Stephan Habscheid, Tim Hector, Dagmar Hoffmann, David Waldecker (eds.) Voice Assistants in Private Homes Media in Action 7 Editorial The open access book series “Media in Action”, conceived by the DFG Collaborative Research Centre “Media of Cooperation”, examines the history and present of net- worked, data-intensive media and their social implications at the interdisciplinary interface of social and media studies. In the tradition of science and technology studies and actor-network theory, German and English monographs, edited vol- umes and dissertations of the series focus on practices, (co-)operations and proce- dures in the use, production and analysis of old and new media. A central challenge the series faces is the development of appropriate ethnographic, digital, sensor- based and design-oriented methods for a new conception of the description of dis- tributed agency between people, computers, bodies and environments. Funded by the German Research Foundation (DFG) – Project number 262513311 – SFB 1187. Stephan Habscheid (Prof. Dr.) is a professor of German studies and applied lin- guistics at Universität Siegen. He is principal investigator of the interdisciplinary project “Un/desired Observation in Interaction: Smart Environments, Language, Body and Senses in Private Homes” at the Collaborative Research Center 1187 “Me- dia of Cooperation”, Universität Siegen (together with Dagmar Hoffmann). His re- search interests include media linguistics, linguistic praxeology, language in insti- tutions and organizations as well as small talk and conversation. Tim Hector (M.A.) works as a research assistant at the Collaborative Research Cen- ter 1187 “Media of Cooperation” in the project “Un/desired Observation in Interac- tion: Smart Environments, Language, Body and Senses in Private Homes” at Uni- versität Siegen. He did a PhD in applied linguistics on the linguistic domestication of voice assistants in private homes. His research is focussed on media linguistics, conversation analysis and linguistic praxeology. Dagmar Hoffmann (Prof. Dr.) is a professor of media sociology and gender media studies at Universität Siegen, Germany. She is principal investigator in the inter- disciplinary project “Un/desired Observation in Interaction: Smart Environments, Language, Body and Senses in Private Homes” at the Collaborative Research Center 1187 “Media of Cooperation”, Universität Siegen (together with Stephan Habscheid). Her research is focused on media and cultural sociology, digital literacy, and polit- ical participation. David Waldecker (Dr.) is a sociologist and an academic librarian in training at Universitäts- und Landesbibliothek Darmstadt. He was a post-doc at the Col- laborative Research Center 1187 “Media of Cooperation”, Universität Siegen, and published his dissertation on Adorno in the recording studio in 2022. Stephan Habscheid, Tim Hector, Dagmar Hoffmann, David Waldecker (eds.) Voice Assistants in Private Homes Media, Data and Language in Interaction and Discourse Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foun- dation) – Project-ID 262513311 – SFB 1187 “Media of Cooperation”. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbib- liografie; detailed bibliographic data are available in the Internet at https://dnb.dn b.de This work is licensed under the Creative Commons Attribution-Non Commercial 4.0 (BY-NC) license, which means that the text may be may be remixed, build upon and be distributed, provided credit is given to the author, but may not be used for commercial purposes. For details go to: https://creativecommons.org/licenses/by-nc/4.0/ Permission to use the text for commercial purposes can be obtained by contacting rights@transcript-publishing.com Creative Commons license terms for re-use do not apply to any content (such as graphs, figures, photos, excerpts, etc.) not original to the Open Access publication and further permission may be required from the rights holder. The obligation to research and clear permission lies solely with the party re-using the material. 2025 © Stephan Habscheid, Tim Hector, Dagmar Hoffmann, David Waldecker (eds.) transcript Verlag | Hermannstraße 26 | D-33602 Bielefeld | live@transcript-verlag.de Cover design: Maria Arndt Cover Illustration: Bence Boros / Unsplash Proofreading: Pip Hare Printing: Elanders Waiblingen GmbH, Waiblingen https://doi.org/10.14361/9783839472002 Print-ISBN: 978-3-8376-7200-8 PDF-ISBN: 978-3-8394-7200-2 ISSN of series: 2749-9960 eISSN of series: 2749-9979 Printed on permanent acid-free text paper. https://dnb.dnb.de https://dnb.dnb.de https://doi.org/10.14361/9783839472002 Contents Voice Assistants in Private Homes. Introduction to the Volume Stephan Habscheid, Dagmar Hoffmann, Tim Hector, and David Waldecker ...............9 I Voice Assistants in Private Homes. Conceptual Considerations The DataEconomy@Home The Private Sphere, Privacy, and the Embedding of Artificial Intelligence Systems into Everyday Life as an Expansion of Economic Data Grabbing Carsten Ochs ....................................................................... 33 Voice Assistants, Capitalism, and the Surveillance of Social Reproduction Markus Kienscherf .................................................................. 57 Machines as Partners Anthropomorphism and Communication Accommodation to Voice Assistants in Disability Contexts Caja Thimm, Phillip Engelhardt, and Julia Schmitz ....................................77 Human-Machine Interaction as a Complex Socio-Linguistic Practice Netaya Lotze .......................................................................105 II Linguistic Exchange with Voice Assistants as a Practical Problem “Oh, Now I have to Speak” Older Adults’ First Encounters with Voice-based Applications in Smartphone Courses Florence Oloff ...................................................................... 147 Stylizing the Ideal User Insights into the Experiences of Turkish-Speaking Voice Assistant Users in Germany Didem Leblebici .................................................................... 181 Linguistic Practices as a Means of Domesticating Voice-Controlled Assistance Technologies Stephan Habscheid, Tim Hector, and Christine Hrncal ............................... 207 III Privacy and Data Protection as Practical Problems Glitch Studies and Smart Speakers A Spotlight on User Experiences of Unexpected Behaviors Christoph Lutz and Gemma Newlands ............................................... 243 The Role of Imagined Sociotechnical Affordances in Shaping Experiences of Privacy in Smart Speakers Jasper Vermeulen and Anouk Mols ................................................. 263 Mostly Harmless? Everyday Smart Speaker Use and Pragmatic Fatalism David Waldecker, Alexander Martin, and Dagmar Hoffmann ...........................291 How to Make GDPR a Threat Again Nikolai Horn in Conversation with Dagmar Hoffmann and David Waldecker .........319 IV Technical Infrastructures as a Practical Problem Demystification of Technology Empowering Consumers to Access and Visualize Voice Interaction Data Dominik Pins, Fatemeh Alizadeh, Alexander Boden, Sebastian Zilles, and Gunnar Stevens ................................................................331 Innovating Alexa amid the Rise of Large Language Models Sociotechnical Transitions in Algorithmic Development Practices Niklas Strüver ..................................................................... 365 List of Authors ................................................................... 403 Voice Assistants in Private Homes. Introduction to the Volume Stephan Habscheid, Dagmar Hoffmann, Tim Hector, and David Waldecker 1. The Emergence of Voice Assistants In 2011, together with the new iPhone 4S, Apple launched a voice assistant called “Siri”1, which it claimed could understand questions and commands in spoken language (initially in English, German, and French) and respond to them as a human conversation partner would (Huq 2011). The announcement was met with great fascination: here was a talking technology, the first “in- telligent personal assistance” system (IPA) to promise to make everyday life easier. With IPAs, it looked like a very popular, prototypical motif of science fiction was finally to become reality (Stresing 2011)2. Just a few years after “Siri”, Amazon followed suit with “Alexa” (2015) and Google with its “Google As- sistant” (2016) (cf. Dürscheid 2023), and by the end of the decade, the systems had become increasingly established in private households (Statista 2021). As well as in smartphone apps, voice assistants have been finding their way into various everyday devices, such as smart speakers, smart TVs, smart watches, or the media interfaces of digitally connected cars. In recent years, however, the high-flying economic and technological hopes initially pinned to voice assistant technologies have been critically reappraised. In 2022, an article in Business Insider asserted that billions of invested dollars 1 Apple (under CEO Steve Jobs) had bought the company of the same name, which had been founded in 2007 and developed the product in 2010 (Wikipedia 2024; see also Dürscheid 2023). 2 This is not the place to retell the media history of talking machines (see, e.g., Volmar 2019). For a detailed media theoretical and linguistic description of the technologies relevant here with a specific focus on smart speakers, see Hector (in preparation). 