HCI as a profession that is still deciding what it is

Ask three colleagues what human-computer interaction is and the answers disagree. The engineer describes the part of the job that keeps software usable enough to ship. The researcher describes the study of how people and computers work together, with its own standards of evidence. The designer describes the discipline that decides what should be built at all. None of them is wrong. Each answer commits its speaker to a different account of good work. Harrison, Tatar and Sengers (2007) call such commitments paradigms. A paradigm is a world view that carries its own practices, expectations and criteria for what counts as knowledge. They argue their field holds three. Human factors optimises the fit between person and machine. Classical cognitivism models the flow of information between computer and mind. A phenomenologically situated paradigm treats interaction as embodied, situated action. Naming them, as this first of ten units does, lets work under one paradigm be valued rather than dismissed by another.

Harrison and colleagues make the split concrete with a single interface. In the 1960s the United States Air Force built automated cockpit warnings that used recorded voices to tell pilots to turn, climb or dive. Human factors supplied the first move. Every pilot and controller was then a man, and the operator the fit was designed for was assumed to be able-bodied and of working age. A woman's voice therefore marked the message as urgent, and it cut pilot errors. Cognitivism supplied the second. Familiar voices, a wife's or a girlfriend's, improved performance further. Familiarity is a variable in the information channel. The third paradigm supplied the objection. The first voice was reputed to have been chosen for its seductive tone. That choice made the cockpit read as male space once women flew. The example is the authors' own illustration rather than a study: it shows the paradigms generate different moves and criteria, not that one produced the better system.

The oldest dispute is whether HCI is a science or a service to engineering

Newell and Card (1985) asked whether psychological science could play a significant part in HCI. They observed that hard science tends to drive out softer science that still has something important to contribute. They analysed four objections, about level, scope, timing and applicability.

Carroll (1997) describes the answer the field gave. HCI is the intersection of psychology and the social sciences with computer science. Over two decades it integrated its scientific concerns with the engineering goal of improving usability. One career can therefore hold a scientific question and an engineering obligation at once.

The engineering conception acquired an apparatus. Nielsen's usability text is organised around a usability engineering lifecycle, with a chapter of that name (Nielsen, 1993). That book codified usability work as a process with stages a project could be planned around. That reading rests on the chapter's title and place rather than its text. A team that plans usability as lifecycle stages asks for estimates and criteria. A team that runs the arrangement Bannon (1995) describes asks for specifications and a final check.

The waves are not the paradigms

Two other three-part accounts are often confused with paradigms. Grudin (1989) traces forty years of interface research and identifies five foci as the interface moves outward, from hardware through software to higher-level cognition and finally to social processes. Bodker (2006) asks how second-wave theory, built around work settings, survives a third wave in which technology spreads into homes and everyday life. Both are historical claims about a widening remit. Harrison and colleagues argue they leave the pervasiveness of intellectual frameworks unaddressed. Mistake a wave for a paradigm and an older conception looks superseded when it may still be running in the next room.

Human actors, not human factors

Bannon (1995) shows what the widening remit did to the practitioner's position. Human factors, he writes, is often incorporated simply as specifications to which the delivered system must adhere. The specialist does task analyses and perhaps some interface retouching near the end, once the design is fixed. The role has therefore been seen as ancillary to building the system. He argues it should instead be fluid and pragmatic, with input when capabilities are discussed, during prototyping and in evaluation. He argues that development requires distinct communities characterised by an essential incommensurability of their world views. The paradigm paper quotes that point back (Bannon, 1995, quoted in Harrison, Tatar and Sengers, 2007).

Design claims its own foundations, and a rival lineage contests them

Fallman (2003) argues that HCI has emerged as a design-oriented field directed at innovation and the construction of new technology while its underpinnings remain poorly understood. He examines three accounts of what design is: the conservative, the romantic and the pragmatic. Through sketching, he shows that the designer works the problem and its solution out together as a coupled pair. He also distinguishes knowledge-generating design-oriented research from artefact-generating research-oriented design. Cross (2007) supplies the foundation, describing a shift in design research from a design science to a design discipline. Dorst (2011) supplies the mechanism. If a designer looks at a problem situation from a particular viewpoint, and adopts the working principle associated with that position, then they create a specific value. The core of design thinking is reasoning for problems that do not yet exist, not a method that travels with the sticky notes.

