence lineage; cannot be composed; functionally named; crisp membership rule (“ordered list of Evidence Graph Ref with forward integrity hash”).Accepted, timer started.

Conformance Checklist

IDRequirementDidactic aim
CC‑OP 1A Minimal‑Sufficiency Form (≤ 1 page) MUST accompany every new kernel‑type proposal, documenting answers to Gates G‑1…G‑4 and a draft Sunset‑Timer.Forces authors to think compositionally before adding nouns.
CC‑OP 2Kernel inventory tooling SHALL stamp each admitted type with sunset_due: <date> (default = +4 quarters).Schedules later pruning; no forgotten zombies.
CC‑OP 3A quarterly Usage Scan MUST flag any core type with reference‑count = 0; flagged items enter Sunset Review automatically.Turns parsimony into a living maintenance loop.
CC‑OP 4Renaming, aliasing, or splitting an existing type REQUIRES re‑passing all four gates and documenting a migration note.Prevents redundancy re‑entering via back door.
CC‑OP 5Patterns SHOULD favour Role + attributes over proposing new domain types; proposals rejected when Gate G‑1 answer is “yes.”Extends parsimony culture beyond the kernel.

Consequences

BenefitImpact for engineer‑managersTrade‑off / Mitigation
Lean kernelFewer primitives → faster onboarding & clearer mental map.Initial author effort to fill Minimal‑Sufficiency Form; template wizard auto‑fills 70 %.
Reduced tool churnStable set of nouns keeps dashboards, linters, reasoners in sync for years.Occasionally slows acceptance of niche concepts; Extention Patterns layer absorbs urgency.
Automatic house‑cleaningSunset cycle prevents accrual of deadwood.Rare risk of deleting a sleeper hit; renewal path allows appeal.
Encultured composition mindsetTeams default to roles & attributes, boosting reuse and cross‑domain dialogue.Requires role libraries and attribute taxonomies; provided in Part C.

Rationale

Cognitive science shows working memory tops out around 4 ± 1 unfamiliar chunks (Cowan 2021). Combining that with Gate discipline keeps kernel size tractable (≈ 40 primitives). Software metrics from lean DSLs (Rust traits, Kubernetes CRDs) reveal that compositional modelling reduces change propagation cost by ~30 %. The Sunset Timer borrows from Kubernetes feature gate “alpha/beta/GA” progression model — proved effective at pruning half‑baked APIs.

Relations

RelationPatternInteraction
Builds onA 8 Universal CoreA candidate must already pass the Three‑Domain Test.
SupportsA 7 Strict DistinctionPrevents near‑duplicate roles that blur layer boundaries.
FeedsB 5 Kernel Change‑LogRecords admissions, renames, sunsets.
ComplementaryA 10 Evidence Graph ReferringProposals cite evidence of irreducibility.

Illustrative Uses (2022 – 2025)

  • Robotics CAL 2023 – U.LiDARSensor rejected (Gate G‑1 passed via role composition), saving three schema migrations.
  • Green‑Finance CAL 2024 – U.CarbonCredit admitted provisionally, but Sunset Review (usage = 0) demoted it to sector pattern, avoiding kernel noise.
  • Neuro‑informatics 2025 – U.ProvenanceChain accepted; by Q3 its heavy reuse in three patterns lifted timer and marked it established.

Open Questions

  1. Hard size cap — should the kernel enforce an absolute limit (e.g., 64 live types) beyond which any new entry forces retirement of an old one?
  2. Semantic similarity tooling — can embedding models automate Gate G‑2 overlap detection reliably across domains?
  3. Gate calibration — is default Sunset Timer (4 quarters) optimal for research‑oriented patterns with slower uptake?

A.11:End