Random walk mixing times reveal how users get trapped in algorithmic clusters on social platforms. This research maps the mathematical structure of these "rabbit holes" to understand information silos. It provides a formal framework for measuring content isolation. Practitioners can use these metrics to audit recommendation engines for diversity and bias.