10 Voice Assistants in Private Homes had been lost, and hoped-for profits had not been achieved – a “colossal fail- ure of imagination”, in the words of a former employee of Amazon (Kim 2022). According to the Business Insider article, most of those conversations were trivial, commands to play music or ask about the weather. That meant fewer opportunities to monetize. Amazon can’t make money from Alexa telling you the weather – and playing music through the Echo gives Amazon only a small piece of the proceeds. (Kim 2022) In addition, since the introduction of new language processing technolo- gies such as text-generating ChatGPT, which was launched in 2022, the earlier voice assistant technology risks appearing unwieldy in comparison (Dürscheid 2023). Against this backdrop, Amazon and other companies are apparently trying to integrate modern generative AI into older voice assistance systems. According to press reports, an assistance system presented by Amazon in 2023 spoke in a far more natural and conversational voice than the friendly-but-robotic one that hundreds of millions have become accustomed to communicating with for weather updates, reminders, timers, and music requests. (Goldman 2024) According to the reports, this ‘new Alexa’ engaged more naturally in conver- sations, delivered more natural voice output, and had a more pronounced personality (ibid.). However, it seems that the version demonstrated has not yet been convincingly implemented into the real performance of the systems (ibid.). Thus, the American magazine Fortune has claimed that Amazon and Apple – once pioneers in the development of talking machines – are now “desperately behind [their] Big Tech rivals Google, Microsoft, and Meta in the race to launch AI chatbots and agents, and floundering (in [their] efforts to catch up)” (Goldman 2024). One reason given for this is that the characteristic technological architecture of older voice assistants is required to retain cer- tain characteristics in order to maintain existing features, but therefore is no longer up to date enough for the integration of recent AI. In addition, these circumstances make it difficult to collect or synthetically produce suitable lin- guistic training data for the further development of the voice assistants. Citing former employees, the article reports that Amazon has therefore repeatedly Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 11 deprioritized the further development of Alexa to focus on the development of generative AI for its cloud computing unit (“Amazon Web Services”) – which could see the existing technology soon becoming a “digital relic” (ibid., see also Herbig 2024). On the other hand, current usage studies show that the number of de- vices with voice user interfaces for different smart technologies is continually increasing. Amazon has not confirmed the reports about Alexa’s economic failure, and it is evidently continuing to invest in such products (Amazon 2023). For example, further development of devices that combine voice user interfaces with camera, monitor, and touch interfaces seems to be ongoing. As Niklas Strüver (2023) points out, smart speakers are conceived as the central hub for the smart home – a field of consumer tech that is clearly continuing to gain ground. Thus, in the smart home, devices like smart speakers are what allow users to manage the entire orchestration of multiple interconnected smart home applications related to the kitchen, housekeeping, or security. While technology companies see internet-enabled devices in the home as a way to increase demand for many such products and associated services, crit- ics point out that many of the devices are too expensive for most consumers and will take years to catch on. Either way, there are ample reasons to examine language-processing ma- chines and their future development from the perspective of interaction re- search and linguistics. Not only does human–machine dialogue offer a fruitful field for investigation, it also points to potential new approaches to research on human–human interactions, as Karola Pitsch (2015) has shown with the example of co-constructions: familiar conversational procedures are “broken open”, making analytical access to more basic conversational phenomena pos- sible. Furthermore, as Martin Porcheron, Joel Fischer, Stuart Reeves, and Sarah Sharples observed in 2018, social interaction among co-present participants changes when the use of machines is incorporated. The linguistic contribu- tions in this volume address human–machine communication as well as hu- man–human communication and can be read together as an overview of cur- rent research in this field. However, many other academic disciplines also address the phenomenon of human–computer interaction (HCI), albeit from different research perspec- tives. In the social sciences, the focus tends not to be primarily on usability or usage modalities, the skills that people need to have in order to operate the de- vices, but above all on exploring how devices integrated into everyday life are changing the ways we live together, how new media and data practices are de- 12 Voice Assistants in Private Homes veloping, and how privacy is being reinterpreted (e.g., Burgess et al. 2022; see Ochs, this volume). Within the social sciences, a field of research is emerging that builds on existing theoretical paradigms (including actor-network the- ory, diffusion research, science and technology studies, surveillance studies, and mediatization research), but which is also developing new innovative and complex methodological approaches. 2. Controversial Discourses, Household Publics, and Everyday Practices Assessments of voice assistants in public discourse vary widely (see Hab- scheid, Hector, and Hrncal, this volume). On the one hand, they are advertised as an addition to a digitally-connected and thus smart lifestyle (Hennig and Hauptmann 2019). As assistance systems, they are said to have the potential to compensate for handicaps and facilitate a self-determined life for older people (Endter, Fischer, and Wörle 2023). On the other hand, they are also subject to critique, because the devices provide manufacturers with users’ voice data from a particularly sensitive context, the private domestic sphere (Sadowski 2020; Turow 2021), largely as a result of “cooperation without consensus” (Waldecker, Hector, and Hoffmann 2023; for the concept see Star 1993). Although voice-controlled assistance systems are embedded in social interaction and everyday practice, those who want to make full use of their functional potential must adapt to technologized dialogue structures and platform logics. In doing so, they have to reveal a lot about themselves that is transmitted beyond the household as ‘data’ where it can be analyzed and exploited in ways and for purposes that are opaque to the user. Furthermore, the creation of social order in such contexts can be distorted by problematic biases (see, for example, Leblebici in this volume). It is in smart home environments that assistance systems as central inter- faces come into their own, while at the same time opening up the household to the outside world far more than ever before. Whereas classic smart speak- ers’ capacity for surveillance was limited to the perceptual mode of hearing (on “eavesdroppers” in physical or electronically mediated presence, see Goffman 1981, 132), smart homes incorporate camera-, monitor- and sensor-based sys- tems and networks including various stationary and mobile devices and in- frastructures, which can massively expand the scope for data collection. Un- der certain circumstances, this is accompanied by a further dissolution of the Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 13 boundaries of privacy, which on the one hand (e.g., in the case of surveillance of household members) may be perceived as abuse, but on the other hand (e.g., from a security perspective) may seem desirable. As with all media, it is an open question as to how users will continue to adapt to new forms of media and how they make media adapt to the circum- stances of their everyday lives. Individuals and households follow public dis- course and interpret it in the light of their own household’s public sphere, their concrete living conditions and interests. There is an ongoing debate within the humanities and social sciences, and especially within the domestication re- search paradigm (Hartmann 2023; Hector et al. 2023) on the adoption of dig- ital media in household use settings. In principle, domestication research is based upon an analogy drawn between the process whereby media are appro- priated and the process whereby cohabitation with farm animals or pets is es- tablished in the course of civilization. Domestication research, as summarized by Waldecker and Hector (2023, 5) “paints media as something that comes into the everyday life of users as foreign and wild, as something that has to be tamed and brought to relate to domestic routines”. The metaphor of “taming” em- phasizes the somewhat unpredictable and sometimes even threatening aspect of media technologies. This contrasts with the private household that often symbolizes a sense of security. With reference to Giddens (1984), Waldecker and Hector point out that this “ontological security” fundamentally establishes trust, supposedly guarantees the stability of one’s own identity, the continuity of life and of the immediate environment. (Media) technologies