The Witness Experience Inventory

Uhde, Alarith; Dreyer, Lianara; Hassenzahl, Marc Article — Published Version The Witness Experience Inventory Interacting with Computers Provided in Cooperation with: WZB Berlin Social Science Center Suggested Citation: Uhde, Alarith; Dreyer, Lianara; Hassenzahl, Marc (2025) : The Witness Experience Inventory, Interacting with Computers, ISSN 1873-7951, Oxford University Press, Oxford, Iss. Advance articles, pp. 1-16, https://doi.org/10.1093/iwc/iwaf010 This Version is available at: https://hdl.handle.net/10419/315673 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Received: May 31, 2024. Revised: January 3, 2025. Accepted: February 22, 2025 © The Author(s) 2025. Published by Oxford University Press on behalf of The British Computer Society. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Interacting with Computers, 2025, 1–16 https://doi.org/10.1093/iwcomp/iwaf010 Article The Witness Experience Inventory Alarith Uhde1,2,*,Lianara Dreyer3and Marc Hassenzahl4 1College of Information Science and Engineering, Ritsumeikan University, Osaka, Japan 2Tokyo College, The University of Tokyo, Tokyo, Japan 3Berlin Social Science Center, Berlin, Germany 4Ubiquitous Design/Experience and Interaction, University of Siegen, Siegen, Germany *Corresponding author: [email protected].jp Abstract Interactions with technology are part of social life, for example in cafés, trains, or parks. This social situatedness not only changes how users experience these interactions. It also influences the situated experiences for other co-located people (“witnesses”). However, despite a large body of research on user experiences, the relation between an interaction and witness experiences, and ways to design for them,remain underexplored.To address this gap, this paper introduces the “Witness Experience Inventory”,a research tool grounded in social-interpretivist theories, that offers a pragmatic approach to study how interactions with technology affect witness experiences. Based on an analysis of eight interactive technologies, we illustrate how the Witness Experience Inventory can inform the design of socially situated interactions with technology to avoid negative and create more positive witness experiences. We provide guidelines for applications of the Witness Experience Inventory in future research and its adaptable coding template. Both build on experiences from our own research, but give future researchers and practitioners the flexibility to adapt the tool to the social settings they study. RESEARCH HIGHLIGHTS •We introduce the “Witness Experience Inventory” (WEI), a qualitative research tool to analyze and design for witnesses experiences of interactions with technology •We provide a detailed overview of aspects that are relevant for witness experiences through the WEI’s coding template, to direct future research •We illustrate how the WEI can inform design decisions to improve witness experiences using three design cases (smartphones, VR glasses, electronic cigarettes) Keywords:social situation; research method; qualitative; experience design; witness experience; social practice. 1INTRODUCTION Interactions with technology are part of social life. Ask any train commuter for a time when some passenger “entertained” everyone else with their phone call. Or go to a stage performance and you will find that it heavily relies on interactive technologies. People take selfies at crowded sightseeing spots, others use drones to film wedding parties. Sidewalk encounters with motorized scooters count as well, just like sharing a table in a café with a laptop user, or “vaping” in the smoking area. Put differently, it would be challenging to find a social gathering today entirely unaffected by interactive technology. Given this ubiquity, we would expect the Human–Computer Interaction (HCI) literature to offer a solid theoretical grounding about how people experience interactions with technology in social situations. In addition, we would expect tools and guidelines grounded in this theory to inform design decisions. And indeed, we do have both theory and tools/guidelines (e.g., Dourish, 2004,Koelle et al., 2018,Korsgaard et al., 2022,Rico & Brewster, 2010,Suchman, 2006). However, we argue here that the two are currently not well integrated. On the one hand, we have extensive theoretical work that has contributed to a better understanding of the complex relations between socially situated interactions and subjective experiences. This work is often more abstract and contributes high-level descriptions of these relations.On the other hand, we have a wide range of tools to assess socially situated experiences, which have grown out of pragmatic needs for data to inform decision-making. These tend to be developed for a specific social setting with a focus on a specific type of interaction, which can cause problems when transferred to other settings or interactions. Some of the findings and guidelines based on these assessments have over time turned out to be less “generalizable” than initially thought (Uhde et al., 2022,2023), and the HCI literature struggles to develop reliable advice to design for social situations. In this paper, we set off to bridge between the theoretical work and the needs of researchers and practitioners. This led us to assemble a research tool based on a combination of existing methods and data analysis approaches, with supportive material for researchers. It is intended to support designing for and evaluating socially situated interactions with technology. We call this tool the “Witness Experience Inventory” (WEI). The WEI combines a semi-structured interview approach with 2|Interacting with Computers, 2025 an open evaluation format, supported by an extensive coding template (the “inventory”). The interview format is based on Interpretative Phenomenological Analysis (IPA; Smith et al., 2009), a method focused on subjective experiences, which reflects the WEI’s grounding in previous (user) experience-oriented research (e.g., Laschke et al., 2013,Uhde et al., 2021). However, we adapted the structure for a user-decentered research focus, informed by social-interpretivist theories of social situatedness and Social Practice Theories (e.g., Reckwitz, 2002,Shove et al., 2012,Suchman, 2006,Uhde & Hassenzahl, 2021). Finally, the coding template is based on the findings of our own initial research about eight interactive technologies outlined below, and thus provides an informed starting point for future work. As the name implies, the WEI focuses on “witnesses”, which stands for all co-situated individuals who do not directly interact with a technology, but whose situated experiences may be influenced by this interaction.1As we illustrate below, the WEI can be used flexibly to study a wide range of interactive technologies and associated witness experiences, and provides concrete insights to improve designs. In sum, we make the following contributions. First, we introduce the WEI and guidelines on how to use it. We present our pilot study about eight common, socially situated interactive technologies, which followed the WEI interview format and served to bootstrap the coding template. Second, the coding template itself provides a starting point for further research about socially situated interactions with technology, and brings together an extensive collection of aspects that can influence situated experiences. Third, we use three practical examples from the pilot study to illustrate how the WEI can inform the (re)design of socially situated interactions with technology. 