Interaction design methods and critique

A design process is a promise about how decisions will be made before anyone knows the answer. It can help a team decide what to build, or it can reassure a client that the decision was rational, and the two look identical on a slide. That is the contested question here: whether process models describe what designers do, or prescribe a ritual that makes an outcome feel earned.

Three claims sit in tension: that design is a learnable sequence of phases, that design knowledge is situated and hard to state, and that a process is a political instrument, because it decides who gets to speak. Hold only one and you will mistake a diagram for a method.

Why the double diamond describes a journey and not a method

The most reproduced diagram in British design is the Double Diamond, published by the Design Council and licensed for reuse (Design Council, n.d.). Its first diamond opens into Discover and closes at Define; the second opens into Develop and closes at Deliver. Discovery means spending time with the people affected rather than assuming the problem. Define means restating the challenge from that insight, Develop means generating several answers, and Deliver means testing them at small scale (Design Council, n.d.).

The Council is candid about the origin. In 2003 it promoted design management with no standard way to describe a process, so Richard Eisermann, then its Director of Design and Innovation, asked his team to describe one. The team codified and renamed an existing idea, and Eisermann says they stood on the shoulders of giants (Design Council, n.d.).

Read as a journey, the diamond gives a team a shared word for where it is, and it protects divergence. Read as a method, it is nearly empty. It does not say whom to speak to, with what instrument, or how to tell a real insight from a memorable anecdote. The Council's page calls it an invitation to get involved rather than an instruction manual (Design Council, n.d.). A constructed case shows the failure: a team runs a two-week Discover phase, produces a wall of quotes, and enters Define unable to say which quote changed a decision. The phase was completed and nothing was discovered.

Design thinking as practised and design thinking as sold

Kimbell's review identifies three accounts of design thinking (Kimbell, 2011). The first treats it as a cognitive style, a way of reasoning designers happen to have. The second treats it as a general theory of design, applicable to almost any problem. The third treats it as a resource for organisations, the version that travelled into business schools and consultancies (Kimbell, 2011).

The cognitive-style account leans on a dualism between thinking and doing, as if a designer plans in the head and then executes. The general-theory account ignores how situated and historically varied designers' practices are. The organisational account makes the designer the main agent and leaves everyone else a resource to be mobilised (Kimbell, 2011). Her alternative is to study design as a set of situated practices.

Take a constructed case. A health service buys a design-thinking workshop to redesign an appointment system, and it produces sticky notes, a persona and a journey map. Six months later the map is on a wall and the system is unchanged, because the workshop never had authority over the booking rules or the rotas. As practised, design thinking was a conversation among people who could not decide. As sold, it promised to solve a problem the buyer could not name.

Participation changes who holds the pen, not only who is asked

Participatory design began in a political context. Muller traces it to the Scandinavian workplace-democracy movement, where researchers allied with organised labour to give workers a voice in the systems that would reorganise their jobs (Muller, 2003). Sanders and Stappers date research on user participation to the 1970s, when the Collective Resource Approach in Norway, Sweden and Denmark engaged workers in building workplace systems (Sanders and Stappers, 2008).

Kensing and Blomberg name three recurring issues: the politics of design, the nature of participation, and the methods and techniques for participation (Kensing and Blomberg, 1998). They also ask whether a participatory result survives once the researchers leave.

The distinction that matters is between consultation and participation. Consultation asks people to react to a decision someone else has framed. Participation gives them a share of the framing. Muller surveys techniques that move the encounter into a shared space: workshops, storytelling through text, photography and drama, games for analysis and design, and prototype co-creation (Muller, 2003). He notes that the field's move into commercial practice has divided it, because a democratic instrument can be bought as a technique for managing consent (Muller, 2003).

Take a constructed case. A hospital plans a shift-handover tool. Consulted, nurses are shown screens and asked whether they like them. Participating, they help decide what a handover must carry, and one criterion changes the data model before any screen exists: the outgoing shift must record an unfinished task as unfinished. The commitments that separate the two procedures are that users are full participants rather than subjects, that design is a political act, and that the aim strengthens the less powerful party (Kensing and Blomberg, 1998). Remove them and the method degrades into consultation with better paperwork.

Co-creation is a partnership, and the word covers a lot of practice

Sanders and Stappers separate two routinely confused terms. Co-creation is any act of collective creativity shared by two or more people. Co-design is that creativity applied across the whole design process, so it is the narrower term (Sanders and Stappers, 2008). Their subject is the front end, where it is not yet known whether the deliverable is a product, service or interface.

The practice is older than the vocabulary: the user-centred approach, with the user as subject, was largely United States driven, while the participatory approach, with the user as partner, was led by Northern Europeans (Sanders and Stappers, 2008).

Take a constructed case. A council redesigns a waste-collection service. A survey returns tidy preferences. A co-design session gives residents materials and a prompt to show how a week of household waste actually behaves, and the models expose a mismatch between the collection schedule and the days people are at home. The tool surfaces a constraint the survey could not, which is why it matters more than the invitation.