that become entangled with this ontological security challenge it and can disrupt it: They become involved in everyday rituals, and even if ontological security is initially called into question by new media technologies (see Silverstone et al. 1992, 17), they (often) lose their threatening character as they are successively woven into everyday life, i.e., they become domesticated (see Bausinger 1984, 349–350). In the process, everyday routines take new forms, and new practices emerge. First of all, new practices are required to get the novel devices and services to work at all. Further practices serve again and again to overcome the systems’ technical unwieldiness and resistance. At the same time, the new usage practices become more or less deeply embedded in everyday life (see Waldecker and Hector 2023): They may (re-)shape, for example, the structur- ing of time between ‘work’ and ‘leisure’, ways of dealing with privacy, or the design of rooms and furnishings in the home. When users live together with other people in households, they must negotiate among themselves who uses which media, when, and how. In such contexts, economic decisions are also 14 Voice Assistants in Private Homes discussed in connection with political and ultimately moral issues, such as whether to subscribe to streaming services, and if so, from which provider(s). Deliberation of such questions involves not only members of the household with its own power dynamics, but also voices from beyond the home; advice may be asked of friends, or sought in online forums or among reviews in which “online warm experts” reflect in accessible language on the possible uses of consumer technology as well as their limitations (Neville 2021; see also Waldecker and Hoffmann 2023). 3. Media Appropriation as a Linguistically Mediated Practice Changing everyday practices as a consequence of media use is also observable at the linguistic level of everyday practice, all the more so when the technol- ogy concerned has a linguistic surface. This is the case, for example, for tele- vision, which is one of the classic mass media that has attracted particular interest in domestication and appropriation research. Unlike smart speakers, television does not require verbal input from users, neither at the level of con- tent nor at the level of operation. Television broadcasts unidirectional com- munication, yet users have been shown to participate nonetheless. For tele- vision, “parainteraction” is characteristic, as Ayaß (1993) – drawing on Horton and Strauss (1957) as well as on Horton and Wohl (1986) – has shown: In uni- directional communication, forms of direct address and staged connection to everyday practices are used by on-screen performers to create an impression of interaction with those watching. Such utterances counterfactually imply that bidirectional interaction ‘through’ the screen could be possible (see also Böck- mann et al. 2019, 145), and under certain circumstances, viewers pick up on this with forms of “parasocial” pseudointeraction in front of the screen (Ayaß 1993, 36). The fact that in many cases the use of media is anchored in linguistic and interactional practice has been emphasized especially strongly by linguistic studies. These have addressed, among other topics, speaking while watching television together (Holly, Püschel, and Bergmann 2001) and intermission talk in theater (Gerwinski, Habscheid, and Linz 2018). It has been shown that viewers use the semiotic material their TV brings into the home as a resource for mutual “orientation” with respect to public issues (Holly 2001, 11–13). The studies also revealed that the appropriation of media – technologies as well as content – is affected not least by the possibilities of linguistic interac- Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 15 tion during and after reception. Examples include the format of “response cries” (Goffman 1981) and other forms of “terse speaking” (Baldauf 2002) in television-accompanied speech or reenactments and other reconstructive genres related to theater dialogues in intermission talk (Schlinkmann 2021). Accordingly, to study the appropriation of smart speaker technologies, it is necessary to ask how the linguistic conditions of their use enable and limit appropriation. Unlike traditional television, internet technologies are two-way media: To put it bluntly, they not only bring the world into the household, but also the household into the world, with the latter in the form of specifically collected, aggregated, and processed data. The use of this data impacts on everyday life in ways that are noticed but cannot be traced, for example, in the form of person- alized advertising or sensor-based environments in the smart home that adapt to usage habits. Thus, in the case of digital household technologies, not only are digital media domesticated in the home, households are also “externalized” (Brause and Blank 2020), or, in Hepp’s words “deeply mediatized” (Hepp 2020). The world that comes into the home with smart speakers is also linguistic on the surface – to a certain extent, it resembles the spoken language of inter- personal interaction. However, linguistic exchange with the machines differs not only in that dialogue involves non-human conversation partners, but that the technical language-processing systems upon which the latter depend have a limited ability to cooperate (Suchman 2007). The linguistic contributions to this volume discuss the range of forms such conversation can take: focusing on the human–machine dialogues, the social interaction they take place within, and the everyday practices that are realized – or not – as a result. The sociotechnical relationships under discussion also raise fundamental questions for social theory. From a conversational linguistics perspective, the ANT approach, whereby all participating entities are conceptualized as equally significant actants (Latour 2005) seems unsatisfactory to us. For example, lan- guage-processing machines like Alexa are participants in practice, but not participants in social interaction as it is understood by conversation analysis (Habscheid 2023; Hector, in preparation; Habscheid, Hector, and Hrncal, this volume). From an ethnomethodological perspective, it can be shown that users orient towards machinic conversation partners with attitudes that, depending on the situation, sometimes reflect a more anthropomorphizing and at other times a more instrumentalized approach to the technology. Accordingly, An- tonia Krummheuer (2010) characterizes the sociotechnical dialogue with an embodied conversational agent (ECA) as a “hybrid” or “ambiguous” exchange: 16 Voice Assistants in Private Homes The exchange between human and machine shows similarities to interper- sonal interaction, which is simulated to a certain extent (see also Hennig and Hauptmann, 2019), but also differences that require users to adapt to the limited communication capabilities of the machines (see also Lotze, this volume). Agency of the voice assistants is an object of negotiation both in everyday practice (Habscheid, Hector, and Hrncal 2023) and at the level of public discourse (Lind and Dickel 2024). 4. Smart Speaker Use and the Social Consequences for Everyday Reality The use of digital technology is just as much a part of everyday life as the use of many other devices and communication with people who are physically present (Keppler 2018, 73). With the integration of a smart speaker into one’s private household, this is extended by a technical artifact that is designed to function as a kind of interaction partner. Based on studies of social robotics, Michaela Pfadenhauer and Tobias Lehmann (2021) propose that a smart speaker can also be regarded as an “artificial companion” in everyday life. Smart speakers are expected to execute various commands as reliably as possible, search for and provide information, manage operation of networked devices, and offer ser- vices. Although their dialogue capabilities are still limited (Habscheid 2023) and communication is prone to disruption and often inconclusive or unpre- dictable (Pins et al. 2020; see also Lutz and Newlands, this volume), it can be assumed that this will improve significantly in the future, not least through the implementation of artificial intelligence. As an everyday companion, the smart speaker is certainly part of household communication: as an omnipresent third party. This participation at the locus of everyday life not only creates a social and emotional relationship with the device or with devices, but will also change how we communicate socially in everyday life. In the words of Hepp (2015), the communicative figuration of households, i.e., the communicative arrange- ment and role behavior of their members, is currently undergoing transforma- tion. It is therefore of sociological interest to explore the extent to which the artificial companions can be regarded as “vehicles to cultural worlds of expe- rience” (Pfadenhauer and Lehmann 2021) and prompt new fundamental ques- tions of sociality (see also Hepp et al. 2022). Furthermore, sociological investigation into sociotechnical practices and their consequences for the protection of privacy is called for. Through the Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 17 appropriation and use of smart technology, users reveal personal data about themselves (such as their taste in music, their shopping behavior, their