Hypotheses Engineering: First Essential Steps of Experiment-Driven Software Development

Abstract

Recent studies have proposed the use of experiments to guide software development in order to build features that the user really wants. Some authors argue that this approach represents a new way to develop software that is different from the traditional requirement-driven one. In this position paper, we propose the discipline of Hypotheses Engineering in comparison to Requirements Engineering, highlighting the importance of proper handling hypotheses that guide experiments. We derive a set of practices within this discipline and present how the literature has tackled them up to now. Finally, we propose a set of research questions that could guide future work towards helping practitioners.

Hypotheses Elicitation in Early-Stage Software Startups Based on Cognitive Mapping

“…Another interesting aspect of hypothesis types is that they can act as elements to help in prioritization, an essential aspect of hypotheses engineering [3]. For instance, regarding the types we identified in HyMap, if customers do not feel the problems, it is hard for the product to succeed, and without them, the whole map would not exist.…”

“…The authors of these models analyzed the existing literature and current practices of companies applying experimentation. These models presented the process as cyclical approaches consisted of some steps executed continuously: identify, specify, and prioritize hypotheses, design an experiment, execute it, analyze the results, and update the hypotheses accordingly [3]. Nevertheless, these models do not describe how hypotheses could be systematically identified.…”

“…In a recent position paper [3], we compared different models of experimentation and observed that, at the process beginning, they suggest the team to identify, specify, and prioritize hypotheses. Drawing a parallel to Requirements Engineering activities employed in a requirement-driven approach, we argued the need for a set of practices called Hypotheses Engineering (HE) to identify, specify, and prioritize hypotheses in experimentation.…”

“…One of the reasons for this limited adoption of experimentation is the lack of clearly defined practices [1]. Therefore, an essential step in the direction of a better implementation of experimentation practices is a systematic way to specify and handle hypotheses [3]. In this study, our goal is to develop a novel technique to identify the hypotheses on which early-stage software startups base their products.…”

Towards a Framework to Guide the Creation of Development Practices for Software Startups

“…Hypotheses Engineering [16] is a proposal to support experiment-driven software development, as a parallel to Requirements Engineering in the conventional, requirements-driven software development. Rather than describing features to be implemented, hypotheses define uncertain questions about the business model or other software product aspects that the team could assess with an experiment.…”