Paper short abstract:

Characterizing scholarly software development as a knowledge practice, this paper presents forms of code description and analysis through which computational physicists and digital humanities researchers negotiate technical constraints and scholarly commitments.

Paper long abstract:

How and when is software development an epistemological issue in digital scholarship? In both the sciences and the humanities, strategies for designing, building, and maintaining software are increasingly recognized as supportive of, or detrimental to, the knowledge claims derived from computational processes. Software development is typically conceptualized as the practical matter of implementing specifiable computational tools, not as an iterative process of materializing theoretical and epistemological orientations. Building on understandings of software development as an ongoing, human-centered practice, on the one hand, and theories of scholarly work that emphasize the materialization of knowledge objects, on the other, this paper presents software development as a knowledge practice. Particular attention is given to forms of description and analysis through which aspects of code are linked to specific scholarly commitments. While some practices for representing and understanding scholarly software are derived from the tradition of software engineering, others are endogenous to specific academic communities. Examples from case studies of two academic communities---computational physicists developing plasma simulations and digital humanities scholars building web-based publishing tools---are presented. Ethnographic fieldwork and technical histories of observed software practices are used to understand how scholarly and technical commitments are negotiated and sustained during the development and maintenance of software at each site.