Stop Worrying and Love the Mess: Critical Review of Reeves, S. (2015). Human-computer interaction as science.

This paper reflects on Human-Computer Interaction (HCI) and its self-image as a ‘science’. While at times written in a complex style, it has as simple message: Stuart Reeves argues that HCI has two long standing disciplinary anxieties: incoherence and inadequacy. The incoherence problem means that there is no disciplinary order and no uniform body of theory since so many different disciplines have influenced HCI. Inadequacy points to the perception that HCI is not a respected science, in particular in comparison with the ‘model science’ physics, as it lacks its rigour. This argumentation impacts many disputes in HCI along the lines of ‘soft vs. hard science’ or ‘qualitative vs. quantitative methodology’.

The central argument of the paper is that in the formative years of HCI, in the 1980s, design of interactive technology was conceptualised as an optimisation problem, thus making it computational. This is largely due to the strong influence of cognitive science in HCI at the time. This framing, beginning with the notion of a ‘scientific design space’, narrowed what kind of questions and methods one can apply to design. Reeves argues – and I agree with him – that ‘design’ as a scientific activity and science in design as a support got conflated, reducing design to a “system with regular, discoverable laws”. Over the years, a few alternatives of ‘designerly thinking’ discourses emerged, but HCI’s scientism proved quite resilient. This conception is, however, not appropriate, and Reeves rightfully concludes that HCI should stop peering at the natural sciences and instead develop its own notion of rigour, i.e. its own methodology and theory based on the subject of study. Additionally, HCI should stop worrying about becoming a discipline and instead embrace itself as an ‘interdiscipline’.

While I agree with Reeves’ conclusions, I was surprised that such a paper is still necessary in 2015. His points echo much of the philosophy of science’s criticism of the positivistic reductionism, which narrows the field of how scientific knowledge can be produced, and the unjustified continuous dominance of the natural sciences as the model all other sciences are measured against. To question this might be new to HCI, but similar arguments have been made before in many other disciplines, in particular the social sciences. I therefore also disagree in abandoning the term ‘scientific’ in favour of ‘rigorous’. Reeves’ call for HCI as an interdiscipline is also not particularly new. What Reeves describes might be more appropriately called a transdiscipline, as the term better describes the breaking down of disciplinary borders instead of just an exchange of knowledge and methods between disciplines. In any case, inter- or transdisciplinarity is high on the agenda of many research programmes these days. Finally, his criticism of scientific approaches to doing design is left open as he doesn’t give a conclusion on how – in his eyes – design as a creative and problem-solving activity could be supported and directed scientifically using appropriate methods and theories. In summary the issues the paper raises are nevertheless valid and worth discussing.

The paper that exemplifies HCI for me is Brynjarsdottir, H., & Håkansson, M. (2012). Sustainably unpersuaded: How persuasion narrows our vision of sustainability. Proc. CHI ’12, 947–956. While the paper has a specific subtopic of HCI, (persuasion and sustainability) I think it is a good example of how the discourse of how ‘scientific’ HCI is plays out in practice.