Reputation Decay

Applies a constant decay factor (λ) to historical observations, causing reputation weight to decrease exponentially with age. Each time step reduces the contribution of past events by a fixed percentage.

• Formula: weight(t) = e^(-λt) where t is the age of the observation
• Half-life: The time required for an observation's weight to halve, providing an intuitive tuning parameter
• Use case: Rapidly discounting stale performance data in high-velocity environments like algorithmic trading or real-time bidding systems

A hard cutoff mechanism that completely discards all observations older than a defined retention horizon. Only events within the active window contribute to the current reputation score.

• Fixed window: A static time range (e.g., 90 days) that drops old data entirely
• Adaptive window: Dynamically adjusts the window size based on observation density or system volatility
• Advantage: Computationally efficient and provides clear audit trails for compliance requirements

Calculates reputation as a weighted moving average where recent interactions carry higher multipliers than older ones. Unlike binary windowing, this preserves all historical data but with diminishing influence.

• Linear decay: Weight decreases proportionally with age
• Harmonic decay: Weight decreases as 1/t, creating a long-tail of residual influence
• Application: Ideal for systems where past behavior retains some predictive value, such as author credibility scoring in academic databases

A parameter (α) in recursive estimation algorithms that controls how aggressively the model updates its beliefs based on new evidence versus retaining prior estimates.

• High α (near 1.0): Strong emphasis on recent data; rapid adaptation to behavior changes
• Low α (near 0.0): Conservative updating; reputation changes slowly
• Kalman filter integration: Used in Bayesian reputation systems to dynamically adjust the forgetting factor based on observation noise and system uncertainty

Applies non-linear decay boosts when specific critical events occur, such as a security breach or policy violation. This mechanism overrides standard temporal decay to immediately penalize trust.

• Penalty multipliers: A violation applies a 10x decay factor to all historical data
• Cool-down periods: After a triggering event, the system enforces a minimum observation period before reputation can recover
• Use case: Sybil resistance in peer-to-peer networks where a single malicious act must rapidly outweigh years of benign history

Gradually reduces the reputation score of entities that cease contributing to the network, reflecting the principle that trust requires continuous validation.

• Dormancy threshold: Decay begins after a defined period of inactivity (e.g., 30 days)
• Reactivation boost: A returning entity may receive a partial restoration of their pre-atrophy score to avoid cold-start penalties
• Application: Critical for validator reputation in proof-of-stake networks where offline nodes pose a liveness risk

The logical property that allows trust to flow through a network. If entity A trusts entity B, and entity B trusts entity C, then A can derive a measure of trust for C.

• Decay Interaction: Transitive trust is highly susceptible to reputation decay; a stale trust link between B and C can poison A's entire downstream trust calculation.
• Application: Used in Web of Trust and EigenTrust algorithms to compute global reputation from local observations.
• Risk: Without decay, transitive paths amplify outdated or compromised trust indefinitely.

A statistical approach that uses Bayesian inference to update the probability distribution of an entity's trustworthiness based on sequential observations.

• Decay Integration: Decay is implemented as a forgetting factor that reduces the weight of older observations in the posterior distribution.
• Mechanism: Beta distributions are commonly used, where decay shifts the effective sample size of historical positive and negative outcomes.
• Advantage: Provides a mathematically rigorous framework for expressing uncertainty in reputation scores.

A programmable penalty mechanism in proof-of-stake and reputation protocols that destroys a portion of a validator's staked assets or reputation score for provably malicious behavior.

• Decay Distinction: Slashing is a discrete, punitive event, while decay is a continuous, time-based reduction. Both serve to prevent the indefinite accumulation of unearned trust.
• Examples: Ethereum's Casper FFG slashes validators for double-signing; reputation systems slash for Sybil attacks.
• Design: Slashing conditions must be cryptographically provable to avoid disputes.

A distributed reputation management algorithm for peer-to-peer networks that calculates a global trust value for each peer by analyzing transitive trust relationships.

• Decay Role: EigenTrust relies on historical transaction satisfaction; without decay, peers with past good behavior but current malice retain high scores.
• Computation: Uses eigenvector centrality on a normalized trust matrix, iteratively converging to a global trust vector.
• Vulnerability: Static scores are susceptible to whitewashing attacks, which decay mechanisms help mitigate.

A type of probabilistic logic that explicitly models uncertainty and belief ownership, representing trust as a composite of belief, disbelief, and uncertainty masses.

• Decay Modeling: Decay can be expressed as a temporal discount operator that shifts mass from belief to uncertainty over time, reflecting increasing doubt in stale evidence.
• Operators: Supports fusion, discounting, and consensus operators for combining multiple trust opinions.
• Use Case: Ideal for supply chain provenance where sensor data trust degrades with latency.

The process of assigning initial trust values to new entities that lack historical interaction data, addressing the cold start problem.

• Decay Relationship: Aggressive decay policies exacerbate the cold start problem by rapidly erasing hard-earned reputation, making bootstrapping strategies more critical.
• Strategies: Include default trust, vouching by established entities, and Soulbound Token transfers.
• Trade-off: High initial trust accelerates onboarding but increases Sybil vulnerability; low initial trust creates friction.