Conversation about HCI interdisciplinarity, its ‘core’, pedagogy and being kicked around by technology

This post is an attempt to summarise a conversation between Stuart Reeves and Henrik Korsgaard that took place in Aarhus, Denmark, on the 2nd of November 2022. Our discussion was broadly animated by a couple of concerns. Firstly, the long-running conversation in HCI about its ‘status’ (as a field, as a discipline, as a respectable pursuit within the academic superstructure) and its constitution (as an un-/non-discipline and as concerns its interdisciplinarity). And secondly, the overall feeling of what it’s like being ‘kicked around’ by technology, which we sensed much of HCI (and other fields) have experienced for a long time (including ourselves). The conversation is depicted here as a series of turns between Henrik and Stuart but this is mainly for the purposes of presentation. In reality, Henrik initiated some notes, and Stuart then provided further commentary on the notes.

Henrik: What do we mean by theory and why do we need it? There are two challenges here. First, pedagogy (the second being collective memory within our field, I will return to that). What do you teach from HCI research within a CS program? While there are many available textbooks, they are often design and technology centric and do not constrain a strain or common core of HCI theories. I jokingly say that there are more pictures of technology in our textbooks than references to evidence-based research and theory. Other fields have classics that present how the discipline or field developed and used to think about particular matters. They are still taught either as an introduction to more complex ideas or as part of becoming a member of a particular discipline. A typical example is that we are still teaching Newtonian ideas as partly introductory (because they are accurate enough for most people) and as understanding the history of physics, before progressing to more complex stuff. We also discussed parallels to psychology that also engage with old ideas in classics (see Green 1997) as a background to understanding traditions and more recent ideas. Thomas Kuhn (1970) talks a lot about textbooks as a way to understand current paradigm of a particular field, as a source of authority, and when there is a paradigm shift, they retain some of the history of the field and a sense of discipline:

“Textbooks thus begin by truncating the scientist’s sense of his discipline’s history and then proceed to supply a substitute for what they have eliminated. Characteristically, textbooks of science contain just a bit of history, either in an introductory chapter or, more often, in scattered references to the great heroes of an earlier age. From such references both students and professionals come to feel like participants in a long-standing historical tradition.” (Kuhn 1970, p.137)

I guess, what I am missing here is not only a technology-centric history and/or tradition (remember, Grudin’s “The Computer Reaches Out” (1990) is about interfaces and not the theories and concepts used to make sense of these changes).

Stuart: We started this discussion initially because of your immediate practical concerns about teaching. I think I faced similar fundamental problems (and reflections on ways to move forwards with these problems) when teaching an introductory HCI module some time ago. Teaching HCI leads you to question how the mass of complexity of a whole research field’s work gets packaged up in a ‘textbook’ way for students. This then makes you wonder “what is at the core of HCI and how stable, substantial, and consistent is it?”. You think “why aren’t there these books that everyone knows in common which I can point to as eternal features of a shared intellectual landscape?”. The first question that springs to mind is whether the ‘textbook’ view of research is similar to or representative of the lived experience of it as part of a community of researchers? Are we fooling ourselves into dreaming of a thing that never was? It is true that there are probably fewer ‘classic texts’ which can provide the level of ongoing intellectual nourishment that other fields benefit from. But maybe this is a function of the technology driving things rather than ideas driving things? The second thing that this conversation here assumes (or at least doesn’t draw attention to) a certain intellectual arrangement of universities, institutions etc. that have a history of specific ways of interrogating and arranging things like ‘disciplines’. HCI and its pedagogy, its forebears, the institutional and cultural situation it finds itself in is a product of that environment and the cultures it has emerged from (many of which are rightly being challenged today). The third thing to mention is that SIGCHI has had an Education Task Force for some time, and EduCHI is now a thing. There’s been prior work here e.g., started by Jenny Preece, Anne Bowser and Elizabeth Churchill some time ago. I’m not sure how this conversation connects with that (yet). Finally I want to say that the go-to analogy is physics and we reach for it immediately here. I don’t think you are implying that there is somehow some commensurability, but rather the general perceived sense of ‘stability’ is an important component of the trouble sensed with HCI. I characterise this as ‘trouble’ but we should remember that trouble is not necessarily there to be ‘solved’ (see Reeves & Beck, 2019).

You also mention discipline. But there are many basic things which HCI researchers do not agree upon or align on. To me the only points of agreement (and maybe even this is wrong) is that HCI is ‘about’ technology (usually interactive in some way) and design. The rest is up for discussion. Even notions about ‘interactivity’ are themselves in question. Further, as you point out, the history of HCI shapes both its biases, its homes, and its relationships to pedagogy (and cultural differences therein). In the UK, my understanding is that HCI is mostly (mostly!) in a strong relationship with computer science, psychology, human factors, and perhaps also mathematics. There are few historic routes in the UK to HCI that I know of which have come via communication studies, for example, that we see elsewhere like the USA. Maybe Europe is the same? This then informs the shape of HCI in the UK, its recruitment, its structure, organisation, funding models, outcomes, etc. These are choices that have been made bit by bit and yet their broader import and cultural significance with respect to ‘the discipline’ is often not foregrounded or remarked upon much.

