Probes: instruments built to inspire, and the cost of judging them as evidence

The probe entered HCI as an instrument never meant to answer a question. Gaver, Dunne and Pacenti sent packs of tasks and artefacts to households. What came back was meant to unsettle the designers, not confirm what they assumed. They described the method as gathering inspiration and uncertainty (Gaver, Dunne and Pacenti, 1999). Hutchinson and colleagues turned the idea into a deployment method. It aims to understand what people want in a real setting, to field-test the technology, and to inspire (Hutchinson et al., 2003).

That origin is the unit's problem. This second of ten units asks what an instrument built to inspire may be asked to show, and the first fixed what a deployment can claim. If a probe is built to produce inspiration, asking it for an effect size is a category error: it measures the instrument against a job it was never given. The field has made that error and has tried twice to repair it. One repair relocates a probe's value to the design process around it. The other reframes uncertainty as a resource to work with.

The probe was defined against the answer

The three goals are not ranked, and that design decision creates the trouble. A probe must understand a setting, field-test a technology and inspire at once, so its material cannot be read against one goal alone. Hutchinson and colleagues deployed a messageProbe and a videoProbe with families in France, Sweden and the United States. They state that a probe is not a prototype, because it is not revised during a study (Hutchinson et al., 2003).

The originators justified the cultural probe by its refusal to be precise. The packs gather fragments of other people's lives. The knowledge they sought was partial and ambiguous by intention (Gaver, Dunne and Pacenti, 1999). A probe is closer to a conversation opener than to a measurement device, and success is whether a design team sees differently.

The originators defended uncertainty, and the field's error was an unstated standard

When the field began to demand more rigour, the originators pushed back. To control a probe and make it evidential defeats its purpose. The uncertainty it deliberately produces is the point, not the flaw (Gaver et al., 2004).

The pressure is visible in the method's spread. Mattelmaki separated four reasons for applying probes in product development: inspiration, information, participation and dialogue (Mattelmaki, 2005). Information and dialogue pull a probe towards being a source of findings. Once a study asks a probe for information, it is a short step to asking for reproducible results.

Boehner, Vertesi, Sengers and Dourish traced how cultural probes were adopted and adapted across HCI and found a proliferation of divergent uses (Boehner et al., 2007). Their disagreements, the authors argue, are epistemological rather than technical: they concern what counts as valid interpretation and what relationship a method has to the methodology behind it. The error is not judging probes by the wrong standard; it is leaving the standard unstated.

A frame that makes an inspiration-seeking instrument arguable

Research through design rescues the probe from bare inspiration. Zimmerman, Forlizzi and Evenson proposed it as a legitimate HCI method. Its object is to make the right thing rather than test a claim. They set out four lenses for evaluating the contribution (Zimmerman et al., 2007). An inspiration-seeking instrument, on that frame, has its own criteria.

Vines and colleagues give the frame a worked instance. They ran participatory design workshops with ten people over eighty on new banking technologies. They handed out concept cards carrying deliberately provocative sketches, and asked participants to write and draw on them (Vines et al., 2012). The participants criticised both current banking practice and most of the proposed concepts, and their objections became the material for the next iteration. Critique, on this showing, is the instrument's output. What the case does not show is whether any concept survived into a service, or whether the objections would have appeared in a conventional requirements interview.

Gaver made those criteria explicit. Research through design, he argued, is likely to produce theories that are provisional, contingent and aspirational. The community should be wary of impulses towards convergence and standardisation (Gaver, 2012). His proposal is that practice-based research treat theory as an annotation of realised design examples and portfolios.

The repair relocates the value to process and to uncertainty

The most concrete repair relocates a probe's value from its artefacts to the work around them. Wallace, McCarthy, Wright and Olivier gave a synthetic account of probe design and use across projects spanning a decade. They treat probes as directed craft objects, which serve as tools for design and tools for exploration. They relate a probe's properties to its use in a project (Wallace et al., 2013). A probe is not a bad questionnaire; its meaning is produced by the design activity it enables.

A second repair reframes the uncertainty itself. Epp, Rosen, Salovaara and Sanchez argue that uncertainties can be generative resources in research through design. They identify three dynamics: deviating widely, repeated focusing and oscillating between uncertainties (Epp et al., 2024). Giaccardi, Murray-Rust, Redstrom and Caramiaux frame the same shift and name the condition design as a probabilistic outcome (Giaccardi et al., 2024).

