Show & Tell 30.3 — GoalPrioritizer Architecture

[web3guru888]

Show & Tell: Phase 30.3 — GoalPrioritizer Architecture

Overview

The GoalPrioritizer ranks candidate goals using multi-objective Pareto optimization, detects conflicts, and dynamically adjusts dimension weights based on current drive states.

Pareto Front Visualization

  Importance
  ▲
  │     ★G3          ← Pareto Front 0 (non-dominated)
  │   ★G1    ★G5
  │              ★G7
  │  ·G2    ·G4      ← Pareto Front 1
  │     ·G6
  │  ··G8             ← Pareto Front 2
  │
  └───────────────────▶ Urgency

★ = Front 0 (dominant goals)
· = Front 1+ (dominated goals)

Priority Dimension Scoring

Goal Assessment Matrix:
              URGENCY  IMPORTANCE  FEASIBILITY  ALIGNMENT  NOVELTY
Goal A:       0.9      0.7         0.8          0.95       0.3
Goal B:       0.3      0.9         0.6          0.90       0.8
Goal C:       0.6      0.5         0.9          0.85       0.5

Weights:      0.25     0.25        0.20         0.20       0.10
              ▲                                             ▲
              (dynamic: ↑ when CURIOSITY high)     (dynamic: ↑ when CURIOSITY high)

Conflict Detection Categories

┌─────────────────────────────────────────────────────┐
│              ConflictDetector                         │
│                                                       │
│  ┌─────────────┐  ┌───────────────┐  ┌────────────┐ │
│  │  Resource    │  │  Temporal     │  │  Logical   │ │
│  │  Conflicts   │  │  Conflicts   │  │  Conflicts │ │
│  │             │  │               │  │            │ │
│  │ Shared      │  │ Overlapping   │  │ Mutually   │ │
│  │ compute/    │  │ deadlines     │  │ exclusive  │ │
│  │ memory      │  │ with insuff.  │  │ precondit. │ │
│  │ budget      │  │ capacity      │  │            │ │
│  └─────────────┘  └───────────────┘  └────────────┘ │
│                                                       │
│  Output: list[tuple[goal_a, goal_b, conflict_type]]  │
└─────────────────────────────────────────────────────┘

Dynamic Reweighting Flow

DriveState                      Weight Adjustment
─────────────                   ─────────────────
CURIOSITY = 0.8  ──────────▶   NOVELTY ×1.5, URGENCY ×1.2
HOMEOSTASIS = 0.7 ─────────▶   FEASIBILITY ×1.5, NOVELTY ×0.5
COMPETENCE = 0.3  ─────────▶   (minimal adjustment)

Before reweight: [0.20, 0.20, 0.20, 0.20, 0.20]
After reweight:  [0.24, 0.15, 0.27, 0.18, 0.16]  (renormalized)

Pipeline Summary

list[GoalSpec] → Score(5D) → ParetoSort → ConflictDetect → Penalize → list[RankedGoal]
     20            20×5         3 fronts       5 pairs         -0.3        20 ranked

Ref: Issue #645 | Wiki: Phase 30 Goal Prioritizer