Dourish (2001) notes that computer science is an engineering discipline. It is also a philosophical enterprise in how it represents the world and models people and action. He develops embodied interaction, in which meaning and action rest on skilled, engaged practice rather than disembodied rationality. A misread interface is an information problem to the cognitivist and a failure of meaning to this lineage.

A team's paradigm shows in its artefacts before its vocabulary

Cooper (1999) describes the condition the design lineage reacted to. The mechanics of building an application take precedence over the aims of the project. As a result, user, designer, programmer and manager alike end up without what they want, and the industry inadvertently put programmers and engineers in charge. A usability test at the end of a build and no research before it means human factors in Bannon's qualified sense. A plan of lifecycle stages with criteria means the usability engineering conception. Arguments about what the product should be rather than whether it works mean the design lineage is in charge. A finding disputed not on its quality but on whether it counts as evidence at all means the clash Harrison and colleagues describe. Reading which is in play, before proposing anything, is the skill this unit teaches.

Contested question: one discipline with a shared core, or several paradigms sharing a name?

Harrison, Tatar and Sengers (2007) hold that the paradigms are separated by different epistemologies, so what counts as legitimate knowledge differs. Measurable utility governs the first, controlled contrastive demonstration the second, and situated interpretation the third. HCI has always been a hybrid discipline using the intersection or union of its constituents' legitimate practices. Third-paradigm work therefore tends to be held to first and second paradigm standards.

The opposing position treats the field as one enterprise with a maturing repertoire. Carroll (1997) describes a cumulative integration of science with the engineering goal of usability. Grudin (1989) describes a continuous outward movement in which required skills change. Bodker (2006) describes earlier concepts that remain usable if repositioned. On this reading, paradigm talk relabels ordinary methodological breadth as incommensurability.

Rogers (2012) sharpens the dispute without resolving it. She charts HCI theory from a nascent field that borrowed theories from cognitive science. She argues HCI is an eclectic interdiscipline rather than a well-defined science. On that reading the paradigms look less like world views in conflict than inherited borrowings put to different purposes.

What would settle the disagreement is a comparison of consequences rather than of definitions. If the paradigms are incommensurable as claimed, two teams working the same problem under different paradigms should diverge predictably in what they measure, accept and build. The divergence should be derivable from the paradigm rather than the people. If the unity account is right, comparable projects should show a shared vocabulary, local method variation and no systematic difference in decisions. The literature contains no such comparison. The weaker test, which asks whether published evaluations would satisfy all three legitimacy standards, remains unrun.

The two sides are unevenly supported. The paradigm case rests on one cockpit anecdote rather than a comparison of projects. The unity case is historical reviews by Carroll, Grudin and Bodker reading the same record as cumulative rather than divided. Neither is an experiment. While the dispute is open, a designer should name the paradigm the brief is read under and the criteria that follow from it. The strongest objection is that the debate is academic: employers reward shipped software, and a graduate who argues paradigms instead of delivering may be the one replaced.

Where the argument is heading

Rogers (2012) closes by asking whether theory will continue to have a role in HCI and what it might be. She points towards a more design-oriented relationship between theory and practice, rather than further assimilation from elsewhere. Bodker (2006) points at third-wave challenges for which second-wave concepts must be repositioned. Fallman (2003) points at the difference between generating knowledge about design and generating designs. The canon runs from 1985 to 2012, but the identity question has not closed. Sharma, Kumar and Nardi (2023) argue that HCI has absorbed an assumption of economic growth without examining it, turning its own self-image into an object of critique.

