REP Score Architecture

The Four Components

REP is composed from four distinct behavioral components that measure different aspects of protocol participation and structural credibility:

REP_total = REP_Contribution + REP_Accuracy + REP_Integrity + REP_Network

REP_Contribution

What it is: REP earned by contributing intelligence to the protocol.

Contribution REP is awarded for:

• Submitting structural insights (channel analysis, narrative identification, creator mapping)
• Adding to the signal graph through verified observations
• Publishing research that is validated by subsequent model outputs

Earning rate: Scales with contribution quality, not volume. A single high-precision insight earns more than dozens of low-signal submissions.

Weight in REP_total: High — contribution is the primary legitimate path to building REP.

REP_Accuracy

What it is: REP earned by correctly identifying structural events before they are broadly recognized.

Accuracy REP is the most valuable component of REP. It is earned when:

• A breakout prediction (citing AVI/BPM data) is validated by subsequent channel growth
• A fragility warning (citing SIS/TRI data) is validated by subsequent collapse
• Narrative strength assessment (NSM-based) is validated by cross-creator propagation

Earning rate: Binary validation — predictions are scored as correct or incorrect based on observable outcomes within a defined time window.

Why it is the most valuable: Accuracy REP requires being right about structure before the market prices it. This is the hardest signal to manufacture and the most valuable to the protocol.

REP_Integrity

What it is: A component that decreases with protocol-damaging behavior.

REP_Integrity starts at its maximum contribution and is reduced by:

• Submitting spam or low-quality signals at high volume
• Coordinated manipulation attempts (identified via behavioral pattern analysis)
• Artificial engagement generation
• Misrepresenting channel ownership or creator identity

Earning rate: Does not earn — only loses. Integrity is preserved by clean behavior, not by actions.

Recovery: REP_Integrity can recover over time after a violation, but recovery is slower than the initial loss. Severe violations can result in permanent integrity penalties.

REP_Network

What it is: REP earned through high-quality referrals and creator onboarding.

Network REP is earned when:

• A referred Telegram creator achieves a SIS > 50 within 60 days of joining
• Onboarded channels demonstrate sustained organic growth (not spike-then-collapse)
• Referred users themselves build meaningful REP through accurate contributions

Weight in REP_total: Lower than Contribution and Accuracy — network is a supplementary path, not the primary one.

Anti-gaming mechanic: Network REP is only awarded when referred entities meet structural thresholds. Referring bots or hollow channels earns nothing. Referring fake accounts triggers an integrity penalty.

Influence Weighting

REP does not translate linearly to influence. The protocol uses a logarithmic influence model to prevent score dominance by a small number of very high-REP entities:

InfluenceWeight = 1 + log(1 + REP)

Influence Weight Examples

The logarithmic compression means:

• Going from REP 0 → 99 gives significant influence gain
• Going from REP 600 → 1000 gives diminishing additional influence
• A single entity with REP 1000 cannot dominate a group of 10 entities at REP 200

This is the anti-whale mechanic — it prevents reputation concentration from translating into disproportionate protocol control.