ARI Scoring Model

Fixed Model (Protocol Constants)

• Dimensions: 5
• Weights: [0.30, 0.25, 0.20, 0.15, 0.10]
• dimension_score range: 0..200
• time_decay = exp(-lambda * days_since_action)
• volume_confidence = min(1, valid_actions_count / 100)
• Final ARI range: 0..1000

Engine State

Per agent state:

• decayedSum[5]
• lastUpdate
• totalActionsCount
• validActionsCount
• firstActionAt

Fixed-Point Decay

• Decay uses fixed-point factors (1e18 scale).
• Precomputed decay table is used for base horizons.
• Large horizon values are handled with chunked multiplication for deterministic behavior.
• daysSince is saturated for DoS-safe bounds (default 10000 days).

Update Flow

On each new valid action:

1. sync(agent) applies elapsed decay.

2. New score vector is added to decayedSum.

3. totalActionsCount and validActionsCount increment.

Dispute Correction Flow

When a scorecard is invalidated:

1. sync(agent) first.

2. Calculate daysSince from action timestamp.

3. Compute exact decay factor with fixed-point chunked math.

4. Subtract decayed contribution from each dimension (clamp at zero).

5. Decrement validActionsCount only.

Tier Mapping

• UNVERIFIED: 0-99
• PROVISIONAL: 100-299
• ESTABLISHED: 300-599
• TRUSTED: 600-849
• ELITE: 850-1000

Tier Policy Reference

Trust tiers are intended to inform access, review, and risk controls for partner integrations.

Final staking thresholds, dispute outcomes, and fee routing remain governance-controlled parameters.

Governance Parameter Note

Mainnet governance parameter set controls final staking thresholds, slash rates, and fee routes.

This document records default policy direction and testnet-aligned behavior.