A small social history of software architecture

Abstract

This paper presents an analysis of software architecture as social artifact, that is, something that software developers talk about and use in their work. This analysis is historical in nature, relying on interviews with software developers with experience spanning four decades and the software engineering literature. We found that 1) only large teams have architecture; 2) architecture is more easily found in discourse than in source; and 3) architecture does not happen at a fixed time in the software lifecycle. These observations taken together suggest that software is a boundary object that developers use to explain the system to each other, thereby making it possible to work together.