On the interest of architectural technical debt: Uncovering the contagious debt phenomenon

A known problem in large software companies is to balance the prioritization of short‐term and long‐term business goals. As an example, architecture suboptimality (Architectural Technical Debt), incurred to deliver fast, might hinder future feature development. However, some technical debt generates more interest to be paid than other. We conducted a multi‐phase, multiple‐case embedded case study comprehending 9 sites at 6 large international software companies. We have investigated which architectural technical debt items generate more interest , how the interest occurs during software development and which costly extra‐activities are triggered as a result. We presented a taxonomy of the most dangerous items identified during the qualitative investigation and a model of their effects that can be used for prioritization, for further investigation and as a quality model for extracting more precise and context‐specific metrics. We found that some architectural technical debt items are contagious, causing the interest to be not only fixed, but potentially compound, which leads to the hidden growth of interest (possibly exponential). We found important factors to be monitored to refactor the debt before it becomes too costly. Instances of these phenomena need to be identified and stopped before the development reaches a crises.