2BACKGROUND We will first outline the central work on socially situated experiences and common issues around the concept of “social situatedness” that motivated the development of the WEI. To illustrate the core problems, we start with a reflection on existing tools and measurement-oriented approaches to social situatedness. This part mainly refers to the “social acceptability” literature, because it is the currently most thoroughly studied aspect of socially situated experiences. But the same arguments apply more broadly. We then introduce central theoretical work on social situatedness, which indicates why it can be difficult to reliably assess socially situated experiences using strictly standardized tools. Finally, having outlined the key problems with both measurements and theory, we suggest a way to integrate them by defining a potential scope of what a research tool can meaningfully address to inform design for the social. This served as the outset for developing the WEI. 2.1 Measuring Socially Situated, Technology-Mediated Experiences: The Case of Social Acceptability Whether people choose to interact with a certain technology in the social does not depend on its functionality alone. We know 1This use of the term “witness” is in line with some recent work (e.g., Uhde et al., 2022), but slightly different from earlier uses. In particular, Reeves (2011) used the term “wittingness” to describe whether another person is an “insider” or “outsider” of a group where an interaction takes place (similar to “teams”; Goffman, 1959), and whether their attention is intended or expected. We use it here to describe to what extent other people are aware of the interaction happening at all, which may shape their experience even if they are not part of the in-group. from examples such as Google Glass and smartphones that people often end up not using a technology at all, or only under certain circumstances (Koelle et al., 2017,2015,Uhde & Hassenzahl, 2021). The social acceptability literature is concerned with why people interact with some technologies in social situations but not others, and how they experience such interactions. One of its goals is to develop design principles that reduce non-functional barriers of interaction in social situations (e.g., embarrassment or disturbing others; Monk et al., 2004a,b,Montero et al., 2010). In more abstract terms, the social acceptability literature looks for ways to improve how people experience socially situated interactions with technology. A detailed review of that literature can be found in Koelle et al. (2020), with more specific discussions of challenges as seen within the field. Here, we focus on three inherent issues of the social acceptability approach, representative of research primarily focusing on measurement. These include: (1) an isolated focus on measurable elements such as the form of an interaction, independent of the interaction’s specific social situatedness; (2) the goal to develop objective measures, even if they may not meaningfully represent the variety of (subjective) experiences; and (3) a framing of social situatedness driven by the needs of measurements. First, from a practical perspective, it would be desirable if we could isolate situational elements to make them measurable and comparable. For example, social acceptability research has a strong interest in the form of an interaction, and several studies attempt to analyze the form independent of other situational elements. This approach aligns with the problem framing and the aspiration to develop generalizable guidelines for social acceptability. There are various examples in the literature of suggestions for more and less “acceptable” forms of interaction (see Koelle et al., 2020 for an overview), assuming that such forms come with an inherent “acceptability value”. Here, we focus on “subtle” interactions as an example. Subtle interactions resemble casual and natural movements, such as foot tapping, to control a technology. They have been suggested as a way to increase social acceptability (Rico & Brewster, 2010). Although such generalizable design patterns and effects of the form of interaction on peoples’ experiences would be desirable (e.g., how acceptable they find it), there are practical issues. One reason is that the form is not necessarily the cause of “unacceptability”, and it is unclear in what way it contributes to it. In the case of public photography, for example, a reason for low social acceptability could be that the interaction as such is seen as intrusive of someone else’s privacy (see also Koelle et al.,2018). If someone takes photos of a stranger, simply making the form of interaction more “subtle” may not resolve the problem, or it could even make it worse. For example, “upskirting” photography became a growing problem in Japan in the 2000s and manufacturers reacted by making the form of interaction less subtle (with an immutable clicking sound; McCann et al., 2017,Sharon & Koops, 2021). But even if we assume that the form is (part of) the reason for low acceptability, a subtle form may not be suitable for a certain socially situated interaction (as shown for face-to-face conversations; Pohl et al., 2019,Uhde et al., 2022). Thus, although ideally we could study the form of interaction independently, it ultimately needs to be considered in relation to other variables, which can be interaction-specific. The second issue is that social acceptability research tends to imply an inherent “true acceptability value” of a certain technology with its form of interaction and other characteristics. However, in practice, people can have fundamentally different judgements of this acceptability, and it is unclear how we could Uhde et al. |3 resolve their disagreement. There is some awareness of this problem in the social acceptability literature already, which shows in the concepts of “early” and “late adopters” of technologies (e.g., Montero et al., 2010). This categorization implies that, at one point in time, some people (the early adopters) find interactions with certain technologies already acceptable and start using them, while others (the late adopters) do not—although the group names imply that they eventually will. The problem is that if such groups exist, simple acceptability scores are necessarily normative. Whose opinion should they be based on if people fundamentally disagree? It also seems problematic that this alignment of people on an “acceptance timeline” depoliticizes the technology, although there may be valid arguments both for and against its adoption. To an extent, the acceptance timeline prescribes a seemingly inevitable future of adoption. This, in turn, contradicts the goal of social acceptability