Henrik: We touched upon recent discussions circulating around the lack of common motor themes in HCI (Kostakos 2015; Liu et al. 2014) and tensions and challenges from being an interdisciplinary field (Reeves 2015; Blackwell 2015). The keyword analysis of Liu et al. (2014) — and relatedly, Kostakos (2015) — and the reported the lack of “motor themes” may be methodologically misdirected, but they point to something that is missing in HCI research (see also Oulasvirta & Hornbæk 2016). What the missing component is, is open. Even the question whether HCI needs some common core or motor themes is debatable. I guess what I feel is missing (and it is a feeling and experience, not something I can firmly back up) is something enduring whether we call it classics, motor themes (based on something else than keywords), theory, concepts, canonical examples for textbooks (again, Kuhn would say that a paradigm in textbooks do not separate the theory from the canonical examples from the canonical exercises). Here, I come back to the question of theory. To me, theory plays an important role as a field or disciplines collective memory. It saves the phenomenon between the concrete and the abstract. It helps us avoid the ‘nomadic’ life Kostakos presents. And it can help us repeat work, perpetually starting from first principles etc.

Stuart: I dislike the scientism of Kostakos and Liu et al. and the method; there are fairly obvious possible ‘cores’ (subcommunities, shared interests, common orientations, etc.) one could easily point to, simply by being a researcher involved in HCI for enough time. Pretending that we cannot do that unless it is part of a ‘scientific analysis’ is to my mind extremely limiting. I also think that the notion of ‘accumulation’ — unstated but nevertheless runs through our discussion — needs to be reflected upon as a ‘science norm’ which may be based on a specific ‘picture’ of research activity (see Feyerabend (1975), and Kuhn of course). And being reflective scholars we should be cognisant of the manner of our investigations too (the big E word, i.e., ‘epistemology’). Thus, treating HCI unreflectively as compared with a picture of ‘normal science’ isn’t really a useful way to think about what HCI could possibly be / be doing. That said, the ‘trouble’ Kostakos and many others seem to feel is nevertheless genuinely felt even if the mode of dealing with it is — in my mind — misguided. I’d also add that the whole discourse on this ‘meta HCI’ stuff — how we each think about our research and what doesn’t get put into papers — is nevertheless vital to properly understand HCI. This stuff is not in the research record. These are more like ‘pub debates’ i.e., things left for the interstitial moments of research life — after all, research is communities of practice too. To not acknowledge that is to ignore a huge chunk of the mundane reality of research activity, just as it is to ignore its many political or cultural aspects.

You suggest that theory may be described as collective memory. It may be that HCI does actually ‘do’ what people would comfortably call ‘theory’, yet does not label it as such. Thus by foregrounding particular things (application, domain, technology, etc.) we may be missing what are actually aspects that do act as a ‘core’ for some people (i.e., reframing things dissolves the perceived ‘problem’). This goes back to HCI’s attitudes towards technology — and the tail wagging the dog. HCI’s beholdenness to tech may disguise parts of its collective programme that are actually substantially contributing to ‘theory’, thus also feeding into the concerns about ‘theory’ and ‘cores’. Finally, I note that some years ago there were increased challenges to the role of corporations at CHI, such as industry sponsorship and so on. I wonder how the specific US (and European) biases of CHI and the structures, organisations of things that this bias has maybe promoted contribute to a kind of atmosphere where the ‘core’ becomes a problem. Here we loop back to a critique familiar to us all of which focus on specific ways of knowing, of which a concern for ‘theory’ features heavily.

Henrik: We often have to repeat and account for aspects of methodology, epistemology, traditions and theoretical orientations that we feel should be somewhat established now, 50 some years into HCI research. There is a sense that, given the strong commitment to design (implications) and informing HCI practitioners, years of insights remain local to particular topics and individual studies. This may also be unevenly distributed, in the sense that user studies and laboratory experiments feel more established than for instance ethnomethodological work and theoretical work.

