Failures Supporting the Evolutionary Design in the Wild of Interactive Systems for Public Spaces Vinicius Ferreira ( ✉ ) , Junia Anacleto, and Andre Bueno Federal University of São Carlos, São Carlos, Brazil {vinicius.ferreira,junia,andre.obueno}@dc.ufscar.br Abstract. In this paper, we describe the lessons learned from an experience of deploying an interactive public installation adopting a methodology that inter‐ twines aspects of in-the-wild study and evolutionary design. This methodology shrinks the cycle of design of a prototype and allows researchers and practitioners continuously design improvements while they learn from evaluating the prototype in the wild. Thereby, multiple settings can be explored, minimizing the need to conduct new experiments that demand time and resources. Considering the meta‐ phor of a wishing well, we designed a public interactive installation, allowing people to make wishes using their personal or a shared device to throw a virtual coin into a real water fountain augmented with a wall-sized screen displaying a “pool of wishes.” We deployed this interactive installation in a passageway of building for eight days, collecting data from observations, questionnaires, inter‐ views, photos and video recordings. Based on the failures in our first cycle of design with the prototype, we present our findings and directions to apply our methodology. We claim the need of a pilot study in situ and having a team committed in collecting and analyzing data, discussing the insights and changes. In addition, the system design must support predictable, orderly and managed evolution. Finally, we contribute to the ubicomp field, demonstrating the impli‐ cations for evaluating evolutionary prototype in the wild. Keywords: Prototyping methods · In-the-wild · Evolutionary design · Ubicomp 1 Introduction Interactive technologies are increasingly spreading through the urban environment, moving Human Computer Interaction (HCI) into a new research paradigm. Researchers are shifting from their labs into the real world, developing and deploying pervasive technologies for people [15]. This in-the-wild movement emphasizes performing in situ user studies with new technologies to uncover design probes in real-life context settings [16]. Moreover, this approach differs from previous ethnographic approaches applied to interaction design that propose to observe existing practices to elicit system require‐ ments and general implications for the design [15]. In-the-wild approach focuses on learning from users experimenting and adopting a prototype in situ. From this approach, several insights can emerge that demand modifi‐ cations on the prototype and require new experiments, which consumes resources and © Springer International Publishing AG 2017 M. Kurosu (Ed.): HCI 2017, Part I, LNCS 10271, pp. 283–296, 2017. DOI: 10.1007/978-3-319-58071-5_22