account data, their everyday routines, their address book) and allow their home envi- ronment to be (acoustically) recorded. Huge volumes of data are transmitted to tech companies, stored, and evidently used as training data or for other purposes. Users are not always aware of this and it is largely beyond their control, although within the EU at least the Digital Service Act is intended to ensure greater transparency (see the conversation with Nikolai Horn in this volume). On this matter, it is important to examine users’ own attitudes and explanations for how they deal with data protection and privacy. The narrative ‘nothing to hide, nothing to fear’ is expressed by many users as a pragmatic approach to data protection settings and issues for a variety of reasons (see Waldecker, Martin, and Hoffmann, this volume). Existing studies of ways of dealing with and justifying decisions concerning the data protection settings of digital applications have tended to neglect to consider those indirectly affected, such as visitors to households in which such devices are installed and used as a matter of course (e.g., Hoffmann 2023). Discourses on media and critiques of corporate data practices not only shape public debate, but are also negotiated in the private sphere (see Vermeulen and Mols, this volume). It remains to be seen how these smart technologies and media practices will ‘conventionalize’ in the future and how social scientists will study the ongoing developments. 5. On the Contributions in this Volume This volume presents a wide spectrum of recent research on voice-operated systems and services, including analyses focusing on their (linguistically me- diated) use and appropriation, on users’ appraisals of them, and on the ques- tion of the exploitative utilization of the data they transmit. Perspectives from conversation analysis and media linguistics, media sociology, media studies, surveillance studies, the critique of political economy and related aspects of consumer research, domestication research, pragmatist and praxeological so- ciology as well as critical theory are brought together to shed light on the prac- tical entanglement of users, devices, algorithms, data, and corporate interests. By encompassing these diverse approaches, this volume sets out to analyze the phenomenon of IPAs at multiple levels: from that of interaction, to everyday practices in households, to the level of users’ perceptions and evaluations, and 18 Voice Assistants in Private Homes not least in relation to global processes of data processing and exploitation. Our aim is to provide a comprehensive view of the transformation and persis- tence of everyday practices under platformized conditions and usage practices mediated by novel interfaces. The majority of the contributions to this volume have evolved from pre- sentations given at the conference “Voice Assistants in Private Homes. Media, Data, and Language in Interaction and Discourse”, which took place on May 8 and 9, 2023, at the University of Siegen, Germany, organized by the re- search project “Un/desired observation in interaction: Intelligent Personal Assistants”, which from 2020–2023 empirically investigated media practices with voice assistants as a key technology in the field of data-intensive digital media, taking a dual approach combining media sociology (Waldecker, Mar- tin, and Hoffmann, this volume) and applied linguistics (Habscheid, Hector, and Hrncal, this volume). The project was part of the Collaborative Research Center “Media of Cooperation”, which brings together numerous sub-projects investigating diverse phenomena but all taking as their point of departure a praxeological media theory paradigm that conceptualizes practice as the “mutual making of common goals, means and processes” and, in this context, media as “cooperatively created conditions of cooperation” or, in short, as “media of cooperation” (Schüttpelz 2017, 24). The “means” that can be coop- eratively produced as “media” can – but do not have to – be of a linguistic nature (see Goodwin 2018; Habscheid, Hector, and Hrncal, this volume). In accordance with the interdisciplinary agenda of the Collaborative Research Center, as editors of this volume we seek to examine the complex phenomenon of data-intensive, AI-based assistance systems by addressing its multiple layers. The aim is to shed light on the intricate interrelationships between use and users, language, devices, algorithms, data, organizations, and economic exploitation. The volume is structured in four parts. The first section – Voice Assistants in Private Homes. Conceptual Considerations – focuses on the theoretical foundations of key areas of IPA research and showcases various method- ological approaches and findings of empirical studies. Carsten Ochs begins by examining the affective reactions of people who wonder why the users of smart speakers seem so unconcerned about their privacy. He traces the emer- gence of the modern practice of privacy protection, which was established in the 20th century, and now, since the advent of smart technologies in private homes, is being renegotiated. Ochs attempts to show what actually happens to the data collected and processed by smart speaker infrastructures that reach Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 19 into private homes, and concludes that the term “surveillance capitalism” is an apt one under the circumstances. Taking a Marxist approach and drawing on feminist theory, Markus Kienscherf’s contribution investigates the role of voice assistants in the reproduction of labor and capital. The appropriation of user-generated voice data by smart speakers is positioned within a more gen- eral history of the role of surveillance in the (re)production of capitalist social relations. The author shows that surveillance is central to the appropriation of surplus value in the spheres of production, the social reproduction of labor power, and the management of circulation and consumption. He then looks at the business models of tech companies and argues that the appropriation of user-generated data transmitted via smart speakers represents an extension of capitalist surveillance into the sphere of social reproduction. The chapter by Caja Thimm, Phillip Engelhardt, and Julia Schmitz deals with anthropomorphism and communication accommodation to voice assis- tants. The focus is on how assistance systems with VUIs (voice user interfaces) are used and affectively engaged with in multi-person households, based upon a case study with households including physically impaired people with special support needs. The authors observe that these users’ assumptions, attitudes, and expectations were not stable but varied according to contextual factors. As a theoretical basis for the research, “Communication Accommodation The- ory” (CAT) is developed and adapted for the study of HCI constellations, fo- cusing on strategies of anthropomorphization, which are shown to partially – and perhaps increasingly – influence the ways people interact with machines as well as to shape the discourse, interface design, and self-image of users. Last but not least, the authors reflect on the different insights into usage gained by their methodological combination of interviews and media diaries. The last contribution in this first section by Netaya Lotze traces the development of a complex sociolinguistic model that can bring findings concerning the anthro- pomorphization of HCI technologies together with evidence of cognitive and linguistic adaptation to the (more or less) limited communicative capacity of machines. After a comprehensive research overview, Lotze presents the results of her own studies conducted since 2000, which she summarizes and inter- prets in the light of the model (and vice versa). The model integrates various approaches from the philosophy of language, computer science, cognitive sci- ence, and linguistics, and is structured to take into account ‘external factors’, ‘system variables’, and ‘user variables’, while incorporating a user typology as well as enabling diachronic analysis. 