Prototypes as arguments: what a prototype can know

A prototype is usually introduced by its finish: a paper sketch is low fidelity, a clickable mock-up medium, a coded build high. Houde and Hill (1997) proposed describing prototypes by the questions they answer. The comparison experiments that followed still treated finish as the variable to manipulate (Walker et al., 2002; Virzi et al., 1996).

This unit, the fourth of the ten, argues that a prototype is a claim about a design space, not a rough draft of the product. Where the previous unit took the problem, this one takes the artefact. Its value lies in what it makes visible to a particular audience, so fidelity is a choice about what a test can detect and not a measure of progress. The unit ends where the literature disagrees about what fidelity is.

A prototype is defined by the question it answers

Houde and Hill (1997) replace fidelity with three questions. A role prototype asks whether an artefact would be useful in someone's life. A look-and-feel prototype asks what it would be like to use. An implementation prototype asks whether it can be built. These are different questions with different audiences, and one artefact usually mixes them. Finish therefore says nothing about which question is answered. Their own space-planning project produced all three at once.

Lim, Stolterman and Tenenberg (2008) give the vocabulary its theoretical statement. A prototype is a filter that constrains what a designer can see, because the medium decides which aspects are visible, and a paper sketch is closed to a screen-reader user whatever it tests. It is also a manifestation that makes a design idea knowable to someone else.

Buxton (2007) supplies the distinction the thesis rests on. Getting the design right and getting the right design are separate questions. A sketch exists to suggest, and a prototype exists to test. A team that treats its first prototype as a small version of the product has answered the second question before asking the first.

Fidelity is an evidential choice

The comparison experiments do not settle which fidelity is better. Walker, Takayama and Landay (2002) compared user testing with low- and high-fidelity prototypes in paper and in computer media. They found the two prototypes equally good at uncovering usability issues, with results independent of the medium.

Virzi, Sokolov and Karis (1996) compared the usability problems found with low- and high-fidelity prototypes across two products. The products were a CD-ROM electronic book and an interactive voice response system. The two conditions uncovered substantially the same problems, and individual problems were detected by a similar proportion of participants.

Sauer and Sonderegger (2009) manipulated prototype fidelity and the aesthetics of the design in usability tests. They measured behaviour, subjective evaluation and emotion. Both shaped what participants did and reported, so a prototype's finish can bias the evidence about the design.

Practice reached a related conclusion earlier. Rudd, Stern and Isensee (1996) describe IBM's move from high-fidelity to low-fidelity prototyping, reporting that low-fidelity prototypes found more usability problems earlier and at lower cost. It is an experience report, not a controlled study.

A prototype makes visible what the team did not know to ask

If a prototype is an argument, its most interesting result is a question nobody had formulated. Buchenau and Suri (2000) describe experience prototyping as engaging designers, users and clients first-hand with the conditions of use. They claim it changes what a team knows, and not only what it has shown users.

Jensen, Elverum and Steinert (2017) give that claim a case base. They analysed 19 prototypes across eight engineering design companies. They describe how those prototypes elicited requirements that were unknown unknowns to the team. The organisation of the work, and not only the artefact, turns a prototype into an instrument for finding out.

The evidence beneath the advice is thin

Once prototyping is described as a way of knowing, the question is what the field knows about it. Camburn and colleagues (2017) reviewed prototyping strategies, techniques and guidelines in engineering design. They report wide variation in practice and weak cumulation of findings.

Menold, Jablokow and Simpson (2017) show the attempt to convert prototyping craft into a structured method: Prototype for X, specified from a literature review and a study of novices' mental models, reporting early evidence rather than a controlled trial. Bjarnason, Lang and Mjoberg (2023) reach the same diagnosis from software, combining a mapping study with a multi-case study, modelling prototyping as a requirements practice in agile development and reporting that the area lacks theory.

What a prototype makes visible depends on who is looking

The audience is part of the instrument. Codner and Lauff (2023) analysed children's comprehension of physical toy prototypes at several levels of fidelity across a semester-long design project. They identify which aspects help or hinder understanding of form and function, in particular mental models, signifiers and affordances.

Chen and Yoon (2024) show the practice moving. They interviewed ten practitioners and educators in the software industry. They find paper and low-fidelity prototyping declining in hybrid, digitally tooled workplaces. Their framework treats a prototype as three components: fidelity, medium and dimension.

Mitomi and Shirasaka (2025) test the recipient directly. They modelled how a person forms a final product image, then tested it with forty participants at two levels of fidelity. Recipients of the low-fidelity prototype projected more of their own knowledge onto it, so the gap between a designer's intention and a recipient's reading is systematic rather than incidental.