research to redesign certain characteristics of technology, such as its form, to make it more acceptable. If people eventually adopt the technology anyway, independent of its form or other inherent characteristics, there needs to be something going on beyond the technology itself that leads to this “acceptability”. In other words, we cannot only focus on characteristics of the technology itself without further context. An empirical example for such disagreements around acceptability can be found in previous research about witness experiences of phone calls on trains. Some passengers find other peoples’ phone calls “annoying” or “intrusive” (Monk et al., 2004a). But in other studies, passengers enjoyed listening in (Love, 2001, Norman & Bennett,2014). Put differently,the same type of interaction can cause opposing experiences for different witnesses, and there is no common agreement among them. Third, as a measurement-oriented approach, the social acceptability literature reduces a social situation to measurable factors that supposedly characterize it. But the relationship between these “measurables” and actual socially situated experiences is not clear. Some of the simpler tools summarize all social situations into one abstract description (e.g., “in public”; Montero et al., 2010)or rely on a single,representative situation (e.g.,“on a busy sidewalk”; Koelle et al., 2018). Similar to the supposed agreement among people described above, such summaries imply some commonality among “social situations” in how they influence situated experiences. However, some interactions such as phone calls can lead to fundamentally different experiences, depending on the social situation. For example, if the grandparents call during a family breakfast, a phone call may lead to positive, shared witness experiences (at least for some family members). In contrast, a phone call in a library could lead to mostly negative experiences (Uhde & Hassenzahl, 2021). Within this measurement-oriented line of thinking, the currently most sophisticated approach to distinguish between social situations is based on measurable proxy markers for situations. Specifically, the “Audience-and-Location Axes” (ALA) differentiate between six “audiences” who witness the interaction (alone, partner, friends, colleagues, strangers, family) and six “locations” where it is performed (home, pavement or sidewalk, driving, passenger on a bus or train, pub or restaurant, workplace; Rico &Brewster, 2010). The idea is that such markers could indicate in which locations and with which audiences an interaction would lead to more positive or negative experiences, and accordingly be more or less acceptable. Following this line of thinking, the way towards improved measures of socially situated experiences would be to add further markers beyond audiences and locations (e.g., daytime, temperature, or noise), until they sufficiently “capture” a social situation. However, this more sophisticated approach is also limited. By sorting audiences and locations into “categories”, it implies a certain coherence within each category and differences between them. But in practice, there can be drastic differences within a category (e.g., two family members can experience the same group call differently). In addition, different categories can overlap (e.g., a family member can also be a colleague). Thus, the meaning of a “family” audience score is not obvious, and its difference to a “colleague” score not necessarily meaningful. The more fundamental critique of this approach is that it attempts to measure social situations as discrete, clearly delimited entities, which are separate from the interaction itself. The interaction is modeled to happen within the situation. An alternative view is to understand the interaction as part of the situation, interwoven with other elements (Dourish, 2004). This brings us to the work on social situatedness grounded in social-interpretivist theories. 2.2 Theoretical Work on Social Situatedness The previous section covered several ways in which it is difficult to simply measure social situations and their relation to user and witness experiences. Here, we will introduce some of the theoretical reasons behind these difficulties. Suchman (2006)outlined many of these in her work on the situatedness of verbal interactions with “intelligent” agents. A key concern she raises is that the idea of a “situation” as a pre-existing and objectively measurable entity, that somehow shapes an interaction, does not match with how actual interactions and experiences relate to each other. Interactions such as verbal conversations often rely on ad hoc actions and reactions. They are based on reciprocal, subjective interpretations and momentary predictions about possible consequences of situated action. Experiences are then shaped retrospectively by post hoc reasoning. For example, the length of a pause between (speech) acts is contingent on other situational characteristics. It can carry various meanings that may change over time and people can interpret them “correctly”or not. In practice, if two speakers start to talk simultaneously after a pause, they will notice the problem and one of them will usually stop talking (Sacks et al., 1974), which allows for a continued, meaningful conversation. The meaning of such situated elements as the length of a pause is not objectively predetermined. It relies on subjective interpretations, anticipations of each other’s interpretations, and ad hoc repair work. We can observe similar performances, interpretations, and repair work with non-verbal interactions as well. For example, Goffman (1959) has developed a reciprocal model of how humans interpret each other’s behavior in their efforts to manage social interactions. In that sense, a seemingly simple performance, such as a mobile phone call on a train, could be interpreted by witnesses to mean many things. It can be a power demonstration or a purposeful violation of social norms. It may also just be caused by the cultural unawareness of a tourist. Or it may be seen as a practice that is acceptable or even enjoyable for other passengers. Likewise, the performer can anticipate and suggest certain “misunderstandings” that work in their interest (e.g., “I’m just a tourist who does not know the rules”), which the witnesses may again anticipate and so forth (see Goffman, 1959,for an elaborate discussion of such processes). In sum, previous work grounded in social-interpretivist perspectives on social interactions highlight their fundamental ambiguity. Different interpretations of why or how someone performs an interaction are crucial to understand 4|Interacting with Computers, 2025 the associated experiences. This makes clear, objective analyses difficult. Dourish (2004) discussed this problem of situatedness further, with a particular focus on research in HCI. He identified two seemingly incommensurable approaches, as illustrated above, grounded in either positivist or phenomenological/socialinterpretivist research paradigms. The positivist approach encompasses measurement-oriented work such as the social acceptability literature. It