Stuart: As Taylor (2015) has argued, there is a particular “default mode” of HCI which he calls “Engelbartian interaction”. This means some things are left unexplicated because that is just ‘normal HCI’. Other approaches have to be constantly justified because they breach various unstated norms and taken-for-grantedness. We could connect this back to our discussion about the history of HCI, geographic distribution, etc., and how that particular path or set of paths taken by different communities has led us to this point. At this moment since you mention it we should also talk practice and practitioners, particularly HCI’s relationship to ‘them’ (as a monolith) and industry. Often I think industry doesn’t care much about HCI even though HCI wants its love to be sent to ‘the practitioners’. This relates back to your point about ‘classic texts’ and pedagogy. Our impacts in HCI are very heavily on practitioners but probably not in any direct sense HCI typically conceives as ‘for practitioners’. We collectively train many of them in our degree and masters programmes — this is real, but long-term impact which takes career lengths to actually come to fruition. If we teach them in a way that renders HCI research inchoate and (for want of a better word) ‘scatty’, or subject to the whims of technology and its market infrastructures, then what long-term, deeper orientations are we leaving them with for the rest of their lives when things no doubt change? Will we just leave them chasing surfaces — vagaries of technologies — and not underlying concepts?

Henrik: So, back to that which may be missing. We cannot pinpoint directly what is missing, but some enduring constructs, synthesis work and a more grounded understanding of the full range of HCI methodology seem to be what the conversation lingers around. I suggest that theory should step into this space as something enduring (that is one of the roles theory has played in other fields). It is striking that contemporary contributions still lament that the field cannot account for basic phenomena (Hornbæk & Oulasvirta 2017) or the state of these constructs (Reeves & Beck 2019). Again, the missing bit or hole in HCI research creates trouble for the field. Internally, because it is hard to synthesise work into teachable constructs, compare decades of effects proven in the lab, themes from the field and ideas. When other areas look to HCI research for answers, models and principles, it is unclear to us where they should look and what they will find.

Stuart: You raise a point about appearances, not only practical problems introduced by the needs of pedagogy (and the specific structures / arrangements / assumptions within that sits, situated within — in our limited experience — (specific) European contexts) — but also issues of respectability, how HCI ‘looks’ to others outside of HCI. It should be the case that we can openly talk about how these typically unvoiced concerns may play a role in the discussion here, and the drives towards producing what you call “the enduring”.

Henrik: Yes, and not only respectability. Depending on how one counts, HCI is a field with 50+ history and work (I typically count from Licklider (1960), because I like the paper. Others might count from the first CHI conference). So when other areas encounter challenges, they may look to us for answers — theories, constructs, guidance. There are many contemporary challenges where we face problems that HCI and CSCW should have some textbook answers to. The massive and hastily migration to remote work under COVID19 for instance. Where are the synthesis insights, models, recommendations, theories for designers and engineers at Zoom, Gathertown, Discord, Slack, Miro to scrutinise to develop better support for everything remote? People working on remote work prior to COVID19 often had problems convincing reviewers of the relevance.

Or with the increased trend of having prompt-based command line interactions with ML models (so-called prompt-based learning), where is the synthesis (and continued) work on command line interaction? Similarly, work on Human-AI collaboration in Nature seems to have ignored both HCI and CSCW (e.g. 2021, 2022). Maybe they had a quick look and concluded they were better off starting from scratch. Or when Infovis started to look at interaction as important, it seems like they had to do a first-principle approach instead of learning from HCI (see Yi et al. 2007). And what about newcomers to HCI? If you cannot easily get ‘some’ overview over ‘some’ parts of HCI without having to read Grudin, Bødker, Rogers, Shneiderman and multiple others in parallel, how do you contribute? Or rather, when Rogers says “anything goes in HCI research” (Rogers 2012, p. xii), then it’s hardly surprising that people skip the classics to go for the quick contribution.

While it is difficult to confidently and conclusively pin-point that something specific is missing, it is much easier to examine some of the reasons for the absence of more enduring constructs or commonly known HCI classics.

Stuart: On this point, there are papers which do try to do ‘meta theory’ or construct ‘meta taxonomies’ to arrange HCI work in response to these concerns about “enduring constructs”. Papers immediately that spring to mind are Van Turnhout et al. (2019), Stolterman and Beck (2022), and various papers by Hornbaek & Oulasvirta. Sometimes HCI textbooks try to do this implicitly by virtue of their way of arranging HCI into something ‘readable’. But the central problem facing this task is that of incommensurability. In other words, I think HCI is just the community interface where we share ideas, but owing to HCI’s non-disciplinary nature, the ways we come to those ideas cannot come under some shared schema no matter how ‘meta’. That said, if we re-frame efforts to construct a ‘meta theory’ of HCI into something seemingly less grand but far more practical like ‘a tourism guidebook’ then we may benefit from the things which that analogy entails. Guidebooks try to be accurate but they do not present a grand theory of a city or other tourist destination. Guidebooks get out of date, but they still retain value, and even may provide historical significance some time later. They contain impressions and opinions on ways to engage with the location of interest, but they are not the only way to get value. They try to boil down complexity into something manageable. That said, the notion of the ‘HCI guidebook’ is still a bit different to your idea of ‘classics’.

