Universal Selection Engine — USE v0.1.0
A constitutional doctrine for selection on ICore: given verified constraints, capabilities, policies, and objectives, select the best compatible option while preserving constitutional outcomes.Status: Doctrine Constitutional dependencies: USCP (six primitives) · USC · URS v0.1.0 (specs.initialcore.net/urs.html) · CORE-LD v0.1.0 (specs.initialcore.net/core-ld.html) · UAM v0.2.0 (specs.initialcore.net/uam.html) Scope: All ICore systems — constitution, applications, agents, runtimes — at every decision point between Verification and Execution Registry prefix: USE
# Part I — USE Specification
1. Definition
USE is the constitutional Selection rung of the UAM ladder, the decision point between Architecture and Capabilities:
*Given verified constraints, capabilities, policies, and objectives, select the best compatible option while preserving constitutional outcomes.*
Selection is the constitutional act; switching is a consequence. A switch changes from one option to another; a decision chooses a course of action; a selection identifies the best compatible option according to constitutional rules. Once selected, the system may switch, route, execute, negotiate, schedule, adapt — or do nothing. Selection precedes every adaptive behavior, which is why it is the constitutional primitive and the rest are implementations.
2. The Operational Triad
Verify — what is valid? Select — what is best? Execute — what is done?
| Phase | Question | Constitutional home |
|---|---|---|
| Verify | What is valid? | TRUST-LD, verification checkers (CORE-LD) |
| Select | What is best? | USE — this rung |
| Execute | What is done? | POLICY-LD, MEMORY-LD audit trail |
No selection without verification; no execution without selection. A selection of an unverified option is verification theater; an execution without a selection is drift.
3. Universality
Selection is universal, not merely software. Evolution selects; physics selects stable states; mathematics selects valid solutions; logic selects valid inferences; networks select routes; compilers select optimizations; AI selects models and actions. Selection is the universal bridge between possibility and actuality. USE governs that bridge where it crosses constitutional territory.
4. Invariants
| # | Invariant | Mechanical verification |
|---|---|---|
| USE-1 | **Verified options.** No unverified option enters selection; every considered option carries a verification result from a declared checker. | Selection record lists every considered option with its verification result before any selection |
| USE-2 | **Invariant preservation.** The selected option preserves every constitutional invariant the system touches; no selection may override an invariant. | Invariant-preservation check passes and is recorded in the USE log prior to execution |
| USE-3 | **Declared best-compatible.** Selection follows declared objectives and canonical defaults; the preference order is declared, never implicit. | Declared objective or preference order present in the selection record; selected option maximal under it (machine-checkable) |
| USE-4 | **Selection record.** Every selection is recorded before execution — options, verification results, rationale, outcome — in the constitutional audit trail. | Signed USE log entry (TRUST-LD) precedes execution; MEMORY-LD append |
| USE-5 | **Rung boundary.** Selection operates at the Selection rung only: upper rungs are preserved untouched; lower rungs are replaceable but never elevated to requirements. | Capability-free path exists (UAM-4); no selection record constrains an upper rung |
| USE-6 | **Do-nothing default.** If no verified compatible option preserves all invariants, selection yields no option: the system does nothing rather than violate. | Empty-selection record cites the blocking invariant(s) |
5. Four-Question Stress Test
| Claim | Why constitutional | What invariant | How verified | If absent |
|---|---|---|---|---|
| Verified options | USCP Verification — unverified inputs cannot be governed | USE-1 | Verification results attached to every considered option | Selection of the unverified: verification theater |
| Invariant preservation | USCP Constraint — invariants bound the solution space | USE-2 | Invariant check recorded before execution | Selection erodes constitutional outcomes |
| Declared best-compatible | USCP Transformation — governed choice among transformations | USE-3 | Preference maximality machine-check | Arbitrary choice: selection ungovernable |
| Selection record | USCP Identity — every governed act traces to its actor | USE-4 | Signed audit trail precedes execution | Governance blind: no accountability |
| Rung boundary | USCP Existence — the system exists independently of any environment | USE-5 | Capability-free path test (UAM-4) | Capabilities become constitutional dependencies |
| Do-nothing default | USCP Constraint — violation is never a compatible outcome | USE-6 | Blocking invariant cited in empty-selection record | Constitution becomes negotiable |
# Part II — USE Doctrine
1. The Boundary Between Certainty and Adaptability
The Selection rung is the constitutional boundary. Everything above it is intended to be stable — Truth, Principles, Constitution, Architecture, Verification. Everything below it is intended to be replaceable — transports, storage engines, runtimes, AI models, protocols, platforms, interfaces. USE is the decision point that keeps the two orders honest: stability above, adaptability below, and a governed act between them. A system that selects from unverified options is not adaptable — it is ungoverned. A system that never selects is not stable — it is frozen.
2. Selection Is the Act; Everything Else Is Implementation
Switching, routing, negotiating, scheduling, adapting — each is a consequence of a prior selection. Treating the switch as the primitive inverts the constitution: adaptation becomes arbitrary and capability drift erodes constitutional outcomes. USE fixes the order — verify, select, execute — so that every adaptive behavior carries its constitutional provenance.
3. Selection in the Executing Constitution
USE is not philosophy; it is the middle of the first executing governance pipeline. In the CodeLab, TRUST-LD attests (verify), USE selects the compatible policy and action (select), POLICY-LD constrains the execution, MEMORY-LD appends the auditable trail (execute). USE's invariants are the contract that pipeline honors: no attestation unverified, no action unselected, no selection unrecorded, no invariant overridden. The CodeLab is USE's first conformance exercise; USE flips Doctrine → Frozen only when that exercise passes with its own conformance suite.
4. Canonical Selection Defaults
Opinionated defaults, replaceable only by constitutional process (UAM §3 rules):
| Domain | Canonical default |
|---|---|
| Selection input | Verified options only — verification result required (USE-1) |
| Selection objective | Declared objective or preference order, never implicit (USE-3) |
| Selection outcome | Preserve invariants first, optimize second (USE-2) |
| Selection record | Signed USE log entry before execution (USE-4) |
| No compatible option | Do nothing — cite the blocking invariant (USE-6) |
Compatible alternatives are conforming if and only if they preserve USE-1..6, declare their compatibility profile (CCA §4), and are reversible.
5. Derivation Chain and Freeze Gate
Derivation: USCP → USC → UCE/UCM/UCC → URS → CORE-LD → UAM → USE. USE is the final rung of the doctrine chain; nothing below it is constitutional.
Freeze gate: USE v0.1.0 flips Doctrine → Frozen upon the CodeLab (TRUST-LD → USE → POLICY-LD → MEMORY-LD) executing green against a USE conformance suite covering USE-1..6 — then, and only then, UCF attestation (claim discipline: "passed its own suite," never external validation).
End of USE Doctrine v0.1.0 — August 1, 2026. Doctrine. Derives from USCP + CORE-LD v0.1.0 + UAM v0.2.0.