20 Voice Assistants in Private Homes Section 2, Linguistic Exchange with Voice Assistants as a Practical Prob- lem presents studies from the field of linguistics examining the practical use of and critical discourse about language assistants. The chapter by Florence Oloff provides an empirically underpinned perspective on the usability and learnability of voice assistants as everyday technologies. Oloff examines spe- cific instances of older users’ first encounters, during adult education courses, with hitherto unknown voice-operated applications. She shows how, in non- profit, professionally guided practical training sessions, participants explore the potential benefits and problems of multimodal interfaces – the first stage of appropriation. Furthermore, a mismatch between the actual needs of the learners, the spatial, temporal, and medial limitations of the settings, and the teaching methods used by instructors to deal with these factors in an impro- visational way becomes clearly evident. Oloff makes some suggestions on how to improve teaching and learning in these contexts. The contribution by Didem Leblebici provides insights into the experiences of Turkish-speaking users of non-Turkish-speaking voice assistants in Germany. The author expands upon a media linguistics interest in voice user interfaces by drawing on theoretical un- derstandings of multilingualism from sociolinguistics and critical discourse analysis. The chapter, which is based on the linguistic analysis of ethnographic interview data, advances a critical discussion of the ways that language-pro- cessing technologies reinforce the standardization of language. Detailed ex- amples are drawn upon to illustrate and analyze different phenomena of styl- ized language use in interaction with IPAs. A contribution by Stephan Hab- scheid, Tim Hector, and Christine Hrncal concludes this second section. The authors present an overview of the results to date from the linguistic strand of the project “Un/desired observation in interaction: Intelligent Personal As- sistants”. In the theoretical part of the chapter, the conceptual foundations of the “Media of Cooperation” Collaborative Research Center are further elabo- rated from a linguistic praxeological perspective, discussing approaches taken in interaction research on the one hand and linguistic media research on the other, as well as the domestication approach in media and communication re- search. This is followed by analyses of empirical findings from the research project, which underscore how instrumental linguistic practices are in embed- ding smart speakers into domestic routines, and illustrate how newly acquired technology reshapes social practices and communication within households. The third thematic section brings together contributions that deal with the issues of Privacy and Data Protection as Practical Problems. Concerns relat- ing to the extraction of personal data and its subsequent use have been raised Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 21 for almost as long as digital media technologies and applications have been available, with the disclosure of data from the domestic sphere often attracting particularly critical attention. One instance when contextual privacy is called into question is when devices behave unexpectedly. Glitches occur, which can appear as technical anomalies and expose critical privacy vulnerabilities. Tak- ing a glitch studies approach, Christoph Lutz and Gemma Newslands focus on users’ experiences of malfunctions, which can also have wider societal implica- tions and raise questions about surveillance, data security, and the ethical re- sponsibilities of technology companies. Although glitch studies is an interdis- ciplinary field that tends to use qualitative methods, Lutz and Newslands draw on quantitative data to identify the four most common glitches experienced by Amazon Echo users and how they categorize the consequences of those glitches in relation to levels of trust and concerns about privacy. The findings high- light a critical aspect of smart speaker technology: the delicate balance between their perceived benefits and the fears of potential negative consequences of us- ing them. Such considerations and fears also play a major role when people decide whether (or not) to purchase voice-operated devices in the first place. In their chapter, Jasper Vermeulen and Anouk Mols present a multi-methods study that investigated the privacy perceptions of users and non-users of smart speakers. Based upon data from in-depth interviews and focus groups, they elaborate on Dutch users’ and non-users’ assessments of risks and benefits. They found that users generally appreciated affordances such as controllabil- ity, support, conversation, linkability, and recordability, while some indicated they would prefer greater transparency regarding corporations’ use of data. Non-users associated recordability and locatability with privacy risks that were seen as significant enough to not use such technology at all. In addition to ra- tional considerations, the study also pointed to the role of emotions in shaping adoption considerations and decisions. David Waldecker, Alexander Martin, and Dagmar Hoffmann also look at users’ attitudes towards data protection is- sues in connection with the use of smart speakers and, in particular, how they deal with them. In doing so, they draw on various studies that show the extent to which users of digital media technologies develop a kind of “online apathy”, “data protection cynicism”, or even “digital resignation”. Based on qualitative interviews with smart speaker users in Germany, the authors report how users cultivate certain attitudes towards the devices and the discourse surrounding them and how they explain their usage routines and pragmatic considerations. In addition to the findings of the studies cited, the authors’ analysis of their own interviews reveals an attitude that Andreas Pettenkofer has termed “prag- 22 Voice Assistants in Private Homes matic fatalism”. Users who adopt such a stance more or less accept the data practices of companies and at the same time declare them to be irrelevant to their everyday lives. To conclude this section, Nikolai Horn, data protection ex- pert and currently political advisor to iRights.Lab, discusses in a conversation with Dagmar Hoffmann and David Waldecker the legal and political aspects of protecting voice-based data. The new possibilities offered by AI and natural language processing are also addressed. Questions are raised about the extent to which voice recordings can be used to draw conclusions about identity char- acteristics of users and how voice recordings could be misused. The interview also explores the question of how users can be made more aware of data pro- tection issues and how EU regulations such as the GDPR can ensure greater transparency in data use and give users more control of their own data. The final (fourth) section – Technical Infrastructures as a Practical Prob- lem – brings together a contribution from the field of social informatics and one from the sociology of technology to focus more explicitly on the IT processes and infrastructures that enable smart speaker technology but are not always transparent for users. Over a period of three years, Dominik Pins, Fatemeh Alizahdeh, Alexander Boden, Sebastian Zilles, and Gunnar Stevens used the living lab approach to investigate users’ uncertainties with regard to the data collected as a consequence of their use of smart speakers in everyday life. Based on findings from interviews, field research, and participatory design workshops with 35 households, the authors developed a tool called “CheckMyVA” that supports users in accessing and visualizing their own VA data. The observations and findings presented in the chapter offer suggestions for tools and design strategies that could foster data literacy and enable users to reflect on their long-term interactions with VAs, ultimately “demystifying” the technology. The final chapter, by Niklas Strüver, takes a look behind the scenes to explore the practices involved in the ongoing development of auto- matic language processing. Amazon was once a pioneer in this field, but the launch of new large language models (LLMs) has posed major challenges for the company. Strüver conducted expert and narrative interviews with partici- pants from university research teams who competed in the most recent Alexa Prize Competitions (APCs) to advance Alexa technology. These interviewees are able to offer fascinating insights into development practices, especially concerning the integration of LLMs into existing technology. Examining how the participants in these competitions deal with the conditions set by Amazon and the resources it makes available to competitors, Strüver outlines Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 23 some of the path dependencies, risks, benefits, and structuring aspects that participants encountered in their attempts to innovate Alexa. It can be summarized that research in the field of smart technologies will certainly continue to be necessary, and that lines of inquiry are always shaped by disciplinary conventions, hence interdisciplinary exchange should continue to be promoted in the future. Acknowledgements As stated above, most of the contributions to this volume were developed from presentations held at the conference “Voice Assistants in Private Homes. Media, Data and Language in Interaction and Discourse” organized by the research project “Un/desired observation in interaction: Intelligent Personal Assistants”. We express our gratitude to all those who participated in the conference and to all the authors who have contributed to this book. We would also like to thank our student assistants Chris Dangelmaier, Sarah Diehl, Aileen Halbe, Johanna Klein, Alexander Martin, Franziska Niersberger-Guéye, and Leonie Tittel, who helped to make the conference such a success. The research project was part of the Collaborative Research Center “Me- dia of Cooperation” at the University of Siegen, Germany, in its second fund- ing phase (2020–2023). In its first funding phase (2016–2019), the project was led by Wolfgang Ludwig-Mayerhofer and investigated the strategic practices by which young people using social media tried to attract ‘requested’ atten- tion and avoid ‘undesired’ observation. After Wolfgang Ludwig-Mayerhofer’s retirement, Stephan Habscheid and Dagmar Hoffmann took over as princi- pal investigators in the second funding phase, continuing to follow the general idea of un/desired observation but applying it to smart speakers with VUIs. This volume is a key publication that resulted from the research conducted in those four years. In its third funding phase (2024–2027), the project is once again shifting its focus, this time to smart home environments, in order to explore further transformations accompanying the recent massive expansion of interfaces as well of registrable data in smart homes through sensor-based mechanisms. The project team in the second funding phase of the Collaborative Research Center (2020–2023) was led by principal investigators Stephan Habscheid and Dagmar Hoffmann, with Tim Hector, David Waldecker, Christine Hrncal, and Kathrin Englert as (post-)doctoral researchers. As the editors of this volume, 24 Voice Assistants in Private Homes we would like to thank all of our colleagues for their long-term collaboration and their contribution to the success of this book. We want to thank Pip Hare for her extremely dedicated comprehensive copy editing as well as Sarah Diehl, Franziska Niersberger-Guéye, and Christopher Wegner for their extensive ed- itorial work. This research was funded by the Deutsche Forschungsgemeinschaft (Ger- man Research Foundation) – Project ID 262513311 – SFB 1187 Media of Cooper- ation. References Ayaß, Ruth.1993. “Auf der Suche nach dem verlorenen Zuschauer.” In Medienre- zeption als Aneignung. Methoden und Perspektiven qualitativer Medienforschung, edited by Werner Holly, and Ulrich Püschel, 27–41. Opladen: Westdeut- scher Verlag. Baldauf, Heike. 