Henrik: I like the idea of a HCI guidebook for now. I do believe that there is a space for more theory and more pluralistic theory. And your point on incommensurability is key here. Right now I think that the HCI theory space suffers more from meaning variance and lack of well-described positions and constructs.. Once some of this synthesis and theorising work develop more clearly, then, we can explore common philosophy of science issues as they face us. Incommensurabilities and other issues feel like a downstream topic and an entirely different conversation.

Henrik: Kostakos (2015) identifies the strong focus on implications for design as problematic. We share that focusing on ‘HCI practitioners’ as a primary audience for direct research results is perhaps a bit of a misunderstanding in the field. Typically, one would expect that practitioners of a field get their knowledge and ideas from textbooks and lectures, and not through shifting through CHI proceedings for ‘strong concepts’ and ‘implications for design’. If we rely (only) on synthesising our knowledge as implications for design, these insights will suffer in multiple ways:

• They will likely remain local to particular research publications
• We have no tradition of or systematic ways of comparing insights (like other fields compare results, models and theories)
• They are likely (seen as) valid only to particular technologies or classes of technologies.
• And unless we are mistaken and the main readership of CHI proceedings is busy HCI folks in industry, then the most recognised epistemological construct in HCI research does not have an audience

This brings us to a final summary of the sense of something missing and/or why something is missing. We largely feel that HCI research is getting kicked around by technology. This is perhaps Kostakos’s point as well, but his analogy with nomads feel inadequate to our experiences. For the inventive part of HCI research, new technologies may bring excitement and opportunities for riding a new wave of novelty, relevance, funding and contributions. For us, the experience is different. One does not have to read that much outside HCI before the idea that ‘new’ technologies are rarely new (history doesn’t repeat itself, but it rhymes; history of infrastructure and bureaucracy and computing and all that jazz). Except for a few radical inventions, technology is only new when deliberately positioned in an “we live in unprecedented times” / epochalism narrative (Morozov 2013). Hence, the fascination with the ‘new’ often becomes an instance of having to relate to the ‘new’ in our own publications, funding proposals, institutions etc., while we at the same time must resist the temptation to critiquing the new, which distracts us from our own research.

The way we put technology front and centre in publication titles is another way we are being kicked around by technology. (And this is perhaps where Kostakos’s keyword analysis is hitting the actual problem.) Putting specific technologies (social media), specific instances of technologies (Facebook) or the catchy names of our own inventions in paper titles seem counterproductive to developing enduring constructs and classics that can speak to more than the latest technologies. We are hiding a lot of the contributions to our collective knowledge (whatever that is) in the deep end of our publications. How can you pick up an idea, theory, concept, implication etc. and engage productively with it, if finding things that speak to each other is almost impossible?

Stuart: To that I would add that while there are strong pockets of work within HCI that are definitely ‘kicking back’ at technology, almost everything seems to be working against us doing this. There is only a relatively small amount of work questioning the largely corporate-driven and individualist basis of most technologies that get play in HCI. I can point the finger directly at myself included in this regard — recently we have had a number of relatively successful studies of voice interfaces (e.g., Porcheron et al. 2018) — but you could quite reasonably critique this style of work as essentially glorified (albeit sceptical) product research. Much HCI is like this, unquestioning of the technical fundamentals which underpin and form core assumptions, even when this work is designated ‘critical’. Maybe I shouldn’t be so hard on this work — it is not unreasonable to explore the objects, infrastructures, systems in the world as you find them. Social computing has done a roaring trade in that. But, alternatives, different arrangements of and conceptions of technology seem to be rumbling there under the surface all the time. By that I mean state-, civic-, or community-oriented technologies — although considered in HCI’s body of work — nevertheless have this extant substrate to work with (devices, infrastructures), the assumptions of which — largely privatised, and individualised — bring their own difficulties when retooling for these alternate purposes.

Henrik: I would like to make a final (post-conversation) note on some of these issues. It doesn’t have to be this way. There are multiple open questions and ambitions to work toward. I do not think that theory should replace anything, but I would also like to be able to work just on HCI theory as a more creative and indigenous genre of HCI contributions. CHI puts up a high bar for theory contributions. A high bar that does not leave room for creativity and experimentation with indigenous theory. Similarly, it is not given that HCI should maintain its fascination and uncritical stance toward industry or the very idea that HCI practitioners and companies are the sole or primary audience for HCI research. Finally, briefly returning to the question of discipline, interdiscipline or speciality. This is not given either. I think it is important to remember that when we give HCI attention in our educational programs and courses, we are also bringing up a new generation of researchers who may feel that this is more a pluralistic (inter)discipline than a meeting between multiple disciplines.