Social networks are searchable, allowing people to find distant targets in a few steps. This paper presents a model explaining this searchability through social identities defined along multiple dimensions. The model defines searchable networks and a search method applicable to various network problems, including peer-to-peer networks and the web. In 1969, Travers and Milgram showed that messages could reach a target in about six steps, revealing short paths in social networks. This suggests that people can find short paths despite limited knowledge of the network. Searchability exists in certain network types, but these are not satisfactory models of society. The paper proposes a model based on social structures that explains searchability. It is derived from six contentions about social networks. Individuals have identities defined by social groups, which hierarchically cluster the world. Group membership influences social interactions, and individuals are more likely to connect with similar others. Social distance is measured across multiple dimensions, and individuals use local information to forward messages. The model shows that searchability depends on parameters like the number of dimensions (H) and homophily (α). A greedy algorithm, where each person forwards a message to the neighbor perceived to be closest to the target, is effective. The model's results match Milgram's experiment, showing average chain lengths of about six steps. The model is relevant to decentralized search problems, such as peer-to-peer networks, where centralized servers are not available. It allows for efficient searches by leveraging multiple social dimensions. The model's parameters align with empirical findings, and it captures the essence of real-world small-world problems. The paper concludes that searchability is a generic property of social networks, applicable to various data structures where elements can be compared along multiple dimensions.Social networks are searchable, allowing people to find distant targets in a few steps. This paper presents a model explaining this searchability through social identities defined along multiple dimensions. The model defines searchable networks and a search method applicable to various network problems, including peer-to-peer networks and the web. In 1969, Travers and Milgram showed that messages could reach a target in about six steps, revealing short paths in social networks. This suggests that people can find short paths despite limited knowledge of the network. Searchability exists in certain network types, but these are not satisfactory models of society. The paper proposes a model based on social structures that explains searchability. It is derived from six contentions about social networks. Individuals have identities defined by social groups, which hierarchically cluster the world. Group membership influences social interactions, and individuals are more likely to connect with similar others. Social distance is measured across multiple dimensions, and individuals use local information to forward messages. The model shows that searchability depends on parameters like the number of dimensions (H) and homophily (α). A greedy algorithm, where each person forwards a message to the neighbor perceived to be closest to the target, is effective. The model's results match Milgram's experiment, showing average chain lengths of about six steps. The model is relevant to decentralized search problems, such as peer-to-peer networks, where centralized servers are not available. It allows for efficient searches by leveraging multiple social dimensions. The model's parameters align with empirical findings, and it captures the essence of real-world small-world problems. The paper concludes that searchability is a generic property of social networks, applicable to various data structures where elements can be compared along multiple dimensions.