is mainly concerned with ways to quantify situational elements, to then study their effects on experiences. Such research uses objective measures of concrete situational features (e.g., the ALA). However, its ability to meaningfully predict situated experiences is limited. It considers elements in isolation and overlooks their mutual dependence, and the subjective, interpretational character of interactions and experiences. On the other side, Dourish positions phenomenological / socialinterpretivist approaches, which focus on situated experiences. This research conceptualizes interactions within a mesh of situated elements that may meaningfully relate to each other. However, these “meaningful relations” between elements depend on the interactions themselves. Following this approach, measuring elements as independent “features”to describe a situation is often not useful. To illustrate this point, think of a living room with a carpet full of Lego bricks as a “situational element”. These can be relevant for the (painful) experience of a “walking barefoot” interaction. But they are less relevant for “sleeping on the couch”. In other words, the relevance of this particular element for situated experiences depends on the interaction of interest. With today’s computing power, we could still try to collect all elements and possible interactions in a giant database to create a compendium of mutual relevance. But this leaves the problem of situated, subjective interpretations unsolved. The barefoot walker may wonder who put the Lego bricks there, and whether they represent a deliberate offense, a funny prank, or just carelessness. This interpretation (independent of the “real” story behind the Lego bricks) can change their situated experience, which further complicates how we can study them, least of all “objectively”. Uhde & Hassenzahl (2021) extended this perspective to social situations. They illustrate the social dynamics and meaningful relations between co-located interactions, performed by different people. To that end, they suggest a thought experiment based on a common, “constant” interaction: a mobile phone call. While keeping the phone call itself “unchanged” (at least initially), they construct a range of social scenarios around it, where other people perform various co-located interactions. This highlights how these “witness interactions”, independent of the phone call itself, can affect the experiences it creates. For example, a phone call in a library, where other people read, would usually be experienced as disturbing because it disturbs the reading interactions. But during a rock concert, where most people dance and sing, the “same” phone call would not bother anyone, because it does not affect their situated interactions (while the phone call itself would be strongly disturbed). They argue that the relations between colocated interactions, such as “making a phone call”, “reading”, or “dancing” are key to understanding experiences around interactions with technology. Thus, as a way forward, they suggest moving away from situation categories, such as “library” or “rock concert”, and rather take these as shorthand for “a situation where people tend to read and work”, or “a situation where people dance and sing”. These constellations of interactions determine which other interactions “fit in” and which will be experienced as disturbing. 2.3 Social Situatedness As Constellations of Practices To analyze such constellations of interactions (or “practices”; Reckwitz, 2002,Shove et al., 2012) seems to bring more complexity, compared with simple situation categories. If we want to study how people experience, say, phone calls on a train, we now need to consider all other co-situated interactions, find out how each of them relates to the phone call, and how people interpret each other’s interactions. This will potentially produce ambiguous data (e.g., phone calls have both positive and negative effects on witness experiences), which we need to integrate into something useful. In addition, the specific constellation of interactions can change momentarily, which again transforms situated experiences. Thus, this analysis confronts us with practical challenges. But this increased complexity also helps overcome problems elsewhere. Situation categories cannot account for intra-category differences (e.g., silent vs. regular compartments on a train), intercategory similarities (e.g., a train that is also someone’s workplace), and category overlaps (e.g., a restaurant on a train). But the social-interpretivist approach can.Similarities and differences can be described based on shared and distinct patterns within the constellation of practices. For example, silent and regular compartments share certain practices (e.g., sitting, working), but not others (e.g., making a phone call). If we introduce a new interaction with technology and study how it relates to these interactions on the train, the specific compatibilities and conflicts can serve as indicators for situated experiences that can transfer to other train situations. For example, the insight that the new interaction conflicts with specific silent compartment practices, but not with the regular compartment practices, can be valuable for designers. In sum, we have outlined two different approaches to studying socially situated interactions with technology and their associated experiences. Both come with their own challenges. The measurement-oriented approach provides easily obtainable data, but it is often unclear what these data indicate about socially situated experiences. The social-interpretivist work highlights ways in which interactions and experiences are embedded in a mesh of situated elements and interpretations. However, there is a lack of tools to make such a social-interpretivist perspective accessible for practitioners to inform design decisions and understand experiential effects of a concrete, socially situated interaction with technology. The Witness Experience Inventory (WEI) we introduce in the following is designed to address this gap. 3THE WITNESS EXPERIENCE INVENTORY Given the practical difficulties of the social-interpretivist approach and the problems with existing measurement-oriented approaches, we first set a scope that we think a pragmatic tool can meaningfully address to produce useful findings. To that end, we made the following decisions for the scope of the WEI: 1) The WEI focuses on understanding the relationships between situated elements—primarily co-located interactions of users and witnesses—and highlights subjective interpretations. 2) Instead of prescribing a fixed set of social situations, we start from the experiential outcome. Witnesses describe situations that led to particularly positive and negative experiences in relation to an interactive technology. From there, we try to understand how these experiences come about Uhde et al. |5 and which situational characteristics seem most essential for that experience. 3) The goal is not to derive an overall “witness experience score”. Instead, the WEI highlights the variety of different witness experiences in relation to an interaction with technology, and experiential risks and opportunities. 