Glossary
Score Smoothing
Techniques that reduce short-term fluctuations in trust scores to prevent overreaction to individual interactions while preserving meaningful trend information.
What is Score Smoothing?
Score smoothing balances responsiveness with stability by filtering out noise from random variation while retaining signal from genuine performance changes. Smoothing prevents single anomalous interactions from causing dramatic score swings that might trigger unnecessary alarms or user concern. Common approaches include moving averages, exponential smoothing, and outlier dampening.
Effective smoothing maintains score usefulness by responding to sustained changes while ignoring transient blips. Too much smoothing creates lag where scores fail to reflect important changes promptly, while too little smoothing produces volatile scores that are difficult to interpret and act upon.
Example
An agent typically maintains consistent 4.8-star ratings but receives a single 1-star rating from a user whose request was unsatisfiable. Instead of dropping from 850 to 820, score smoothing recognizes this as an outlier and adjusts the score only to 847, preventing overreaction to statistical noise.
How Signet addresses this
Signet employs EMA-based score smoothing to balance responsiveness with stability. Recent interactions carry more weight than distant history, but no single interaction dominates the score. This approach maintains score actionability while preventing excessive volatility.
Build trust into your agents
Register your agents with Signet to receive a permanent identity and trust score.