2002. Knappes Sprechen. Tübingen: Niemeyer. Bausinger, Hermann. 1984. “Media, technology and daily life.” Media, Culture and Society 6: 343–351. https://doi.org/10.1177/016344378400600403. Böckmann, Barbara, Dorothee Meer, Michelle Mohn, Anastasia-Patricia Och, Ilaria Paltrinieri, Alina Renelt, Christine Ramdorf, Daniel Rettinghausen, Katharina Staubach, and Martin Tenz. 2019. “Multimodale Produktbewer- tungen in Influencer-Videos auf YouTube: Zur parainteraktionalen Kon- struktion von Warenwelten.” Zeitschrift für Angewandte Linguistik 70: 139–171. https://doi.org/10.1515/zfal-2019-2003. Borbach, Christoph. 2018. “Sprachspiele|Stimmsynthesen: Zur Nachrichten- technischen Genese Des Auditiven Pendants Von ELIZA.” In Hello, I’m Eliza: Fünfzig Jahre Gespräche Mit Computern, edited by Marianna Baranovska and Stefan Höltgen, 177–98. Bochum: projektverlag. Brause, Saba Rebecca, and Grant Blank: “Externalized domestication: smart speaker assistants, networks and domestication theory.” Information, Com- munication & Society 23(5): 751–763. https://doi.org/10.1080/1369118X.2020. 1713845. Burgess, Jean, Kath Albury, Anthony McCosker, and Rowan Wilken. 2022. Ev- eryday data cultures. Cambridge: Polity Press. Dürscheid, Christa. 2023. “Mit Menschen und Chatbots. Digitale Kommuni- kation früher und heute. Conference Paper: LingColl 2022 in Tampere.” Re- https://doi.org/10.1177/016344378400600403 https://doi.org/10.1515/zfal-2019-2003 https://doi.org/10.1080/1369118X.2020.1713845 https://doi.org/10.1080/1369118X.2020.1713845 Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 25 search Gate. Accessed August 10, 2024. https://www.researchgate.net/publ ication/369506979_Mit_Menschen_und_ Chatbots. Endter, Cordula, Florian Fischer, and Tobias Wörle. 2023. “Doing Home by Us- ing Digital Assistive Technologies. On the Role of Meaning and Materiality in the use of Digital Assistive Technologies at Older People’s Domesticity.” Digital Culture & Society 9 (1): 179–200. https://doi.org/10.14361/dcs-2023-01 09. Englert, Kathrin, Dagmar Hoffmann, and David Waldecker. 2022. “Tut mir leid, ich verstehe nicht ganz – Smart Speaker als vermeintliche Gesprächs- partner*innen.” merz | Zeitschrift für Medienpädagogik (66) 2: 24–34. Gerwinski, Jan, Stephan Habscheid, and Erika Linz. 2018. Theater im Gespräch. Sprachliche Publikumspraktiken in der Theaterpause. Berlin and Boston: de Gruyter. Giddens, Anthony. 1984. The Constitution of Society. Outline of a Theory of Structura- tion. Cambridge: Polity Press. Goffman, Erving. 1981. Forms of Talk. Philadelphia: University of Pennsylvania Press. Goldman, Sharon. 2024. “How Amazon blew Alexa’s shot to dominate AI, ac- cording to more than a dozen employees who worked on it.” Fortune, June 13, 2024 https://fortune.com/2024/06/12/amazon-insiders-why-new-alex a-llm-generative-ai-conversational-chatbot-missing-in-action/. Goodwin, Charles. 2018. Co-Operative Action. New York: Cambridge University Press. Habscheid, Stephan. 2023. “Socio-technical dialogue and linguistic interac- tion. Intelligent Personal Assistants (IPA) in the Private Home.” Sprache und Literatur 51 (2): 167–196. https://doi.org/10.30965/25890859-05002020. Habscheid, Stephan, Tim Hector, and Christine Hrncal. 2023. “Human and Non-Human Agency as Practical Accomplishment: Interactional Occa- sions for Ascription and Withdrawal of (Graduated) Agency in the Use of Smart Speaker Technology”. Social Interaction. Video-Based Studies of Human Sociality 6 (1). https://doi.org/10.7146/si.v6i1.137378. Hartmann, Maren, ed. 2023. The Routledge Handbook of Media and Technology Do- mestication. London and New York: Routledge. Hector, Tim. In preparation. Smart Speaker als Gesprächsbeteiligte? Sprachliche Praktiken im Prozess der Domestizierung mit Voice-User-Interfaces. Dissertation, University of Siegen, Faculty of Arts and Humanities – Faculty I. Defense: April 12, 2024. https://www.researchgate.net/publication/369506979_Mit_Menschen_und_ https://www.researchgate.net/publication/369506979_Mit_Menschen_und_ https://doi.org/10.14361/dcs-2023-0109 https://doi.org/10.14361/dcs-2023-0109 https://fortune.com/2024/06/12/amazon-insiders-why-new-alexa-llm-generative-ai-conversational-chatbot-missing-in-action/ https://fortune.com/2024/06/12/amazon-insiders-why-new-alexa-llm-generative-ai-conversational-chatbot-missing-in-action/ https://doi.org/10.30965/25890859-05002020 https://doi.org/10.7146/si.v6i1.137378 26 Voice Assistants in Private Homes Hector, Tim, David Waldecker, Niklas Strüver, and Tanja Aal, eds. 2023. The- matic issue: “Taming Digital Practices – On the Domestication of Data-Driven Tech- nologies”. Digital Culture & Society 9 (1). Hennig, Martin and Kilian Hauptmann. 2019. “Alexa, optimier mich! KI-Fik- tionen digitaler Assistenzsysteme in der Werbung.” Zeitschrift für Medien- wissenschaft 11 (2): 86–94. https://doi.org/10.25969/mediarep/12636. Hepp, Andreas. 2020. Deep mediatization. Abingdon and New York: Routledge. Hepp, Andreas. 2015. “Kommunikative Figurationen: Zur Beschreibung der Transformation mediatisierter Gesellschaften und Kulturen.” In Theorien des Medienwandels, edited by Susanne Kinnebrock, Christian Schwarzeneg- ger, and Thomas Birkner, 161–188. Köln: Herbert von Halem. Hepp, Andreas, Wiebke Loosen, Stephan Dreyer, Juliane Jarke, Sigrid Kannen- gießer, Christian Katzenbach, Rainer Malaka, Michaela Pfadenhauer, Cor- nelius Puschmann, and Wolfgang Schulz. 2022. “Von der Mensch-Maschi- ne-Interaktion zur kommunikativen KI. Automatisierung von Kommuni- kation als Gegenstand der Kommunikations- und Medienforschung.” Pu- blizistik 67 (4): 449–474. https://doi.org/10.1007/s11616-022-00758-4. Herbig, Daniel. 2024. “Amazon: Verbesserte Alexa kommt laut Bericht nicht voran.” Heise online, June 14, 2024. https://www.heise.de/news/Amazon-V erbesserte-Alexa-kommt-laut-Bericht-nicht-voran-9763110.html. Hoffmann, Dagmar. 2023. Digitaler Alltag, Datenpraktiken und Subjektau- tonomie. In: merz | Zeitschrift für Medienpädagogik, 67 (1): 51–59. Holly, Werner. 2001. “Der sprechende Zuschauer.” In Der sprechende Zuschauer. Wie wir uns Fernsehen kommunikativ aneignen, edited by Werner Holly, Ulrich Püschel, and Jörg Bergmann, 12–24. Wiesbaden: Westdeutscher Verlag. Holly, Werner, Ulrich Püschel, and Jörg Bergmann, eds. 2001. Der sprechende Zuschauer. Wie wir uns Fernsehen kommunikativ aneignen. Wiesbaden: West- deutscher Verlag. Horton, Donald, and Anselm Strauss. 1957. “Interaction in Audience Participa- tion Shows.” American Journal of Sociology 62 (6): 579–587. Horton, Donald, and R. Richard Wohl. 1986. “Mass communication and para- social interaction: Observation on intimacy at a distance.” In Inter/Media. Interpersonal communication in a media world (3rd edition), edited by Gary Gumpert, and Robert Cathcart, 185–206. New York, Oxford: Oxford Uni- versity Press. Huq, Oliver. 2011. “iPhone 4S: Auf die inneren Werte kommt es an.” Heise online, October 4, 2011. https://www.heise.de/news/iPhone-4S-Auf-die-inneren- Werte-kommt-es-an-1354456.html. https://doi.org/10.25969/mediarep/12636 https://doi.org/10.1007/s11616-022-00758-4 https://www.heise.de/news/Amazon-Verbesserte-Alexa-kommt-laut-Bericht-nicht-voran-9763110.html https://www.heise.de/news/Amazon-Verbesserte-Alexa-kommt-laut-Bericht-nicht-voran-9763110.html https://www.heise.de/news/iPhone-4S-Auf-die-inneren-Werte-kommt-es-an-1354456.html https://www.heise.de/news/iPhone-4S-Auf-die-inneren-Werte-kommt-es-an-1354456.html Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 27 Keppler, Angela. 2018. “Medien, Lebenswelt und Alltagshandeln.” In Medien- soziologie. Handbuch für Wissenschaft und Studium, edited by Dagmar Hoff- mann and Rainer Winter, 71–85. Baden-Baden: Nomos. https://doi.org/10 .5771/9783845264196-70. Kim, Eugene. 2022. “Amazon is gutting its voice assistant, Alexa. Employees describe a division in crisis and huge losses on a wasted opportunity.” Busi- ness Insider, November 19, 2022. https://www.businessinsider.com/amazo n-alexa-job-layoffs-rise-and-fall-2022-11. Krummheuer, Antonia. 