Architectural technical debt of multiagent systems development platforms

Technical debt is candidate to be the next buzzword in software engineering, and the number of studies evaluating the technical debt of software projects is increasing. A particular and dangerous type of debt is the architectural debt, i.e., the consequences of sub-optimal design decisions. Currently, there are no studies about the evaluation of architectural debt in MultiAgent Systems (MAS) and platforms. Hence, in this paper we propose the analysis of four well-known MAS development platforms, with the aim of evaluating their architectural debt and open the discussion in this field. We exploit a tool, named Arcan, developed for architectural smell detection and for the computation of an architectural debt index. The results show that MAS development platforms are subjected to architectural debt, and in particular to the presence of Cyclic Dependency smells. However, there is evidence that the minimum amount of debt is reached when developers report “bug fixes” and “Improvements”.