4) The goal is also not to develop an all-encompassing, generalizable assessment of socially situated, technology-mediated experiences. Instead, we acknowledge the specificities of each situation and interaction with technology, which may be important for how witnesses experience it. 5) Finally, the WEI should encourage reflection on possible design decisions and their impact on different (groups of) witnesses. Often, such decisions have positive effects for some groups of witnesses, but negative effects for others. The WEI should make such trade-offs apparent and support reflection. In line with this scope, the WEI is based on a qualitative research approach using interviews. This helps examine the detailed relations between a socially situated interaction with technology and its surroundings, and keeps it flexible for unexpected situational characteristics. It also helps understand the reasons behind positive, negative, and ambiguous relations among interactions, to later inform design decisions. We selected and adapted a specific interview method (IPA; Smith et al., 2009), because it is already established in experienceoriented HCI research (e.g., Laschke et al., 2013,Uhde et al., 2021). It is easy to use with limited resources and a small sample, and flexible to be adapted for various situations and interactions of interest. For our initial study, it also allowed us to collect data in the same way as intended for future uses of the WEI, while bootstrapping the coding template described below to facilitate future analyses. As a general format, the WEI focuses on positive and negative witness experiences of an interaction with technology. Participants first describe their own experience as witnesses. Then we collect a list of other co-located (witness) practices and ask participants to relate them to the interaction, to learn about potential other witness experiences. This format corresponds with the social-interpretivist perspectives outlined above (e.g., Dourish, 2004,Uhde & Hassenzahl, 2021). 3.1 Interview Study: Overview To turn the theoretical considerations into a practical tool and test its usefulness, we ran an interview study following the WEI interview format as described below (see also Figure 1). This study served three main goals: Study Goal 1: To create initial empirical data about the complex relations between interactions with technology and witness experiences Study Goal 2: To test and refine the WEI interview procedure, and to bootstrap the coding template, which can be used as an informed starting point for the evaluation of future work Study Goal 3: To test how the WEI can be used to produce insights for (re)designing socially situated technology 3.2 Method 3.2.1 Participants We recruited 15 participants through an agency (female =8, male =7, meanage =34.73,sdage =14.57,rangeage =[18, 60]) in a large city in Germany and ran the interviews in one of Fig. 1. The procedure for the WEI, as used in the study and with some recommendations for future uses: We had no restricted number of witness practices in our study and ran a full template analysis with a minimal a priori template. In our study, we repeated the procedure with each participant for two to three interactions. their interview rooms. Participants had various backgrounds and occupations, including office clerks, high school and university students, an electrician, and a project coordinator. The interviews lasted for around 47 minutes on average. 3.2.2 Procedure The interviewer welcomed the participants to the interview room, briefly explained the purpose of the study and his interest in how people experience technologies in social situations. All participants then signed a consent form that informed them about the anonymous data analysis, video and audio recording, and their right to quit the interview at all times without negative consequences. The interviews followed a semi-structured format, focused on two example interactions with technology and their associated witness experiences. We also prepared one backup interaction in case participants had insufficient experience with one of the examples and could not describe how they think it would affect witness experiences. We counterbalanced the choice of interactions between participants. For each interaction, we asked 6|Interacting with Computers, 2025 Fig. 2. Four index cards about interactions with drones and attached sticky notes with comments about co-situated interactions. participants to think about personal experiences they had as a witness. They described their thoughts, feelings, and behavior by reporting their experiences, and through follow-up questions in a laddering format (Reynolds & Gutman, 1988). For each interaction, the interviewer showed the participants two photos of people interacting with the technology and asked if they are familiar with this interaction. The photos served as a simple illustration to clarify the interaction of interest. If participants were familiar with it, he asked them to describe a situation based on their own experience where this interaction with technology has led to positive experiences for the surrounding people. If they had no such experience (e.g., no positive witness experience with a mobile phone call), they could also describe a positive scenario they could think of. Otherwise, the interviewer moved on to the next case and used the backup interaction if there was enough time left. While the participant gave an initial description of the positive experience and social situation, the interviewer noted down the label for the situation used by the participant (e.g., “café”) on an index card and then asked them to name the five or six most typical practices people perform in this kind of situation. The interviewer then wrote down each of these practices on a sticky note and attached it to the index card (see Figure 2). In the next step, he went through each of these co-situated practices one by one and asked the participant how they think this practice relates to the interaction with technology. Although the initial experience for this situation was positive, the relation to individual practices was sometimes described as neutral or negative. The interviewer then asked follow-up questions to better understand how the participant makes sense of this relation between the interaction and each of the other practices, and the associated experiences that unfold. After the participant had finished their description, the interviewer summarized how he understood the relations and checked whether that was in line with the participant’s understanding. He then asked the participant whether they see a way the interaction could be done differently to reduce negative or emphasize positive witness experiences. Finally, he asked the participant to name similar situations that would lead to similar witness experiences. In total, the description of one situation typically took around 10 to 15 minutes. After this positive experience, the interviewer repeated the same procedure for the same interaction, but this time asked the participant to describe a negative witness experience. Finally, he repeated the same procedure for the second interactive technology and, if necessary, with the backup technology. After the interview, the interviewer gave a short debriefing and answered further questions by the participant before thanking them for their participation and ending the interview session. 