2010. Interaktion mit virtuellen Agenten? Zur Aneignung ei- nes ungewohnten Artefakts. Stuttgart: Lucius&Lucius. Latour, Bruno. 2005. Reassembling the Social: An Introduction to Actor-Network-The- ory. Oxford: Oxford University Press. Lind, Miriam, and Sascha Dickel. 2024. “Speaking, but having no voice. Nego- tiating agency in advertisements for intelligent personal assistants.” Con- vergence 30 (3): 1008–1024. https://doi.org/10.1177/13548565231192100. Neville, Stephen J. 2021. “The domestication of privacy-invasive technology on YouTube: Unboxing the Amazon Echo with the online warm expert.” Con- vergence 27 (5): 1288–1307. https://doi.org/10.1177/1354856520970729. Pfadenhauer, Michaela, and Tobias Lehmann. 2022. “Affects after AI: Sociolog- ical perspectives on artificial companionship.” In The Routledge Social Science Handbook of AI, edited by Anthony Elliott, 91–106. London: Routledge. Pins, Dominik, Alexander Boden, Gunnar Stevens and Britta Essing. 2020. “‘Miss understandable’ – A study on how users appropriate voice assistants and deal with misunderstandings.” In Proceedings of Mensch und Computer 2020 (MUC20), 349–359. ACM, Magdeburg. Pitsch, Karola. 2015. “Ko-Konstruktion in der Mensch-Roboter-Interaktion. Kontingenz, Erwartungen und Routinen in der Eröffnung.” In Ko-Konstruk- tionen in der Interaktion, edited by Ulrich Krafft, Elisabeth Gülich, and Ulrich Dausendschön-Gay, 229–58. Bielefeld: transcript. https://doi.org/10.1515/ 9783839432952-013. Porcheron, Martin, Joel E. Fischer, Stuart Reeves, and Sarah Sharples. 2018. “Voice Interfaces in Everyday Life.” Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems – CHI ’18, 1–12. https://doi.org/10.1145/ 3173574.3174214. Sadowski, Jathan. 2020. Too Smart. How Digital Capitalism is Extracting Data, Controlling Our Lives, and Taking Over the World. Cambridge: MIT Press. Schlinkmann, Eva (2021): Kommunikative Rekonstruktion in der Theaterpause. Eine Gattungsanalyse von Pausengesprächen im Theater. Stuttgart: Metzler. https://doi.org/10.5771/9783845264196-70 https://doi.org/10.5771/9783845264196-70 https://www.businessinsider.com/amazon-alexa-job-layoffs-rise-and-fall-2022-11 https://www.businessinsider.com/amazon-alexa-job-layoffs-rise-and-fall-2022-11 https://doi.org/10.1177/13548565231192100 https://doi.org/10.1177/1354856520970729 https://doi.org/10.1515/9783839432952-013 https://doi.org/10.1515/9783839432952-013 https://doi.org/10.1145/3173574.3174214 https://doi.org/10.1145/3173574.3174214 28 Voice Assistants in Private Homes Schüttpelz, Erhard. 2017. “Infrastructural Media and Public Media.” Media in Action 1 (1): 13–61. http://dx.doi.org/10.25819/ubsi/7935. Silverstone, Roger, Eric Hirsch, and David Morley. 1992. “Information and communication technologies and the moral economy of the household.” In Consuming Technologies. Media and Information in Domestic Spaces, edited by Roger Silverstone and Eric Hirsch, 13–28. London: Routledge. “Siri”. Wikipedia article. Accessed August 10, 2024. https://en.wikipedia.org/ wiki/Siri. Star, Susan. 1993. “Cooperation Without Consensus in Scientific Problem Solv- ing: Dynamics of closure in open Systems.” In CSCW. Cooperation or Con- flict?, edited by Dan Daiper, Closton Sanger, and Steve Easterbrock, 93–106. London: Springer. https://doi.org/10.1007/978-1-4471-1981-4_3.  Statista. 2021. “Statista Global Consumer Survey. ID 999790” Accessed August 21, 2024. https://de.statista.com/statistik/daten/studie/1271603/umfrage/ anteil-der-haushalte-in-deutschland-mit-smart-speaker/. Stresing, Laura. 2011. “Siri sagt: Warum die Apple-Sprachsoftware nicht nur Geeks fasziniert.” Tagesspiegel, December 7, 2011. https://www.tagesspieg el.de/gesellschaft/medien-_-ki/warum-die-apple-sprachsoftware-nicht- nur-geeks-fasziniert-6708628.html. Strüver, Niklas. 2023. “Frustration Free: How Alexa Orchestrates the Develop- ment of the Smart Home.” Digital Culture & Society 9 (1): 99–123. https://doi. org/10.14361/dcs-2023-090106. Suchman, Lucy. 2007. Human-machine reconfigurations. Plans and situated actions (2nd edition). Cambridge: Cambridge University Press. Turow, Joseph. 2021. The Voice Catchers: How Marketers Listen in to Exploit Your Feel- ings, Your Privacy, and Your Wallet. New Haven and London: Yale University Press. Volmar, Axel. 2019. “Productive Sounds.” In The Democratization of Artificial Intel- ligence, edited by Andreas Sudmann, 55–76. Bielefeld: transcript. https://d oi.org/10.1515/9783839447192-004. Waldecker, David, and Tim Hector. 2023. “A Praxeological Approach Towards the Domestication of Connected Media Technologies. Introduction.” Digi- tal Culture & Society 9 (1): 5–22. https://doi.org/10.14361/dcs-2023-0102. Waldecker, David, and Dagmar Hoffmann. 2023. “Inszenierung von kritischen Kompetenzen in Nischenöffentlichkeiten. Bewertungen von Smart Spea- kern auf YouTube.” kommunikation@gesellschaft 23 (1). https://doi.org/10.15 460/kommges.2022.23.1.1000. http://dx.doi.org/10.25819/ubsi/7935 https://en.wikipedia.org/wiki/Siri https://en.wikipedia.org/wiki/Siri https://doi.org/10.1007/978-1-4471-1981-4_3 https://de.statista.com/statistik/daten/studie/1271603/umfrage/anteil-der-haushalte-in-deutschland-mit-smart-speaker/ https://de.statista.com/statistik/daten/studie/1271603/umfrage/anteil-der-haushalte-in-deutschland-mit-smart-speaker/ https://www.tagesspiegel.de/gesellschaft/medien-_-ki/warum-die-apple-sprachsoftware-nicht-nur-geeks-fasziniert-6708628.html https://www.tagesspiegel.de/gesellschaft/medien-_-ki/warum-die-apple-sprachsoftware-nicht-nur-geeks-fasziniert-6708628.html https://www.tagesspiegel.de/gesellschaft/medien-_-ki/warum-die-apple-sprachsoftware-nicht-nur-geeks-fasziniert-6708628.html https://doi.org/10.14361/dcs-2023-090106 https://doi.org/10.14361/dcs-2023-090106 https://doi.org/10.1515/9783839447192-004 https://doi.org/10.1515/9783839447192-004 https://doi.org/10.14361/dcs-2023-0102 https://doi.org/10.15460/kommges.2022.23.1.1000 https://doi.org/10.15460/kommges.2022.23.1.1000 Stephan Habscheid et al.: Voice Assistants in Private Homes. Introduction to the Volume 29 Waldecker, David, Tim Hector, and Dagmar Hoffmann. 2024. “Intelligent Per- sonal Assistants in practice. Situational agencies and the multiple forms of cooperation without consensus.” Convergence 30 (3): 975–991. I Voice Assistants in Private Homes. Conceptual Considerations The DataEconomy@Home The Private Sphere, Privacy, and the Embedding of Artificial Intelligence Systems into Everyday Life as an Expansion of Economic Data Grabbing Carsten Ochs Abstract Although voice assistants have been adopted widely in private homes, they still cause bafflement among those who have a negative attitude towards smart speakers. But what is at stake in affective reactions such as these? And why does the issue of privacy fre- quently come to the fore in this context? This contribution sets out to somewhat unsettle the seeming naturalness of problematizing smart speakers as a “privacy issue”, so as to offer a clearer understanding of the whys and wherefores of the issue in the first place. To this end, I first examine the astonishment that is frequently expressed in response to the dissemination of smart speakers (section 2). What is so astounding about installing smart speakers in the private sphere of the home? The next aspect to be investigated (sec- tion 3) concerns an essentially modern privacy practice: it is linked to the expectation that individuals have the right and the means to control which entities may receive which el- ements of their personal information. The idea that in order to constitute oneself as an individual one must have control over who can access one’s personal information came to prevail as the dominant concept of data privacy in the 20th century. Having thus specified the notion of the private sphere on the one hand, and of privacy on the other, I proceed (sec- tion 4) by investigating why some of today’s users willingly relegate these fundamental forms of privacy. And I analyze what actually happens to the data that is collected and processed by smart speaker infrastructures that reach into private homes. To conclude (section 5), I bring together the insights gained in order to support the argument that smart speakers in the private home form part of surveillance capitalism’s expansion into as many social spheres as possible. 