Scoring Model

FlameWire uses a multi-factor scoring system to evaluate miner performance and distribute ALPHA emissions fairly. This document explains every component of the scoring algorithm.

Part 1: Node Score

Each node receives an individual score based on three metrics: correctness, uptime, and latency.

1.1 Uptime Score

Measures how reliable and available the node is. A node that goes offline frequently or falls out of sync cannot serve clients.

Data required:

• Passed health checks — node responded with is_syncing = false and block height matches network head
• Total health checks — number of checks performed

Formula:

Uptime Score = passed / total

Range: 0.00 - 1.00

Example: Stable Node

Example: Unstable Node

1.2 Correctness Score

Verifies that the node returns correct data. A node with incorrect data is dangerous — it can cause financial losses for clients. Any error means total failure.

Data required:

• Correctness tests performed (extrinsic and hash comparisons)
• Result: all correct or not

Formula:

Correctness Score = 1 if all tests pass, 0 if any test fails

Range: 0 or 1 (binary)

Example: Correct Node

Validator tests 3 extrinsics (old, middle, recent):

• Extrinsic block 1,000,000: hash correct ✓
• Extrinsic block 3,500,000: hash correct ✓
• Extrinsic block 5,000,000: hash correct ✓

Correctness Score: 1.00

Example: Corrupt Node

Validator tests 3 extrinsics:

• Extrinsic block 1,000,000: hash correct ✓
• Extrinsic block 3,500,000: hash incorrect ✗
• Extrinsic block 5,000,000: hash correct ✓

Correctness Score: 0.00 — Could be incomplete archive, corrupted database, or fraud attempt.

1.3 Latency Score

Measures response speed compared to other nodes in the network. For an RPC provider, speed is critical for user experience.

Data required:

• Current node latency (ms)
• Minimum latency in network (ms)
• Maximum latency in network (ms)

Formula:

Latency Score = 1 - (node_latency - min_latency) / (max_latency - min_latency)

Range: 0.00 - 1.00

Special cases:

• Fastest node: Latency Score = 1.00
• Slowest node: Latency Score = 0.00
• All nodes same latency: Latency Score = 1.00 for all

Example: Fast Node

Example: Slow Node

1.4 Combined Node Score

Combines the three metrics into a single score per node.

Formula:

Node Score = 0.40 × Correctness + 0.30 × Uptime + 0.30 × Latency

Example: Excellent Node

Example: Problematic Node

Part 2: Regional Factors

2.1 Regional Multiplier

Encourages balanced geographic distribution. Nodes in regions with few nodes are worth more because they help the network be more robust globally.

Data required:

• Number of nodes in each region
• Total number of nodes in network

Formula:

Regional Multiplier = min(2.0, max(0.5, 33.33% / actual_percent%))

Range: 0.50 - 2.00

Limits explained:

• Minimum 0.5: Even if a region is very saturated, you don't lose more than half the value
• Maximum 2.0: Even if a region is very empty, you don't gain more than double

Example: Node in Rare Region

A node in Asia is worth double! The network desperately needs nodes there.

2.2 Diminishing Returns

Prevents node spam in a single region. Each additional node added by the same miner in the same region contributes less and less.

Formula:

Node N Contribution = Node Score × (1 / N)
where N = which node number for this miner in this region

Example: Diversified Miner

Miner has 3 nodes in 3 different regions (all Node Score = 0.90):

Example: Spam in One Region

Miner has 3 nodes, all in Europe (all Node Score = 0.90):

2.3 Regional Score

Combines node contributions with the regional multiplier for each region.

Formula:

Regional Score = Sum of contributions with diminishing × Regional Multiplier

Example Comparison

3 nodes in Europe are worth less than 1 node in Asia!

Part 3: Diversity Bonus

Rewards miners who cover multiple geographic regions. A global network needs presence in all zones.

Part 4: Raw Miner Score

Calculates the miner's total score before smoothing.

Formula:

Raw Miner Score = (Regional Score EU + Regional Score US + Regional Score AS) × Diversity Bonus

Example: Strategic Miner

Example: Non-Strategic Miner

With 4 nodes, this miner earns less than the strategic miner with 3 nodes!

Part 5: EMA Smoothing

Prevents sudden score jumps. Protects good miners from temporary bad luck and prevents manipulation through intermittent performance.

Formula:

EMA Score = α × Raw Score + (1 - α) × Previous EMA Score

Where α (alpha) = 0.1

• 10% from new score
• 90% from previous score
• Changes propagate gradually over ~20-30 epochs

Example: Stable Performance

Example: Sudden Drop Cushioned

Part 6: Final Weight

Transforms scores into percentages that sum to 100%. These become the weights set on blockchain.

Formula:

Weight = EMA Score / Sum of all EMA Scores

Example: Network with 4 Miners

Complete Formula Summary

Step 1: Calculate Node Score

Uptime Score = passed / total
Correctness Score = 1 if all tests pass, 0 otherwise
Latency Score = 1 - (node_latency - min) / (max - min)

NODE SCORE = 0.40 × Correctness + 0.30 × Uptime + 0.30 × Latency

Step 2: Calculate Regional Multipliers

For each region:
    actual_percent = region_nodes / total_nodes
    multiplier = 33.33% / actual_percent
    REGIONAL MULTIPLIER = min(2.0, max(0.5, multiplier))

Step 3: Calculate Contributions per Miner per Region

For each miner, for each region:
    For each node (sorted by Node Score descending):
        CONTRIBUTION = Node Score × (1 / N)
        where N = node number in this region

    REGIONAL SCORE = Sum of contributions × Regional Multiplier

Step 4: Calculate Diversity Bonus

regions_covered = number of regions where miner has at least 1 node

DIVERSITY BONUS =
    1.00 if regions = 1
    1.10 if regions = 2
    1.20 if regions = 3

Step 5: Calculate Raw Miner Score

RAW MINER SCORE = (Regional Score EU + Regional Score US + Regional Score AS) × Diversity Bonus

Step 6: Apply EMA Smoothing

If first epoch for miner:
    EMA SCORE = Raw Miner Score
Else:
    EMA SCORE = 0.1 × Raw Miner Score + 0.9 × Previous EMA Score

Step 7: Calculate Final Weight

WEIGHT = EMA Score / Sum of all EMA Scores in network

Complete End-to-End Example

Network situation:

• Total nodes: 20
• Europe: 12 nodes (60%) → Multiplier: 0.56
• US: 6 nodes (30%) → Multiplier: 1.11
• Asia: 2 nodes (10%) → Multiplier: 2.00

Miner X has:

• Node 1 in EU: Uptime=0.99, Correctness=1, Latency=0.85
• Node 2 in EU: Uptime=0.95, Correctness=1, Latency=0.80
• Node 3 in Asia: Uptime=0.98, Correctness=1, Latency=0.70

Step 1: Node Scores

Step 2: Contributions with Diminishing (EU sorted by score)

Step 3: Regional Scores

Step 4: Diversity Bonus

Miner has nodes in 2 regions (EU and AS) → 1.10

Step 5: Raw Miner Score

Raw = (0.79 + 0 + 1.81) × 1.10 = 2.60 × 1.10 = 2.86

Step 6: EMA

Previous EMA was 2.50:

EMA = 0.1 × 2.86 + 0.9 × 2.50 = 0.286 + 2.25 = 2.54

Step 7: Final Weight

If sum of all EMAs in network is 15.00:

Weight Miner X = 2.54 / 15.00 = 16.9%