The sources this unit rests on

• Bannon, L. J. (1995). From Human Factors to Human Actors. Readings in Human-Computer Interaction. doi:10.1016/b978-0-08-051574-8.50024-8 Contrasts human factors bolted on as specifications near the end of development with people as actors whose work and tools develop together. Argues the specialist's role should be fluid and pragmatic rather than ancillary; the practitioner's place is negotiated, not given.
• Bodker, S. (2006). When second wave HCI meets third wave challenges. Nordic Conference on Human-computer interaction. doi:10.1145/1182475.1182476 Surveys second-wave HCI theory, built around work settings, against a third wave in which technology spreads into homes and everyday life, working through multiplicity, context, boundaries, experience and participation. Shows the object of study widening, so competence is defined against a moving target.
• Carroll, J. M. (1997). Human-Computer Interaction: Psychology as a Science of Design. Annual Review of Psychology. doi:10.1146/annurev.psych.48.1.61 Reviews HCI as the intersection of psychology and the social sciences with computer science. Finds the field integrated its scientific concerns with the engineering goal of improving usability, so one field can be applied science and design work at once.
• Cooper, A. (1999). The Inmates are Running the Asylum. Berichte des German Chapter of the ACM. doi:10.1007/978-3-322-99786-9_1 Argues the mechanics of building an application take precedence over the aims of the project, so user, designer, programmer and manager alike end up without what they want. Proposes goal-directed design; the unit's example of the practitioner-polemic genre.
• Cross, N. (2007). From a Design Science to a Design Discipline. Design Research Now. doi:10.1007/978-3-7643-8472-2_3 Describes a shift in design research from a design science to a design discipline, with attention moving to design cognition and designerly ways of knowing. Carries the claim that design knowledge is a distinct kind.
• Dorst, K. (2011). The core of 'design thinking' and its application. Design Studies. doi:10.1016/j.destud.2011.07.006 Argues design thinking became known for its practices and tools while its core was left unstated, and locates that core in framing and frame creation. The unit's sharpest statement that design competence is a reasoning ability rather than a toolset.
• Dourish, P. (2001). Where the Action Is. The MIT Press. doi:10.7551/mitpress/7221.001.0001 Sets out the philosophical bases of HCI, noting computer science is an engineering discipline and also a philosophical enterprise in how it models reality, people and action. Develops embodied interaction, in which interacting rests on skilled, engaged practice rather than disembodied rationality.
• Fallman, D. (2003). Design-oriented human-computer interaction. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. doi:10.1145/642611.642652 Argues HCI has emerged as a design-oriented field while its underpinnings remain poorly understood, and finds through sketching that problem and solution are worked out together as a coupled pair. Distinguishes design-oriented research from research-oriented design.
• Grudin, J. (1989). The Computer Reaches out: The Historical Continuity of Interface Design. DAIMI Report Series. doi:10.7146/dpb.v18i299.6693 Traces forty years of interface research and identifies five foci, from hardware through software to higher-level cognition and finally to social processes. Notes that the skills required of practitioners change as the focus shifts, so the remit widened repeatedly.
• Harrison, S., Tatar, D. and Sengers, P. (2007). The three paradigms of HCI. Alt. Chi, CHI 2007. http://people.cs.vt.edu/~srh/Downloads/HCIJournalTheThreeParadigmsofHCI.pdf Argues HCI contains three paradigms, human factors, classical cognitivism and a third, phenomenologically situated paradigm, each a world view with its own criteria for what constitutes knowledge. States the authors are not claiming the third is right and the others wrong, only that naming paradigms explains why work in one looks invalid in another.
• Newell, A. and Card, S. K. (1985). The Prospects for Psychological Science in Human-Computer Interaction. Human-Computer Interaction. doi:10.1207/s15327051hci0103_1 Asks whether psychology can play a significant role in HCI, noting hard sciences tend to drive out softer ones even where the softer have important contributions to make. Analyses four objections about level, scope, timing and applicability; the unit's oldest professional anxiety.
• Nielsen, J. (1993). The Usability Engineering Lifecycle. Usability Engineering. doi:10.1016/b978-0-08-052029-2.50007-3 Places usability work inside a lifecycle account of engineering, so usability was codified as a process and its specialist became an engineer of usable systems. The unit's marker of the engineering self-image; the note rests on the record and title, not the text.
• Rogers, Y. (2012). HCI Theory: Classical, Modern, and Contemporary. Synthesis Lectures series. doi:10.1007/978-3-031-02197-8 Charts HCI's theory from a nascent field that borrowed theories from cognitive science, asking what theory promised, how it was used and which made the most impact. Argues HCI is an eclectic interdiscipline rather than a well-defined science.
• Sharma, V., Kumar, N. and Nardi, B. (2023). Post-growth human-computer interaction. ACM Transactions on Computer-Human Interaction. doi:10.1145/3624981 Argues that HCI has absorbed an assumption of economic growth without examining it, and asks what a post-growth orientation would change in the field. Turns the profession's self-image into an object of critique rather than a settled inheritance.