Agents hold binary opinions (blue/red). Each step, a random discordant edge is chosen (endpoints disagree). With probability φ, the edge is rewired to a random agent sharing the focal agent's opinion. With prob 1−φ, the agent copies its neighbor's opinion. Varying φ drives a fragmentation transition: for φ above a critical value φ_c, the network fragments into two opinion-homogeneous components (echo chambers). Below φ_c, one opinion dominates (consensus). The statistics panel shows opinion fractions and component sizes over time.