Ethnography Has Left the Field

Abstract

Ethnography is commonly associated with immersive field observation, where researchers seek to understand behavior through prolonged presence within a setting. Conceptually, however, immersion represents a strategy for acquiring behavioral evidence rather than the defining essence of ethnography itself. Ethnographic reasoning is grounded in interpreting observable human activity within a defined context of interest.

Field observation samples behavior probabilistically, relying on immersion to encounter relevant actions as they occur. Advances in large language models introduce an alternative pathway by enabling intentional specification of research focus through an observable unit of behavior expressed as an

[individual] performing [task] in [context].

This formulation defines the context of interest explicitly and serves as the anchor for ethnographic decomposition.

Starting from this observable unit, ethnographic reasoning can expand tasks into subtasks, constraints, dependencies, artifacts, and role interactions based on behavioral evidence either observed in real-world settings or surfaced through structured decomposition within large language models. The outcome is a structured knowledge graph representing the context of interest as an explorable model rather than a narrative summary.

This separation between ethnography as a reasoning method and ethnography as a field practice transforms research from episodic observation toward continuous context modeling. Field studies remain essential for validation and refinement, but no longer function as the sole mechanism for context discovery.

A context model derived from observable behavior whether observed in reality or surfaced through structured decomposition in LLM provides a shared foundation across disciplines, allowing designers, engineers, and strategists to explore different perspectives while remaining anchored in the same contextual structure. Explicit modeling of relationships among roles, tasks, constraints, and decisions enables ethnographic insight to become operational, reusable, and scalable.

Ethnography therefore evolves into an infrastructural capability: a continuous process of contextual decomposition and synthesis that transforms observable behavior into a structured representation of context of interest, guiding discovery, informing design, and enabling focused real-world validation.