Research / Paper
Graph-Based Human Verification
Trust-flow framework for one-account-per-person communities.
Theoretical work on canonical identity selection, trust propagation, and how to preserve usability without central identity authority.
The mechanism
Trust is finite and flows through the graph under constraints, which lets a canonical identity emerge from local vouching rather than a central authority. The framework covers canonical node selection, duplicate handling, and both local and global trust perspectives.
Evaluation
Assessed across uniqueness, controversy tolerance, and interpretability — the goal being a system that stays usable and privacy-preserving rather than one that merely maximizes detection.