3.2.3 Material: Eight Interactive Technologies We selected interactive technologies for this study to cover a large variety of interactions and associated witness experiences. To that end, we defined two dimensions and looked for examples with some variation on them (see Figure 3). The first dimension was meant to create some variety in “novelty” of interactions, because previous work around early and late adopters of technology indicated that the temporal dimension may have an effect on how people experience an interaction (e.g., Montero et al., 2010). In addition, we selected a range of technologies between more “inward” oriented and more “outward” oriented examples. Such outward-orientation or “purposeful visibility” of an interaction from the witness perspective has been studied before as a central determinant of their situated experiences (Reeves et al., 2005). Taken together, our selection included a mix of less common, more “inward” oriented technologies such as Virtual Reality (VR) glasses, more “outward” but unusual technologies like a drone, and more established technologies like a smartphone, laptop, and camera. 3.2.4 Analysis To analyze the interviews, we first transcribed them in German. These transcripts served as the data set for a Template Analysis conducted by the first and second author. Template Analysis (King, 2004,King et al., 2018) is a “codebook” variant of Thematic Analysis (Braun & Clarke, 2006,2021), with a particular approach to the development of a coding template. It is usually recommended for very large samples where the template can streamline the coding process. However, it seemed particularly useful in our case for three reasons. First, Template Analysis offers an established, iterative process that alternates between individual coding sessions and follow-up discussions. This allowed us to integrate two individual readings of the material (by the two coders) into an agreed version. Second, the Template Analysis process converges into a final template, which serves as a “semisolid” outcome of the analysis (see Figure 4). In our study, this template covered the situational aspects that seemed relevant for witness experiences of the various example interactions in an urban context in Germany. As a side benefit, this template already covers a solid baseline for future applications of the WEI, with many relevant aspects. Nevertheless, the template remains easily adaptable, for example when used in different cultural settings or with different interactive technologies. Thus, we consider this semi-flexibility a useful trade-off between openness (which is necessary given the variety of social situations) and “informed defaults”. Third, Template Analysis can be conducted with an optional “a priori template” in case the researchers already have particular guiding questions. In our case, we had an initial interest in positive and negative witness experiences, situations that seem especially (un-)suitable for the interactions, and how users and witnesses react to the interactions. We were also interested in aspects that concern the technology design and potential ambiguous experiences. Our a priori template included codes according to these interests as a starting point (see Uhde et al. |7 Fig. 3. The eight example technologies used in the study, roughly aligned on the two axes from “inward” to “outward” orientation, and from “established, typical” to “unusual”. The photos in the figure are the ones used as illustrative material in the study. Uhde et al., 2025, or supplementary material for details about the template development). During the analysis, the two coders first used the a priori template to separately code three interviews that covered all eight example interactions. After this first round, they held a debriefing session to discuss unclear text passages and any changes to the template, which led to a first iteration of the template. Then they each coded three further interviews, agreed on a second iteration of the template, and repeated this step another time to develop a third iteration based on all 15 interviews. Finally, the first author used this updated template to code all 15 interviews a second time. This led to a few minor changes both coders subsequently discussed and agreed on for the final version of the template. 4RESULTS The final template (see Figure 4 for an overview) provides detailed descriptions of various situational elements that can be relevant for shaping witness experiences. Here, we summarize the most central aspects from our study, given the specific technologies and cultural settings, in three key themes. Following this general summary, we describe three design cases to illustrate how the findings from the WEI can be used to inform design decisions, and which align with these themes. 4.1 Overall Analysis 4.1.1 Ambiguities We have already argued that interactions with technology can lead to different witness experiences. For example, some people find phone calls disturbing (Monk et al., 2004a)and others like to listen in (Love, 2001). Our study provided further insights into these ambiguous witness experiences. We distinguish between witness-related and user-related aspects of ambiguity. Witness-related aspects: individual differences, mood, social roles, and group memberships. One reason for the ambiguity simply relates to individual differences among witnesses. For example, one participant described a situation where she made a phone call in a café. A woman nearby was notably annoyed by this: “she sat with her back towards me, and she was very annoyed by my phone call. So she turned around again and again” [I1]. In a different situation, with a different witness but also in a café, the witness experience was quite different. The participant made a phone call about her son who had some problems in school: “then she listened in and heard all of that, although I thought nobody would hear that. And then she somehow talked to me and we ... talked about it ... we somehow became friends over this, she also had a similar issue with her daughter and yes ... that seems to have come across as ... positive” [I1]. Relatedly, the current mood of the witness can affect their experience. One participant described a scenario of a phone call on a train: “I sometimes notice that, when I make a phone call somehow, you look at the other passengers a bit, you know? How does he look, is he relaxed, is he a bit grumpy? [...] People’s frustration tolerance can be a bit problematic sometimes, you know? I see that again and again, you wait for the train in Frankfurt, you know? And the train doesn’t come or is 30 minutes delayed, people are already fuming, so it’s a bit difficult uh ... to go about your business, make phone calls etc.” [I10]. Individual differences and interpretations of other people’s mood can be difficult for designers to address. By redesigning the phone call interaction in these examples, designers can support either side. They can side with the witnesses who feel disturbed, by making the interaction less notable (see e.g., Kimura et al., 2019, Li et al., 2019). But they can also support the positive experiences (e.g., Desmet & Hassenzahl, 2012). It may be difficult to develop a solution that caters for both groups of witnesses. 