A method that seeks inspiration needs both a way to fail and a vocabulary for its claim

A method that seeks inspiration must still be able to fail, or it can claim anything. Gaver and colleagues describe the failure of a sensor-based system meant to evoke interpretation and appropriation in domestic settings. They name symptoms under four themes: engagement, reference, accommodation, and surprise and insight (Gaver et al., 2009). An interpretive system, they conclude, can fail incontrovertibly yet instructively; a method that reports its own failure has chosen a different test.

A claim needs a vocabulary as well as a test. Benford and colleagues separate the movements an interface expects, the movements it can sense and the movements an application desires. Comparing the three reveals problems and inspires features (Benford et al., 2005). Applied to three interfaces including the Drift Table, it is the unit's constructive alternative.

The probe is still in use, and now it scaffolds participation

Sun and colleagues designed a workbook probe for families. They deployed it for two to three months with nine families whose children were aged six to eleven (Sun et al., 2026). A probe workbook assumes a household that can read and write. They distilled five methodological lessons from the workbooks and interviews. The study reports a finding about how to run the method rather than about smart homes: the probe's output is documented as method, and no evidential claim is made.

Contested question: has the category error been repaired, or renamed?

The literature disagrees about whether any of this answers the evidential objection. The originators hold that a probe should not be made evidential at all, because controlling it defeats the uncertainty which is its purpose (Gaver et al., 2004). Bardzell, Bardzell and Koefoed Hansen press from the other side. Research through design must clarify how its objects contribute to knowledge. It must also accept that the critical reception of an object is foundational (Bardzell et al., 2015). The repair wing relocates value to process and uncertainty (Wallace et al., 2013; Epp et al., 2024); the critique wing asks for a knowledge claim others can take up. Zimmerman and colleagues and Gaver keep both positions arguable (Zimmerman et al., 2007; Gaver, 2012). Boehner and colleagues deny there is a single question to settle (Boehner et al., 2007).

Gaver and colleagues and the critique wing argue from position rather than data (Gaver et al., 2004; Bardzell et al., 2015). Wallace and colleagues give a synthesis of probe projects across a decade, and Epp and colleagues name uncertainty as a resource (Wallace et al., 2013; Epp et al., 2024). What would settle it is a traceable claim: a probe study would state in advance what it will assert and what would count against it, and another project would take up that assertion. Benford's framework is a candidate vocabulary for such claims (Benford et al., 2005). While the dispute is open, report the probe as method and as participation, and claim no effect. The strongest objection is that a method which can never be wrong has stopped making a claim.

Where the area is heading

The direction of travel is towards probes that are explicit about which kind of study they are. Giaccardi and colleagues frame data-driven design as a field that must work with uncertainty rather than eliminate it (Giaccardi et al., 2024). Epp and colleagues give designer-researchers a vocabulary for doing so (Epp et al., 2024). Bardzell and colleagues keep the demand for a stated knowledge claim in view (Bardzell et al., 2015). The likely convergence is a split register. Probes would be reported as method and as participation, and design-research studies would state their claims in a shared vocabulary such as Benford's (Benford et al., 2005).

The sources this unit rests on

Each note states what the cited document did and found. Titles are in British spelling, and identifiers were checked against registered metadata on 2 November 2026.

