What are heuristic rules in spam filtering?

Heuristic rules are predefined patterns that filters use to identify spam characteristics. They encode expert knowledge about what spam typically looks like, from suspicious phrases to structural anomalies.

A heuristic might flag messages with excessive capitalization, multiple exclamation marks, or phrases commonly found in scam emails. It might detect hidden text, invisible characters, or HTML tricks designed to evade detection. Each matching rule adds points to the spam score.

SpamAssassin is famous for its extensive heuristic ruleset, with hundreds of tests that collectively evaluate a message. Rules have different weights based on their reliability as spam indicators.

Heuristics were the foundation of early spam filtering before machine learning became dominant. Today they work alongside ML models, catching patterns that statistical approaches might miss.

Heuristics are the experienced sailor's instincts, refined rules of thumb that spot trouble before the instruments do.