The first step is a cognitive ethnographical description of the activity, Hutchins proposes 3 levels:

DISTRIBUTED COGNITION FIRST LEVEL OF DESCRIPTION: description of the tools and representations "details of how the algorithm and representation are realized physically" Marr, 1982. Description of the Propagation of representational state across a set of physical devices (charts...) + physical activities of the members of the navigation team. This task specification permits construction of the computational level of description for the individuals.

DISTRIBUTED COGNITION SECOND LEVEL OF DESCRIPTION: cognitive requirements of performance of the task. What are the people in the setting doing? What do they have to do in order to do what they do? The problem is that we cannot directly observe their internal organization, nor can we specify the mechanisms of coordination by which representational state is propagated. I will assume that a principal role of the individuals in this setting is providing the internal structures that are required to get the external structures into coordination with one another. description of the subtasks, list of tasks

p157: DISTRIBUTED COGNITION THIRD LEVEL OF DESCRIPTION: who does what, who manipulates which artifacts/representations described in the 2 last levels. MANIPULATION OF THE TOOLs BY AN INDIVIDUAL

The other step Hutchins proposes is the computer model...