The honest limit is that this is a landscape survey and a manifesto, not an evaluation: it does not measure whether co-design beats the alternatives (Sanders and Stappers, 2008).

Value-sensitive design asks whose values the system already serves

Every design embeds values, whether or not anybody names them. Friedman and Hendry's book is the fullest statement of an approach that names them on purpose. Value sensitive design has been under development for more than two decades, and it brings moral imagination and technical imagination together so that values are engaged at every stage rather than inspected at the end (Friedman and Hendry, 2019).

The practical contribution is a set of seventeen methods, among them stakeholder analysis, value scenarios and multilifespan timelines (Friedman and Hendry, 2019). The first building block is the split between direct stakeholders, who interact with the system, and indirect stakeholders, who are affected without using it. The second is the discipline of writing down value conflicts, so that a trade-off becomes a decision someone owns. The authors treat the investigation of indirect stakeholders as an open question that still needs work (Friedman and Hendry, 2019).

Take a constructed case. A university installs attendance monitoring in lecture halls. Students and lecturers are direct stakeholders. Indirect ones include cleaners whose routes change, staff whose performance is inferred from a proxy, and applicants judged by a system they never met. Naming them turns a technical choice into a question about power, which is the point of the approach.

Speculative design tests futures, and the objections are serious

Critical and speculative design aims at futures rather than products. Dunne and Raby describe a form of design used to create ideas as well as things, posing what-if questions to open debate about the kinds of future people want and do not want, instead of forecasting trends (Dunne and Raby, 2013). It physicalises speculation so that it can be argued with, but does not deploy it for people to use. Examples include a solar kitchen restaurant, a flypaper robotic clock, a menstruation machine and a cloud-seeding truck (Dunne and Raby, 2013).

Tonkinwise's review accepts the case for speculating and presses on what the speculations assume (Tonkinwise, 2014). He argues that a shopping framework and a taste regime sit underneath the work, that the authors' first-person plural slips between meaning themselves and meaning humanity, and that class difference is absent despite the book's focus on the 2008 financial crisis. He adds that the projects rarely make an affirmative case for a future anyone would want (Tonkinwise, 2014).

Take a constructed case. A team builds a household device that shows what a smart meter reveals to a supplier. Used well, it opens a question before a product exists. Used badly, it substitutes provocation for evidence and never reaches the households it claims to speak for. Speculative design is a way to test a future, not a way to prove a claim about one.

Research through design makes knowledge, but not the generalising kind

Research through design asks whether making things can be a way of knowing. Zimmerman, Forlizzi and Evenson propose a model in which designers produce novel integrations of research in order to make the right thing, a product that moves the world from its current state to a preferred one, and so contribute while addressing under-constrained problems (Zimmerman, Forlizzi and Evenson, 2007).

Gaver's later essay is the corrective the field needed. He argues that research through design is likely to produce theory that is provisional, contingent and aspirational, and that the diversity of approaches is natural for a generative endeavour rather than a sign of low standards (Gaver, 2012). His proposal is to treat theory as annotation of realised design examples, and especially of portfolios of related pieces, rather than as general laws to be tested.

That reframing fixes the boundary. Research through design can support statements about what is possible, what a designed thing does, and which concerns it exposes. It cannot support a statement about how a population will behave, because it does not sample one. A project that builds three versions of a reminder service and documents what each made easy supports a claim about a design space and its trade-offs, not the claim that users prefer the third version.

Critique is a craft with a form you can practise

Critique has three moves. Describe what is there, without evaluation. Interpret what it means or does. Judge it against criteria you state. The separation is not pedantry: it is what stops a judgement from being unfalsifiable.

The distinction to internalise is between an objection about evidence and an objection about preference. An evidence objection says a claim is not supported by what was shown, and it can be met by better evidence. A preference objection says the critic would have done it differently, and it can only be met by argument about criteria. Both are legitimate, and they are not interchangeable.

A rationale another designer could act on has a recognisable shape. State the question the design answered, the options that were genuinely available, the criteria used to compare them, the decision and the evidence behind it. Zimmerman and colleagues supply one scaffold for the criteria, the four lenses for evaluating a design research contribution (Zimmerman, Forlizzi and Evenson, 2007), and Gaver supplies the other, the annotated portfolio (Gaver, 2012).

Take a constructed case. A team ships a checkout with no confirmation step and calls it frictionless. A description says the button submits the order. An interpretation says the design treats commitment as an obstacle. The judgement needs a stated criterion, and the honest one is reversibility: an irreversible action should not be the one easiest to take by accident. The evidence objection is that no test showed accidental orders rising or falling, so the gain has not been shown. That objection names the study that would settle it.

A process earns its place by changing a decision, and it fails when it only changes the mood. The structures a designer defends in a rationale, the latency of a search and the way a queue behaves under load, are where the next part of the programme begins.