Situated action and the critique of the planning model

A workflow diagram, a procedure manual and a task model share one assumption. They assume that a course of action is the execution of a plan, so that a faithful account of the plan is a faithful account of the work.

This ninth of ten units argues against that assumption, and asks what a plan can explain. Plans, models and procedures are resources people use to organise action rather than causes that determine it. A system that treats a plan as an executable instruction misconstrues the activity it is meant to support.

The planning model treats the plan as the explanation of the action

Suchman (1987) states the target. The planning model treats a course of action as the execution of a plan. Her evidence was a recorded interaction between novice users and an intelligent help system that regulated a photocopier interface, matching what users did against idealised plans. The machine was the Xerox 8200, advertised in the early 1980s with a figure in a lab coat and the promise that pressing the green button was enough. The help system could perceive only the actions that changed the machine's state: doors, buttons and paper trays. The users' reading of the display and their attempts to correct an error were invisible to it. A record of what the machine's state changed cannot see a screen-reader user either. The machine's model of the task was doing the work of interpretation, and the users' difficulty was structural. The resources ordinary communication relies on, a shared setting, a visible history of what has been tried, a chance to repair a misunderstanding, were unavailable to it.

The finding compresses into a slogan that is exact: plans are resources for action rather than its determinants.

The situated claim is about cognition, not only about interfaces

Lave (1988) moves the claim out of the laboratory. Her Adult Math Project compared "best buy" calculations in the supermarket with tests borrowed from school. The same people who solved 98 per cent of the supermarket problems averaged 59 per cent on the school-style test. They averaged 93 per cent on a best-buy simulation run later at home. Twenty-five shoppers took part, predominantly women and older than twenty-one. The test was built from the very best-buy problems they had met in the aisle. So a capability is not a portable possession of the individual. It is a fact about the relation between activity and setting.

Hutchins (1995) shifts the unit of analysis. His study of ship navigation argues that cognitive work is distributed across people, instruments and representations. The right unit is therefore the whole system of which the navigator is a part. The design question then becomes not what a user knows, but where in the system the knowing sits. It also asks what happens when one component is replaced.

When a plan for use meets an actual office

The paperless office is a plan for use mistaken for a description of use. Sellen and Harper (2002) asked why paper survived a confident prediction of its disappearance. Paper, they found, affords activities the replacement systems did not: reading with a pen in hand, annotating in place, moving flexibly among documents, and seeing one's own progress. Their account records an average increase of 40 per cent in an organisation's paper consumption after email was introduced.

People are actors, not components with measurable properties

Bannon (1991) draws the design consequence. Human factors research treats the person in a system as a component with measurable properties, and the situated view does not permit that. They interpret the situation, they hold purposes the task analysis never gave them, and they can take part in designing the systems they will use. Suchman (1993) states the claim at the level of design work: producing and using a technology are mutually constitutive.

Dourish (2001) turns the critique into a programme. Embodied interaction grounds the design of interactive systems in skilled, engaged practice rather than disembodied rationality. It draws on a phenomenological tradition that puts natural practice before abstract cognition.

Carrying an analysis of practice into a design is a problem, not a pipeline

Dourish and Button (1998) ask what it would mean to approach system design from the ethnomethodological perspective. The relationship to examine is whether ethnomethodology can take a foundational place in the very notion of system design. The alternative is to make it a resource in requirements elicitation and specification. Their name for that relation is technomethodology, and they want it to change what design is rather than supply better inputs.

Anderson (1994) makes the critical version of the point. Designers had taken up ethnography mainly as a method for specifying end-user requirements. He argues that most of this interest rests on a misunderstanding, since the designer had redefined an analytic practice as data collection. He proposes an analytic ethnography for HCI that opens possibilities for design rather than closing them into a requirements list. The test of any claim that fieldwork informed a system is which part of the analysis travelled. Was it a finding about how the work is organised, or a list of features others could have collected?

Contested question: does situated analysis become design, and by what route?

The disagreement is not about whether practice matters. It is about whether an analysis of practice can be carried into a design, and by what mechanism, and every position in it rests on argument and case rather than on a trial.

