Measuring Leadership: A Network Analysis of Prominent Early Sufis (9th-10th cents. C.E.)

By Jeremy Farrell
Submitted to Session P5360 (Cultural Trends in the Abbasid Period, 2018 Annual Meeting
Islamic World;
7th-13th Centuries;
LCD Projector without Audio;
How do leaders gain substantial followings? The organization of human communities is classically associated with hierarchization, the distinction between leaders and the led. In light of problematic categories for describing the attraction between leaders and followers in existing literature, we propose metrics of the leader-follower relationship that expressed through network topology and stylometry. As a case study, we examine the hierarchization of a community of religious figures known as "Sufis" from 9th-10th century Iraq. While data standardization poses issues, we find that the most attractive Sufi leaders exhibit: (1) the strongest ties to followers; (2) the most original message; and (3) the most heterogenous message.

Existing theories of leadership appeal to the concept of attraction to explain hierarchization. For instance, several approaches to leadership consider ascriptive qualities (e.g. lineage, "charisma") as invariably attractive to followers. Conversely, other theories hold that constituencies follow individuals who exert control of over resources (e.g., currency, divine inspiration). Although these theories of attraction explain dynamics in contemporary communities, their application to historical communities assumes strong parameters that are difficult to identify in historical cases.

Network analysis offers an avenue to articulate concepts that are more applicable to historical instances of hierarchization. Three categories for evaluating leader-follower relationships can be expressed using commonly-used network metrics. First, we evaluate the "strength" of ties between leaders and followers. Second, we evaluate the originality of a leader's "message," composed of discrete "sayings." Topologically, this corresponds to the geodesic distance between a statement's originator (node[o]) and a leader (node[j]); thus, an "original" message returns a value of "0", while more "traditional" messages return increasingly higher values. Finally, we evaluate the heterogeneity of a leader's message through stylometric analysis. Using a list of key words, we evaluate their relative frequency in a leader's message: absolute values (0.0, 1.0) indicate homogenous messages, while a median value (0.5) indicates a heterogeneous message.

We apply these categories to a corpus of texts written by Sufis from the 9th-11th centuries, in order to discern the factors that influenced the development of a leader-follower hierarchy. A unique structural feature of these texts is a record of the transmission pathway for sayings by leading Sufis (i.e., x ? y ? z ['saying']); our corpus comprises nine works with ~1,900 transmission pathways.