8|Interacting with Computers, 2025 Fig. 4. Overview of the themes, subthemes, and subsubthemes of our analysis, which are included in the coding template. See Uhde et al. (2025)orthe supplementary material for full code descriptions and further details. A different type of ambiguity relates to the different social roles of witnesses. One participant described how he was responsible for setting up VR glasses to attract potential new employees during a job fair. The candidates could use the VR glasses and others could watch them. But his own witness experience was different because of his role: “if I am part of the crew at the stand, I need to make sure that nothing happens to the future candidates. In that case I have a different task than the interested visitors who look for a job and see ‘ah, they do something over there, that seems fun, I want to have a look!’ So it’s different, depending on the target groups” [I6]. This type of ambiguity can be easier to address. In this case, designers may prioritize the experience of the candidates over the employees. They are also not in an inherent conflict as in the phone call example above, and there may be straightforward ways to design a positive experience for both. Uhde et al. |15 the WEI brings these elements back together. Through the relations of for example the interaction and surrounding practices, it can uncover relevant situational characteristics that underpin witness experiences. We hope that this work inspires further research that takes a social-interpretivist stance on socially situated experiences and helps practitioners integrate this perspective in their design process. DATA AVAILABILITY STATEMENT The (translated) interview guide and coding template can be found in Uhde et al. (2025) and the supplementary material. The interview recordings and transcripts will not be shared publicly to preserve the participants’ anonymity. Please contact the first author for possibilities to access these data. Acknowledgments This project was in part funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Grant No. 425827565, as part of Priority Program SPP2199 Scalable Interaction Paradigms for Pervasive Computing Environments. It was finalized with funding by Tokyo College, UTIAS, The University of Tokyo. We would like to thank our colleagues at Siegen University and at Tokyo College for their valuable feedback, and all participants for their participation. References Ahlström, D.,Hasan, K. and Irani, P. (2014) Are you comfortable doing that? Acceptance studies of around-device gestures in and for public settings. In Proceedings of the 16th International Conference on Human–Computer Interaction with Mobile Devices & Services - MobileHCI’14, pp. 193–202. ACM, New York, NY, USA. Bajorunaite, L.,Brewster, S. and Williamson, J. R. (2023) Reality anchors: bringing cues from reality to increase acceptance of immersive technologies in transit. In Proceedings of the ACM on Human–Computer Interaction (MobileHCI), pp. 1–28. ACM, New York, NY, USA. Baumer,E.P.and Brubaker, J. R. (2017) Post-userism. In Proceedings of the 2017 ACM Conference on Human Factors in Computing Systems, pp. 6291–6303. ACM, New York, NY, USA. Braun, V. and Clarke, V. (2006) Using thematic analysis in psychology. Qualitative Research in Psychology,3, 77–101. https://doi. org/10.1191/1478088706qp063oa. Braun, V. and Clarke, V. (2021) One size fits all? What counts as quality practice in (reflexive) thematic analysis? Qualitative Research in Psychology,18, 328–352. https://doi. org/10.1080/14780887.2020.1769238. Callahan-Lyon, P. (2014) Electronic cigarettes: human health effects. Tobacco Control,23, 36–40. Connelly, S.,Richardson, T. and Miles, T. (2006) Situated legitimacy: deliberate arenas and the new rural governance. Journal of Rural Studies,22, 267–277. https://doi.org/10.1016/j. jrurstud.2005.11.008. da Silva, C. L. C. A.,Dakafay, H. M.,O’Brien, K.,Montierth, D.,Xiao, N. and Ojcius,D.M.(2021) Effects of electronic cigarette aerosol exposure on oral and systemic health. Biometrical Journal,44, 252–259. https://doi.org/10.1016/j.bj.2020.07.003. Desmet, P. and Hassenzahl, M. (2012) Towards Happiness: PossibilityDriven Design. In Zacarias, M., de Oliveira, J. V. (eds), Human– Computer Interaction: The Agency Perspective, vol. 396, pp. 3–27. Springer Berlin, Heidelberg, Berlin, Heidelberg. Dey, A. K. (2001) Understanding and using context. Personal and Ubiquitous Computing,5,4–7.https://doi.org/10.1007/s007790170019. Dourish, P. (2004) What we talk about when we talk about context. Personal and Ubiquitous Computing,8, 19–30. https://doi. org/10.1007/s00779-003-0253-8. Goffman, E. (1959) The Presentation of Self in Everyday Life. Anchor Books, New York, NY, USA. Hassenzahl, M. (2010) Experience design: technology for all the right reasons. Synthesis Lectures on Human-Centered Informatics,3, 1–95. https://doi.org/10.1007/978-3-031-02191-6. Hassenzahl, M.,Diefenbach, S. and Göritz, A. (2010) Needs, affect, and interactive products—facets of user experience. Interacting With Computers,22, 353–362. https://doi.org/10.1016/j. intcom.2010.04.002. Hassenzahl, M. and Tractinsky, N. (2006) User experience—a research agenda. Behaviour and Information Technology,25, 91–97. https:// doi.org/10.1080/01449290500330331. Janlert, L.-E. and Stolterman, E. (2010) Complex interaction. ACM Transactions on Computer–Human Interaction,17,1–32.https://doi. org/10.1145/1746259.1746262. Kimura, N.,Kono, M. and Rekimoto, J. (2019) Sottovoce: an ultrasound imaging-based silent speech interaction using deep neural networks. In Proceedings of the ACM Conference on Human Factors in Computing Systems, p. 11. ACM, New York, NY, USA. King, N. (2004) Using templates in the thematic analysis of text. In Cassell, C., Symon, G. (eds), Essential Guide to Qualitative Methods in Organizational Research, pp. 256–270. SAGE Publications Ltd., Los Angeles, CA, USA. King, N.,Brooks, J. and Tabari, S. (2018) Template analysis in business and management research. In Ciesielska, M., Jemielniak, D. (eds), Qualitative Methodologies in Organization Studies, pp. 179–206. Springer, Berlin, Heidelberg, Germany. Koelle, M.,Ali, A. E.,Cobus, V.,Heuten, W. and Boll, S. (2017) All about acceptability? Identifying factors for the adoption of data glasses. In Proceedings of the 2017 ACM Conference on Human Factors in Computing Systems, pp. 295–300. ACM, New York, NY, USA. Koelle, M.,Ananthanarayan, S. and Boll, S. (2020) Social acceptability in HCI: a survey of methods, measures, and design strategies. In Proceedings of the 2020 ACM Conference on Human Factors in Computing Systems, pp. 1–19. ACM, New York, NY, USA. Koelle, M.,Ananthanarayan, S.,Czupalla, S.,Heuten, W. and