Label Propagation (Raghavan et al., 2007) runs in near-linear O(m) time: assign each node a unique label, then iteratively update each node to the most frequent label among its neighbors. Seed nodes (shown with rings) have fixed labels and act as sources of community identity. The algorithm converges when no node changes label, typically in O(log n) steps. The randomized update order means results can vary between runs — watch how labels spread like dye through the network, filling communities from the seeds outward.