• Gaver, B., Dunne, T. and Pacenti, E. (1999). Cultural probes. doi:10.1145/291224.291235 Introduces cultural probes as packs of tasks and artefacts sent to participants to return fragments of their lives, presented as gathering inspiration and uncertainty rather than extracting data. It founds the tradition whose status this unit examines.
• Hutchinson, H. et al. (2003). Technology probes: inspiring design for and with families. doi:10.1145/642611.642616 Defines the technology probe as simple, flexible and adaptable, with three goals at once: understanding needs in a real setting, field-testing the technology, and inspiring. It reports a messageProbe and a videoProbe deployed with families in France, Sweden and the United States, and is explicitly not a prototype.
• Gaver, W. W., Boucher, A., Pennington, S. and Walker, B. (2004). Cultural probes and the value of uncertainty. doi:10.1145/1015530.1015555 Defends the uncertainty probes deliberately produce, arguing that controlling a probe to make it evidential defeats its purpose. It gives the originators' own counter to this unit's critique arm.
• Benford, S. et al. (2005). Expected, sensed, and desired: a framework for designing sensing-based interaction. doi:10.1145/1057237.1057239 Separates the movements an interface expects, those it can sense and those an application desires, arguing that comparing the three reveals problems and inspires features. Applied to three interfaces including the Drift Table, it gives this unit a vocabulary in which a probe study's claim can be checked.
• Mattelmaki, T. (2005). Applying probes: from inspirational notes to collaborative insights. doi:10.1080/15719880500135821 Examines probe use in product development and gives four reasons for applying probes: inspiration, information, participation and dialogue. The information and dialogue reasons are how the method's success carried the category error with it.
• Boehner, K., Vertesi, J., Sengers, P. and Dourish, P. (2007). How HCI interprets the probes. doi:10.1145/1240624.1240789 Traces how cultural probes were adopted and adapted into divergent and derivative uses. It argues the resulting disagreements are epistemological rather than technical, supplying this unit's diagnosis that the standard is left unstated.
• Zimmerman, J., Forlizzi, J. and Evenson, S. (2007). Research through design as a method for interaction design research in HCI. doi:10.1145/1240624.1240704 Proposes research through design as a legitimate HCI method whose object is to make the right thing rather than test a claim, with four lenses for evaluating the contribution. It sets design-led enquiry beside scientific enquiry, which makes an inspiration-seeking instrument arguable.
• Gaver, W., Bowers, J., Kerridge, T., Boucher, A. and Jarvis, N. (2009). Anatomy of a failure: how we knew when our design went wrong, and what we learned from it. doi:10.1145/1518701.1519040 Analyses the failure of a sensor-based domestic system meant to evoke interpretation and appropriation, naming symptoms under four themes. It shows that an interpretive method can fail incontrovertibly yet instructively.
• Gaver, W. (2012). What should we expect from research through design? doi:10.1145/2207676.2208538 Argues research through design yields theories that are provisional, contingent and aspirational, and warns against impulses towards convergence and standardisation. It proposes theory as an annotation of realised design examples and portfolios, clarifying the claim a probe study can make.
• Wallace, J., McCarthy, J., Wright, P. C. and Olivier, P. (2013). Making design probes work. doi:10.1145/2470654.2466473 Gives a synthetic account of design probe design and use across projects spanning a decade. It treats probes as directed craft objects and relates a probe's properties to its use, relocating a probe's value to the design activity around it.
• Bardzell, J., Bardzell, S. and Koefoed Hansen, L. (2015). Immodest proposals: research through design and knowledge. doi:10.1145/2702123.2702400 Argues research through design must clarify how its objects contribute to knowledge, holding that the critical reception of an object is foundational because a design object's meanings cannot be paraphrased simply. It raises the evidential bar for probe studies.
• Epp, F. A., Rosen, A. P., Salovaara, A. and Sanchez, C. (2024). Uncertainties as generative resources in research through design: three dynamics for moving in a design space. doi:10.1145/3689041 Argues uncertainties can be generative resources in research through design and identifies three dynamics, deviating widely, repeated focusing and oscillating between uncertainties. It supports this unit's claim that a probe's uncertainty is a feature, not a flaw.
• Giaccardi, E., Murray-Rust, D., Redstrom, J. and Caramiaux, B. (2024). Prototyping with uncertainties: data, algorithms, and research through design. doi:10.1145/3702322 Introduces a special issue on uncertainty and prototyping, outlining three histories of uncertainty in design and naming the condition design as a probabilistic outcome. It offers the current theoretical treatment of uncertainty in design research.
• Sun, K. et al. (2026). Designing workbook probes for families: a smart home case study of intergenerational co-speculation. doi:10.1145/3773077.3806138 Deploys a workbook probe with nine families whose children were aged six to eleven, over two to three months, and distils five methodological lessons. It reports the workbook as a participation scaffold rather than a source of findings.
• Vines, J., Blythe, M., Lindsay, S., Dunphy, P., Monk, A. and Olivier, P. (2012). Questionable concepts: critique as resource for designing with eighty somethings. doi:10.1145/2207676.2208567 Participatory design workshops with ten people over eighty about new banking technologies, using deliberately provocative concept cards that participants annotated. Participants criticised current practice and most concepts, and their objections drove further iterations, so critique is treated as a design resource.