Boll, S. (2018) Your smart glasses’ camera bothers me! Exploring opt-in and opt-out gestures for privacy mediation. In Proceedings of the 10th Nordic Conference on Human–Computer Interaction— NordiCHI’18, pp. 473–481. ACM, New York, NY, USA. Koelle, M.,Kranz, M. and Möller, A. (2015) Don’t look at me that way! Understanding user attitudes towards data glasses usage. In Proceedings of the 17th International Conference on Human–Computer Interaction with Mobile Devices and Services—MobileHCI’15,pp. 362–372. ACM, New York, NY, USA. Korsgaard, H.,Lyle, P.,Saad-Sulonen, J.,Klokmose, C. N.,Nouwens, M. and Bødker, S. (2022) Collectives and their artifact ecologies. Proceedings of the ACM on Human–Computer Interaction,6, 1–26. https://doi.org/10.1145/3555533. Laschke, M.,Hassenzahl, M.,Brechmann, J.,Lenz, E. and Digel, M. (2013) Overcoming procrastination with remind. In Proceedings of Designing Pleasurable Products and Interfaces (DPPI’13), pp. 77–85. ACM, New York, NY, USA. Li, R.,Wu, J. and Starner, T. (2019) Tongueboard: an oral interface for subtle input. In Proceedings of the 10th Augmented Human International Conference, pp. 1–9. USA. ACM, New York, NY. Love,S.(2001) Space invaders: Do mobile phone conversations invade peoples’ personal space? In Proceedings of the 18th International 16 |Interacting with Computers, 2025 Symposium on Human Factors in Telecommunications, pp. 125–131. Bergen, Norway. McCann, W.,Pedneault, A.,Stohr, M. K. and Hemmens, C. (2017) Upskirting: a statutory analysis of legislative responses to video voyeurism 10 years down the road. Criminal Justice Review,43, 1–20. Mitchell, R. and Olsson, T. (2017) Barriers for bridging interpersonal gaps: three inspirational design patterns for increasing collocated social interaction. In Proceedings of the 8th International Conference on Communities and Technologies, pp. 2–11. ACM, New York, NY, USA. Mitchell, R. and Olsson, T. (2019) Facilitating the first move: exploring inspirational design patterns for aiding initiation of social encounters. In Proceedings of the 9th International Conference on Communities and Technologies, pp. 283–294. ACM, New York, NY, USA. Monk, A.,Carroll, J.,Parker, S. and Blythe, M. (2004a) Why are mobile phones annoying? Behaviour and Information Technology,23, 33–41. https://doi.org/10.1080/01449290310001638496. Monk, A.,Fellas, E. and Ley, E. (2004b) Hearing only one side of normal and mobile phone conversations. Behaviour and Information Technology,23, 301–305. https://doi.org/10.1080/01449 290410001712744. Montero, C. S.,Alexander, J.,Marshall, M. T. and Subramanian, S. (2010) Would you do that? Understanding social acceptance of gestural interfaces. In Proceedings of the 12th International Conference on Human Computer Interaction with Mobile Devices and Services—MobileHCI’10, pp. 275–278. ACM, New York, NY, USA. Norman, B. and Bennett, D. (2014) Are mobile phone conversations always so annoying? The ‘need-to-listen’ effect re-visited. Behaviour and Information Technology,33, 1294–1305. https://doi. org/10.1080/0144929X.2013.876098. Olsson, T.,Jarusriboonchai, P.,Wo ´ zniak, P.,Paasovaara, S.,Väänänen, K. and Lucero, A. (2020) Technologies for enhancing collocated social interaction: review of design solutions and approaches. Computer Supported Cooperative Work,29, 29–83. https://doi. org/10.1007/s10606-019-09345-0. Pohl, H.,Muresan, A. and Hornbæk, K. (2019) Charting subtle interaction in the HCI literature. In Proceedings of the 2019 ACM Conference on Human Factors in Computing Systems. ACM, New York, NY, USA. Reckwitz, A. (2002) The status of the “material” in theories of culture: from “social structure” to “artefacts”. Journal for the Theory of Social Behaviour,32, 195–217. https://doi.org/10. 1111/1468-5914.00183. Reeves, S. (2011) A framework for designing interfaces in public settings. In Reeves, S. (ed.), Designing Interfaces in Public Settings, pp. 141–175. Springer, London, UK. Reeves, S.,Benford, S.,O’Malley, C. and Fraser, M. (2005) Designing the spectator experience. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems—CHI’05, pp. 741–750. ACM, New York, NY, USA. Reynolds, T. J. and Gutman, J. (1988) Laddering theory, method, analysis, and interpretation. Journal of Advertising Research,28, 11–31. Rico, J. and Brewster, S. (2010) Usable gestures for mobile interfaces: evaluating social acceptability. In Proceedings of the 2010 ACM Conference on Human Factors in Computing Systems, pp. 887–896. ACM, New York, NY, USA. Sacks,H.,Schegloff, E. A. and Jefferson, G. (1974) A simplest systematics for the organization of turn-taking for conversation. Language, 50, 696–735. https://doi.org/10.1353/lan.1974.0010. Schmidt, A.,Beigl, M. and Gellersen, H.-W. (1999) There is more to context than location. Computer Graphics,23, 893–901. https://doi. org/10.1016/S0097-8493(99)00120-X. Sharon, T. and Koops, B.-J. (2021) The ethics of inattention: revitalising civil inattention as a privacy-protecting mechanism in public spaces. Ethics and Information Technology,23, 331–343. https://doi. org/10.1007/s10676-020-09575-7. Shove, E.,Pantzar, M. and Watson, M. (2012) The Dynamics of Social Practice: Everyday Life and How It Changes. SAGE Publications Ltd., Los Angeles, CA, USA. Smith, J.,Flowers, P. and Larkin, M. (2009) Interpretative Phenomenological Analysis: Theory, Method and Research: Understanding Method and Application. SAGE Publications Ltd., New York, NY, USA. Sramek, Z.,Sato, A. J.,Zhou, Z.,Hosio, S. and Yatani, K. (2023) Soundtraveller: exploring abstraction and entanglement in timbre creation interfaces for synthesizers. In Proceedings of the ACM Conference on Designing Interactive Systems (DIS’23), pp. 95–114. ACM, New York, NY, USA. Suchman, L. (2006) Human–Machine Reconfigurations: Plans and Situated Actions (2nd). Cambridge University Press, New York, NY, USA. Uhde, A.,Dreyer, L. and Hassenzahl, M. (2025) Witness Experience Inventory (WEI). Coding Template and Interview Guide. Uhde, A. and Hassenzahl, M. (2021) Towards a better understanding of social acceptability. In Proceedings of the ACM CHI Conference on Human Factors in Computing Systems Extended Abstracts. ACM, New York, NY, USA. Uhde, A.,Laschke, M. and Hassenzahl, M. (2021) Design and appropriation of computer-supported self-scheduling practices in healthcare shift work. Proceedings of the ACM on Human–Computer Interaction,5,1–26.https://doi.org/10.1145/3449219. Uhde, A.,Zum Hoff, T. and Hassenzahl, M. (2022) Obtrusive subtleness and why we should focus on meaning, not form, in social acceptability studies. In Proceedings of the 21st International Conference on Mobile and Ubiquitous Multimedia (MUM’22),p.11. ACM, New York, NY, USA. Uhde, A.,Zum Hoff, T. and Hassenzahl, M. (2023) Beyond hiding and revealing: exploring effects of visibility and form of interaction on the witness experience. Proceedings of the ACM on Human–Computer Interaction (MobileHCI),7, 23. © The Author(s) 2025. Published by Oxford University Press on behalf of The British Computer Society. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Interacting with Computers,2025, 1–16 https://doi.org/10.1093/iwcomp/iwaf010 Article