One position says the route is a discipline that can be taught and followed. Randall, Harper and Rouncefield (2007) set out fieldwork for design in a book, and Crabtree, Rouncefield and Tolmie (2012) develop the programme as design ethnography in a methods text. They are explicit about the exit: findings are carried into specification documents, use cases, storyboards, scenarios and prototypes, and checked with the people studied. The evidence for that exit is worked practice, not a comparison of projects that followed it with projects that did not.

A second position holds that this route, as usually travelled, is the misunderstanding. Anderson (1994) argues it in a critique that reports no study, holding that treating ethnography as requirements gathering defines it as data collection and discards its analytic value. A third position says the relation must be foundational rather than procedural. Dourish and Button (1998) want ethnomethodology inside the notion of system design, changing what design is. A fourth steps back to the field. Kuutti and Bannon (2014) argue, in a conference position paper, that HCI now has a practice paradigm alongside its dominant interaction paradigm.

What would settle it is three checks. The first is a trace. Follow one situated finding from the field record of a published project to a decision in the artefact. A programme that supplies such traces has answered Anderson in practice. The second is prediction. An analysis earns the name if it can say in advance what a proposed design will break. The third is the criterion internal to each position: for design ethnography, whether the influence on the design is traceable, for technomethodology, whether the design's own notion of its user changed, for the practice paradigm, whether the field's problems were reframed.

The strongest counter-argument to this unit's own claim is that the critique rests on a small set of well-chosen cases, and that plan-following is itself a describable practice. The evidence is one help system in Suchman (1987) and twenty-five shoppers in Lave (1988), and the claim that plans are resources rather than determinants has not been tested where procedures are followed closely because lives depend on them.

What to do while the dispute is open

Name the practice, cite the record, and name the design decision the record was supposed to constrain. Record what the analysis predicted before the artefact was built. Where a plan is implemented in a system, say which parts of the work the plan describes and which it leaves to be discovered in the setting.

Where the argument is heading

Suchman and colleagues (1999) give the twenty-year programme in three strands: critiques of technical discourse, ethnographies of work and technologies in use, and design interventions. Their premise is that a technology can be assessed only in relation to the sites of its production and use. Suchman (2007) reworks the original argument, shifting the question from the status of human-like machines to how humans and machines are enacted as similar or different in practice. The unit of analysis keeps widening, from the individual problem solver (Lave, 1988) to the system (Hutchins, 1995) to the sociomaterial arrangement (Suchman, 2007). The design question keeps getting harder, because Dourish and Button (1998) left it a foundational problem.