Contested question: is fidelity one axis, or a relation to a question and an audience?

The literature disagrees about what fidelity is. The first position treats it as a single variable that a team sets and trades against cost, and each of the four comparison studies above varies it. Fidelity is a dial, and its setting is a compromise between effort and the quality of evidence.

The second position holds that fidelity is a set of dimensions in relation to the question and the audience. Its theoretical form is Houde and Hill (1997), whose role, look-and-feel and implementation prototypes are different questions rather than points on a scale, and Lim and colleagues (2008), for whom the prototype is a filter whose medium decides what is visible. Its empirical form is Chen and Yoon (2024), whose framework separates fidelity from medium and dimension, and Codner and Lauff (2023), for whom comprehension depends on signifiers and affordances relative to a child audience. Its experimental form is Mitomi and Shirasaka (2025), who measured how fidelity and the recipient's traits shape what is read into a prototype.

What would settle it is a rating study the literature does not contain. The first position predicts that a declared fidelity score, set on one axis while other dimensions are held constant, will predict the volume and kind of findings. The second position predicts that it will not: two prototypes at the same nominal fidelity, differing in medium and audience, yield different findings, and independent raters do not agree on a prototype's fidelity when the dimensions are unspecified. Virzi and colleagues (1996) came closest, but no source here measured whether a fidelity rating predicted the findings.

The two positions rest on different kinds of evidence. The one-axis position rests on four comparison studies: two controlled experiments, one manipulation of fidelity and aesthetics, and one experience report from IBM. The multidimensional position rests on two theoretical essays, two qualitative studies and one experiment, which measured how fidelity and the recipient's traits shape what is read into a prototype. The experiments support the axis as something a team can set; the theory and the new experiment deny that the setting carries the meaning. Until the rating study exists, the designer should declare fidelity, medium and audience together, treat a bare fidelity score as uninterpretable on its own, and build the cheapest prototype that answers the question. The strongest objection to the unit's thesis is that the two controlled experiments found low and high fidelity uncovering substantially the same problems, so for usability testing the dimensions collapse back into a cost decision.

Where the argument is heading

Named work points one way. The field moves from a fidelity spectrum towards a set of dimensions matched to a question and an audience. Bjarnason and colleagues (2023) try to build prototyping theory from evidence rather than prescription, and report how little theory the area has. Chen and Yoon (2024) record practitioners already treating fidelity as multidimensional. Codner and Lauff (2023) and Mitomi and Shirasaka (2025) show the audience as a variable that changes what a prototype communicates. The missing study would use matched prototypes at a declared fidelity, the same participants, and a measure not only of the problems found but of whether the prototype changed a decision.

