contradiction-tracker

Contradiction tracking works by storing statements in memory with dates and sources, then comparing new claims to past ones. It classifies inconsistencies into types like direct reversal, selective memory, and gradual drift.

When analyzing political statements, the system detects direct reversals, selective memory, moving goalposts, contextual flips, and gradual drifts. These classifications help identify how a public figure's position has shifted.

Start fact-checking inconsistent positions by logging an entity's initial statement with its source and date. Submitting a new claim afterward triggers an automated contradiction check against stored memory.

Yes, memory limits apply when tracking a public figure: up to 20 statements are stored per entity. Topic tagging manages this capacity by enabling targeted contradiction checks and analysis.

Yes, you can organize logged statements using topic tagging. This memory management feature enables targeted checks and analysis when comparing new claims against past entity statements.