The sources this unit rests on

• Suchman, L. A. (1987). Plans and Situated Actions: The Problem of Human-Machine Communication. Cambridge: Cambridge University Press. https://archive.org/details/planssituatedact0000such The book builds the critique of the planning model on a recorded interaction between novice users and an intelligent photocopier help system that could not supply the resources ordinary communication relies on. It argues that plans are resources for action, not its determinants, and it carries the unit's central claim.
• Suchman, L. (1993). Working relations of technology production and use. Computer Supported Cooperative Work 2(1-2), 21-39. doi:10.1007/bf00749282 Suchman argues that producing and using a technology are mutually constitutive, so design and use cannot be separated into a plan and its execution. The argument is conceptual, and it widens the critique from one system to the relations between designers and users.
• Suchman, L. A. (2007). Human-Machine Reconfigurations: Plans and Situated Actions. 2nd edition. Cambridge: Cambridge University Press. doi:10.1017/cbo9780511808418 The reworked edition reconsiders how agency is configured at the human-machine interface, arguing that the sciences of the artificial obscure the performative nature of both persons and things. It shows the argument extended to how humans and machines are enacted as similar or different in practice.
• Dourish, P. (2001). Where the Action Is: The Foundations of Embodied Interaction. Cambridge, MA: MIT Press. doi:10.7551/mitpress/7221.001.0001 Dourish sets out embodied interaction as a foundation for HCI, grounded in phenomenology and in skilled, engaged practice rather than disembodied rationality, and aimed at the information-processing model of the user. It gives the unit its constructive programme.
• Dourish, P. and Button, G. (1998). On "technomethodology": Foundational relationships between ethnomethodology and system design. Human-Computer Interaction 13(4), 395-432. doi:10.1207/s15327051hci1304_2 The paper asks what it would mean to approach system design from the ethnomethodological perspective, distinguishing the use of ethnomethodology as a resource in requirements work from giving it a foundational place in design. Their name for that relation is technomethodology, and it supplies the unit's sharpest statement of why analysis-to-design is a problem rather than a pipeline.
• Lave, J. (1988). Cognition in Practice: Mind, Mathematics and Culture in Everyday Life. Cambridge: Cambridge University Press. doi:10.1017/cbo9780511609268 Lave's Adult Math Project compared supermarket best-buy arithmetic with school-style tests of the same people, and found performance shaped by the activity and setting rather than by a portable competence. It gives the unit evidence that the situated claim is about cognition, not only about interfaces.
• Hutchins, E. (1995). Cognition in the Wild. Cambridge, MA: MIT Press. doi:10.7551/mitpress/1881.001.0001 Hutchins analyses ship navigation to argue that cognitive work is distributed across people, instruments and representations, so the unit of analysis is the whole system rather than the individual mind. It tells the unit where to look when a component of a working system is replaced.
• Sellen, A. J. and Harper, R. H. R. (2002). The Myth of the Paperless Office. Cambridge, MA: MIT Press. doi:10.7551/mitpress/4833.001.0001 Sellen and Harper asked why paper survived predictions of its disappearance and found that it affords reading, annotation and movement among documents in ways the replacements did not. Their account records an average 40 per cent rise in paper consumption after email was introduced, and the case shows a plan for use that modelled the document, not the work.
• Bannon, L. J. (1991). From human factors to human actors: The role of psychology and human-computer interaction studies in system design. In J. Greenbaum and M. Kyng (eds), Design at Work: Cooperative Design of Computer Systems. Hillsdale: Erlbaum, 25-44. doi:10.1016/b978-0-08-051574-8.50024-8, which resolves to the 1995 reprint in Readings in Human-Computer Interaction, 205-214. Bannon argues for moving from a human-factors view of the user as a measurable component to a view of people as actors who can participate in design. It gives the unit the design consequence of treating action as situated.
• Anderson, R. J. (1994). Representations and requirements: The value of ethnography in system design. Human-Computer Interaction 9(2), 151-182. doi:10.1207/s15327051hci0902_1 Anderson argues that design's uptake of ethnography rested on a misunderstanding, since the practice was defined as data collection for requirements rather than as social-scientific analysis. He proposes an analytic ethnography for HCI, and it supplies the unit's caution against the analysis-to-requirements shortcut.
• Randall, D., Harper, R. and Rouncefield, M. (2007). Fieldwork for Design: Theory and Practice. London: Springer. doi:10.1007/978-1-84628-768-8 The book presents fieldwork for design as a practical relationship between theory and practice, with a history of ethnographic approaches in CSCW and HCI and examples from academic and commercial settings. It gives the practical account of design-oriented fieldwork.
• Crabtree, A., Rouncefield, M. and Tolmie, P. (2012). Doing Design Ethnography. London: Springer. doi:10.1007/978-1-4471-2726-0 The authors set out ethnomethodologically informed design ethnography with practical guidelines, describing how findings are carried into specification documents, use cases, storyboards, scenarios and prototypes. It is the unit's constructive answer to Anderson, and it aims to get past ethnography stopping at implications for design.
• Kuutti, K. and Bannon, L. J. (2014). The turn to practice in HCI: Towards a research agenda. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '14). New York: ACM, 3543-3552. doi:10.1145/2556288.2557111 Kuutti and Bannon argue that a practice paradigm has emerged in HCI alongside the dominant interaction paradigm, and that it is a complementary lens rather than a replacement. It explains why the situated argument has become a field-level frame.
• Suchman, L., Blomberg, J., Orr, J. E. and Trigg, R. (1999). Reconstructing technologies as social practice. Am. Behav. Sci. 43(3), 392-408. doi:10.1177/00027649921955335 Suchman and colleagues review a twenty-year programme linking everyday practice to technology design and use in three strands: critiques of technical discourse, ethnographies of work and technologies in use, and design interventions. It gives a synoptic account of the programme its themes belong to.