The sources this unit rests on

• Houde, S. and Hill, C. (1997). What do prototypes prototype? In Handbook of Human-Computer Interaction. Elsevier. doi:10.1016/B978-044481862-1.50082-0 Houde and Hill propose describing prototypes by the questions they answer rather than by fidelity, distinguishing role, look-and-feel and implementation prototypes. It is a conceptual framework from practice, illustrated rather than tested.
• Lim, Y.-K., Stolterman, E. and Tenenberg, J. (2008). The anatomy of prototypes: prototypes as filters, prototypes as manifestations of design ideas. ACM Transactions on Computer-Human Interaction. doi:10.1145/1375761.1375762 Lim, Stolterman and Tenenberg argue prototypes are both filters constraining what a designer sees and manifestations making design ideas knowable, a nature under-theorised next to the fidelity debate. It is developed through design cases. Conceptual analysis.
• Buchenau, M. and Suri, J. F. (2000). Experience prototyping. Proceedings of the 3rd Conference on Designing Interactive Systems. doi:10.1145/347642.347802 Buchenau and Suri describe experience prototyping as engaging designers, users and clients first-hand with the conditions of use, illustrated with projects. It argues the method changes what a team knows, and the claim rests on those projects. It is a design project.
• Walker, M., Takayama, L. and Landay, J. A. (2002). High-fidelity or low-fidelity, paper or computer? Choosing attributes when testing web prototypes. Proceedings of the Human Factors and Ergonomics Society Annual Meeting. doi:10.1177/154193120204600513 Walker, Takayama and Landay compared user testing with low- and high-fidelity prototypes in paper and computer media and found the two equally good at uncovering usability issues, independently of the medium. It supports the unit's scepticism about a single fidelity axis. A single experiment.
• Sauer, J. and Sonderegger, A. (2009). The influence of prototype fidelity and aesthetics of design in usability tests: effects on user behaviour, subjective evaluation and emotion. Applied Ergonomics. doi:10.1016/j.apergo.2008.06.006 Sauer and Sonderegger manipulated prototype fidelity and design aesthetics in usability tests, measuring behaviour, subjective evaluation and emotion. Both shaped what participants did and reported, so finish can bias the evidence. A single experiment.
• Rudd, J., Stern, K. and Isensee, S. (1996). Low vs. high-fidelity prototyping debate. Interactions. doi:10.1145/223500.223514 Rudd, Stern and Isensee describe IBM's move from high-fidelity to low-fidelity prototyping, reporting that low-fidelity prototypes found more usability problems earlier and at lower cost. It is an experience report, and its claims are not measured. Practitioner account.
• Virzi, R. A., Sokolov, J. L. and Karis, D. (1996). Usability problem identification using both low- and high-fidelity prototypes. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. doi:10.1145/238386.238516 Virzi, Sokolov and Karis compared the usability problems identified with low- and high-fidelity prototypes across two products and found substantially the same problems. It is a controlled experiment and the unit's strongest evidence. Only two products.
• Camburn, B., Viswanathan, V., Linsey, J., Anderson, D., Jensen, D., Crawford, R., Otto, K. and Wood, K. (2017). Design prototyping methods: state of the art in strategies, techniques, and guidelines. Design Science. doi:10.1017/dsj.2017.10 Camburn and colleagues review prototyping strategies, techniques and guidelines in engineering design, reporting wide variation in practice and weak cumulation of findings: the unit's evidence that the advice outruns the knowledge.
• Jensen, M. B., Elverum, C. W. and Steinert, M. (2017). Eliciting unknown unknowns with prototypes: introducing prototrials and prototrial-driven cultures. Design Studies. doi:10.1016/j.destud.2016.12.002 Jensen, Elverum and Steinert analysed 19 prototypes across eight engineering design companies and describe how they elicited requirements that were unknown unknowns to the team, through prototrials. It is a multi-case investigation, and eight companies limit its reach.
• Buxton, B. (2007). Sketching User Experiences: Getting the Design Right and the Right Design. Morgan Kaufmann. https://www.sciencedirect.com/book/9780123740373/sketching-user-experiences Buxton argues that getting the design right and getting the right design are separate questions, that a sketch suggests while a prototype tests, and that framing matters more than refining a chosen concept. It is a practitioner text and reports no study.
• Menold, J., Jablokow, K. and Simpson, T. (2017). Prototype for X (PFX): a holistic framework for structuring prototyping methods to support engineering design. Journal of Engineering Design. doi:10.1016/j.destud.2017.03.001 Menold, Jablokow and Simpson develop specifications for Prototype for X from a literature review and a study of novices' mental models. It converts prototyping craft into a structured method, reporting early evidence rather than a trial.
• Chen, J. and Yoon, D. (2024). Exploring the diminishing allure of paper and low-fidelity prototyping among designers in the software industry. Proceedings of the CHI Conference on Human Factors in Computing Systems. doi:10.1145/3613904.3642774 Chen and Yoon interviewed ten practitioners and educators and found paper and low-fidelity prototyping declining in hybrid, digitally tooled workplaces. Their framework treats a prototype as fidelity, medium and dimension.
• Bjarnason, E., Lang, F. and Mjoberg, A. (2023). An empirically based model of software prototyping: a mapping study and a multi-case study. Empirical Software Engineering. doi:10.1007/s10664-023-10331-w Bjarnason, Lang and Mjoberg combine a mapping study of software prototyping with a multi-case study to model prototyping as a requirements practice in agile development. It reports that the area lacks theory and supplies no controlled comparison.
• Codner, A. C. and Lauff, C. A. (2023). Prototyping for children: understanding how fidelity affects children's comprehension of prototypes. Proceedings of the Design Society. doi:10.1017/pds.2023.301 Codner and Lauff analysed children's comprehension of toy prototypes at several levels of fidelity, identifying which aspects help or hinder understanding of form and function, notably signifiers. It shows the audience changes what a prototype communicates. Small study.
• Mitomi, K. and Shirasaka, S. (2025). Understanding how prototypes are interpreted: a structural model of final product interpretation. The Design Journal. doi:10.1080/14606925.2025.2572084 Mitomi and Shirasaka modelled how a recipient forms a product image from a prototype and tested it with forty participants at two fidelity levels; low fidelity invited more projection of the recipient's own knowledge. It measures interpretation, not usability findings.