Are two heads better than one in CAD? a comparison of various CAD working styles

Abstract

Collaboration in Computer-Aided Design (CAD) has existed since the inception of CAD tools. The established norm in multi-user CAD work has been to use top-down modeling techniques wherein a complex model is divided into sub-assemblies (or part files) for individual designers to work on separately. In this process, designers integrate their work through a check-in/check-out process. This style of collaboration does not change regardless of team sizes, product types, or over time. However, recent cloud-based CAD tools are expected to change this by offering real-time collaboration like Google Docs. In this research, we are interested in learning the effects of real-time collaboration on designers' work. We draw heavily from software development research where dyadic work is common and is known as 'pair programming'. We use an experimental approach to investigate research questions pertaining to speed and quality of real-time collaboration. We found that pair work in CAD is not summative. In other words, the work of two designers does not lead to twice the outcome of individuals. This results is contrary to previous real-time CAD collaboration research but consistent with software programming research. However, we also found that the quality of CAD increases in certain pair CAD settings. We observed that sharing control of the CAD software leads to higher quality and parallelizing work leads to worse quality. To elaborate on our results, we reveal specific patterns of participant behaviour based on audio communication and cursor activity. In summary, we establish foundational knowledge in real-time CAD collaboration research. Through our work, we share insights which can inform practicing engineers that are interested in adopting pair CAD work.