In the fast-growing Internet of Things (IoT) industry, algorithmic technology promises 'smart' solutions to everyday problems. Drawing on a design research investigation, this chapter questions and critically examines the embedded epistemology of IoT, in the context of what I call the algorithmic paradigm. This examination reconsiders the prevailing epistemology and offers an alternative based on a second-order cybernetics perspective. This alternative recognises the importance of accounting for the role/agency of the observer/designer/user, the circular causality of user behaviour and technology, and the relationality of 'smartness'. To explore the possibility of a shift in perspective from the current algorithmic paradigm to a second-order one, users are approached as experiential, non-linear subjects rather than as probabilistic and linear ones. Outcomes reveal the value of second-order cybernetics as an epistemological stance and a practical approach to research on the design of 'smart' interactions. The methodological framework demonstrates how design research and secondorder considerations can work together, asking novel questions to inform disciplines with an interest in IoT interactions, from both a design perspective (the way designers approach their practice) and in terms of broader implications for society. Keywords 'Smart', AI, Home, Determinism, Reductionism, Context implication that it is a commodity that can be stored in a computer, dismissing its significance as a relational concept (both human and algorithm are responsible for the 'smartness' of the interaction). In order to reinforce and highlight the complexity of the word, I will place 'smart' in quotation marks. I will also distinguish lower and upper case -upper case 'SMART' indicates that I am contesting the word in my research projects.2 In statistical terms, Gartner [5] has estimated that 25 billion connected 'things' will be in use by 2020, while Cisco projected that in 2020, 50 billion objects would be connected [2]. The disparity in the figures does not necessarily reflect a downward revision after a three-year gap; rather, it signifies how quickly the field is emerging and the uncertainty that comes with that speed. 6 The term 'user' will be used to describe the human component in a human-IoT interaction. In this chapter it has no commercial bias; 'user' is interpreted as a complex human instead of a passive consumer. Here, complexity is defined as "the quality of being complex', i.e.,"having many varied interrelated parts, patterns, or elements and consequently hard to understand fully", or "marked by an involvement of many parts, aspects, details, notions, and necessitating earnest study or examination to understand or cope with." See George Klir's definition from Webster's Third International Dictionary in the International Encyclopedia of Systems and Cybernetics [4].