Your-Fit-Tailored Constitution

Architectural invariants that govern all system design. These are non-negotiable truths that must hold across all operations, regardless of implementation choices.

1. Weekly Cadence Invariant

Statement: 7-day cycles are non-negotiable; exceptions close with recovery, never skip.

Implications:

  • Every Active UserEntity has a deterministic mapping from current time to exactly one $WEEK_ID
  • At most one open CycleEntity per $WEEK_ID per user
  • Deviations from cadence introduce compounding variance into cleaning queues, inventory availability, and user trust
  • A delayed box is not “late shipping” but a broken cycle that must be logged and compensated at the system level
  • Exceptions are allowed only as explicit state changes, not silent delays

Violation: Any silent delay, implicit skip, or cadence drift that is not recorded as an explicit compensating event.

2. Circular Inventory Invariant

Statement: Garments are stateful assets with lifecycle bounds; inventory is closed-loop.

Implications:

  • A GarmentEntity cannot be in two boxes at once
  • A GarmentEntity cannot be both Available and InTransit simultaneously
  • A GarmentEntity cannot re-enter Available without passing through inspection and any required refurbishment
  • Each garment has a maximum allowable lifecycle budget (bounded set of counters)
  • Transitions must enforce retirement when bounds are exceeded
  • Every garment exit path (resale, donation, recycling, downcycling) must be defined at inventory intake

Violation: Double-allocation, phantom inventory, garments returning to circulation without inspection, or lifecycle bounds being ignored.

3. State Truth Discipline

Statement: All entities auditable; single subsystem owns each state class.

Implications:

  • Exactly one subsystem is authoritative for each class of state
  • StateAuthoritySubsystem is the canonical source of truth for entity state and transitions
  • External signals (carriers, facilities, payment rails) are inputs that must be validated, reconciled, and translated
  • The platform must not delegate canonical state transitions to external actors
  • All identifiers and temporal references are tenant- and time-scoped ($TENANT_ID, $WEEK_ID, $CYCLE_ID)
  • State transitions are event-driven, monotonic where possible, reversible only via explicit compensating events

Violation: Split-brain state, external systems treated as authoritative, in-place mutation instead of compensating events, or ambiguous entity ownership.

4. Explicit Failure Handling

Statement: Failures are normal; recovery paths are first-class, not exceptions.

Implications:

  • The system assumes failure as a normal operating condition
  • Each failure must resolve into one of three outcomes: cost absorption, user compensation, or inventory state adjustment
  • No failure is allowed to remain unclassified (unclassified failures accumulate as hidden debt)
  • Recovery paths are designed upfront, not bolted on as exception handling
  • Graceful degradation over brittle optimization

Failure Categories:

  • Inventory shortage before commitment → controlled degradation, automatic skip, clear accounting
  • Packing variance → record variance event, correct or commit to observed set
  • Delivery signal missing → bounded uncertainty state, DeliveredByProxy events
  • Late return → exception control loop, progressive escalation, HoldLogistics state
  • Damage/contamination → immediate quarantine, SafetyIncident event
  • Payment failure → HoldPayment state, AccountsReceivable tracking

Violation: Silent failures, unclassified exceptions, failures that corrupt state correctness, or ad hoc operational interventions.

5. Probabilistic Fit

Statement: Fit is a belief distribution that evolves, not static measurement.

Implications:

  • No subsystem may collapse fit into a single deterministic fact for allocation decisions
  • Allocations must incorporate uncertainty and remain correct even when predictions are wrong
  • Fit confidence is modeled as a rolling posterior probability
  • Each cycle should increase fit certainty or reduce uncertainty
  • Fit belief updates occur after delivery, after wear, and after return inspection
  • Both explicit user signals and implicit behavioral signals inform fit beliefs
  • UserFitBelief includes DriftModel (gradual body changes)
  • GarmentFitBelief includes AgingModel (repeated refurb effects)
  • FitIntelligenceSubsystem must treat explicit feedback as optional

Violation: Deterministic fit matching, ignoring uncertainty in allocation, requiring explicit user feedback, or fit thrash (oscillating between sizes without learning).

6. Cognitive Load Minimization

Statement: System defaults for all routine decisions; user input optional.

Implications:

  • User cognitive load must asymptotically approach zero as the system learns and stabilizes
  • The system must have defaults for all routine decisions
  • Users must never be required to understand the internal state machine
  • Explicit user input is optional and rare, reserved for exceptions that cannot be safely resolved automatically
  • Any UI, communication, or policy that shifts cognitive load back onto the user violates the core premise
  • The user-facing state model is an intentionally simplified projection of the canonical state
  • ExperienceMinimizationSubsystem defines what decisions are optional and what defaults apply

Violation: Requiring users to make routine decisions, exposing internal state complexity, or pursuing “low cognitive load” through lack of transparency rather than high-quality defaults plus precise exception handling.

Cross-Cutting Verification

Any downstream design—whether technical architecture, warehouse layout, or pricing model—should be evaluated by asking:

Does this choice reduce uncertainty, friction, or variance across cycles without increasing cognitive load on the user?

If not, it is misaligned with this constitution.