34 Voice Assistants in Private Homes. Conceptual Considerations 1. Introduction Over the last few years, I participated regularly in University of Kassel’s winter semester lecture series “Der soziologische Blick” (“The sociological gaze”), a course that serves primarily to introduce new students to relevant research fields, topics, and debates addressed by contemporary sociology, but some- times also attracts interested listeners from the general public. As I have spent many years investigating the digital transformation of information privacy (Ochs 2022), I was frequently assigned with the task of presenting to students a sociological perspective on the social role played by the distinction of private versus public in pre-modern, modern, and contemporary societies. At the end of my lecture in the 2022 series, I was approached by an elderly man; I assumed that he had either started to study sociology since retirement, or was simply interested in the topic. He expressed his appreciation of the lecture, before going on to raise some criticism regarding my bad habit of bridging pauses for thought by murmuring filler words like “exactly”, “yes”, “that’s it”, etc. After this assessment of the quality, he shifted to the lecture’s content and pointed out to me that the major current threat to privacy was the implementation of smart speakers, “such as Alexa”, in private homes. That was something that my research should focus upon, he advised, shaking his head with bafflement that anyone could be crazy enough to welcome such devices into their homes. What the anecdote illustrates is a rather common reaction when it comes to voice assistants in private homes, common at least among people who have a negative attitude towards smart speakers and the infrastructures that enable their agency (for an impressive mapping of such an infrastructure, see Crawford and Joler’s 2018 visual rendition and analysis of Amazon Echo’s “anatomy”). It is perhaps unsurprising that the practice of using smart speak- ers seems particularly alarming to an elderly generation that has witnessed the state surveillance in East Germany and/or the resistance to the West German census in the 1980s and the Federal Constitutional Court’s assertion of the right to informational self-determination. And yet, we should not presume that it is only the elderly who are concerned. But what is at stake in affective reactions such as these? What exactly was it that made the lecture attendee shake his head at the idea of allowing smart speakers into private homes? And why does the issue of privacy come to the fore in this context? This contribution sets out to somewhat unsettle the seeming naturalness of problematizing smart speakers, such as Echo, and voice assistants, such as Alexa, as a “privacy issue”. It is not my aim to applaud the proliferation of these Carsten Ochs: The DataEconomy@Home 35 devices and infrastructures, nor to absolve them of criticism, but rather to offer a clearer understanding of the whys and wherefores of the issue in the first place. To this end, I will distinguish three different aspects and consider them in succession before consolidating the insights gained to formulate the main argument of my contribution. The first aspect to be examined in the next section (section 2) concerns the astonishment that is frequently expressed in response to the dissemination of smart speakers. What is so remarkable or astounding about installing smart speakers in the home? As I will explain, there is nothing “natural” about assump- tions that the home as a private sphere should be shielded from techno-eco- nomic agencies such as the Amazon Echo. Yet, many people do perceive the idea of connecting their household to Amazon’s global infrastructure as an in- vasion into the domestic private sphere that threatens the established norms of the private/public distinction in contemporary society. While the notion that the sanctity of “local privacy” (Rössler 2001, 25; 255; cf. Roessler 2004) must be upheld already had genealogical precedents in pre- modernity even if it took on a more specific form in modern societies, the next aspect to be investigated (section 3) represents an essentially modern practice: it is linked to the expectation that individuals (the owners or residents of pri- vate homes, for example) have the right and the means to control which entities may receive which elements of their personal information. The idea that in or- der to constitute oneself as an individual one must have control over who can access one’s personal information came to prevail as the dominant concept of data privacy in the 20th century. It is intimately tied to the idea that ‘the individ- ual’ is not static or given but rather evolves over an individual trajectory of self- development, i.e., as an individual “career” (Luhmann 1989; 1997) with the self becoming a “project” (Giddens 1991). By this point, then, we should have gained a deeper understanding of the reasons that led to the lecture attendant’s head- shaking after the 2022 lecture on the private/public distinction. The installa- tion of smart speakers such as the Amazon Echo in people’s private homes af- fects two basic types of privacy at the same time – and two that are guaranteed by basic rights: the security of the private spatial sphere, and that of personal information. For some, it is hard to imagine why anyone would willingly rele- gate these fundamental forms of privacy. The next section (section 4) will present some explanations for this appar- ent “carelessness” on behalf of smart speaker users, and consider them along- side an analysis of what actually happens to the data that is collected and pro- cessed by smart speaker infrastructures that reach into private homes. 36 Voice Assistants in Private Homes. Conceptual Considerations To conclude (section 5), I bring together the insights gained in order to support the argument that smart speakers in the private home form part of surveillance capitalism’s expansion into as many social spheres as possible (Zuboff 2019). The difficulties that data protection bodies have in adapting to this expansion, I propose, are due to the historical context in which measures to protect privacy protection were originally developed – they were tailored to the sphere of labor, and to the practices of work. Whereas individuals’ control of their own personal information is undermined by the requirements of digital, networked self-constitution, practices that take place in the private sphere of the home have only recently been dragged into the realm of social datafication. 2. Private Spheres: Genealogical Remarks on the Private Home In a 2018 essay accompanying their impressive analytical mapping of the sociotechnical planetary infrastructure that constitutes Amazon’s Machine Learning (ML)-based Artificial Intelligence (AI) agent Alexa, Kate Crawford and Vladan Joler sketch out the underlying user scenario propagated by Amazon: A cylinder sits in a room. … It is silently attending. A woman walks into the room, carrying a sleeping child in her arms, and she addresses the cylin- der. ‘Alexa, turn on the hall lights?’ The cylinder springs into life. ‘OK.’ The room lights up. … A brief interrogative conversation – a short question and a response – is the most common form of engagement with this consumer voice-enabled AI device. But in this fleeting moment of interaction, a vast matrix of capacities is invoked: interlaced chains of resource extraction, hu- man labor and algorithmic processing across networks of mining, logistics, distribution, processing, prediction and optimization. The scale of this sys- tem is almost beyond human imagining. (Crawford and Joler 2018, 1) What is so astonishing about the idea of implementing a technical agent that is “silently attending” in one’s private home? Why do some people shake their heads when Echo/Alexa users connect their private homes to a sociotechnical global system that “is almost beyond human imagining”? The first and almost automatic response to this question is that many people find it disturbing to envisage inviting a silent listener that is connected to some infrastructure ‘out Carsten Ochs: The DataEconomy@Home 37 there’ into their private homes. Do we not usually expect external listeners to remain firmly outside our private sphere, the spatial privacy of our homes, where we engage with family and friends, i.e., with those who do not play a functional role, but with whom we choose to share our lives with? Do we not expect these domestic interactions, which constitute our lifeworld, to be none of the economy’s business? Indeed, upholding the spatial privacy of the home is a long-standing social practice that can be traced back to the ancient world of Greco-Roman antiquity and is still performed today, with the sanctity of the home in Germany guaran- teed by article 13 of German constitutional law1. While it therefore might seem somewhat natural to us to expect the private sphere to form a separate realm within society, there is nothing natural about this separation whatsoever. In fact, the status of the spatial private sphere as an experiential realm in its own right, shielded from authorities’ access, and clearly separated from the world of work, is a product of the social history of European societies from antiquity to the present day. As Hannah Arendt has explained, in ancient Greek society, the “oikos” was the homestead of the extended families of Greek patriarchy. It served both as a discrete spatial realm in which families went about their daily business, and as the site of economic reproduction that guaranteed the social position and standing of the family head in the public agora, and thus in Greek society (Arendt 2002, 76–77). In this way, “the distinction between private and public correspond[ed] to a division between two institutional domains – the private domain of the household and the public domain of the body politic” (Gobetti 1997, 104). Notwithstanding that European medieval societies differed, of course, in many respects from those of Greco-Roman antiquity, the family and its home- stead in the Middle Ages continued to play the role of a base from which to op- erate. Even if the head of this medieval type of family did not act in any realm 1 As this remark indicates, the issues dealt with in this chapter are approached from a European perspective by an author based in Germany. The ideas and explanations presented thus relate to the social history of the ‘province’ of Europe, which is not to say that similar developments might not have occurred elsewhere too. For example, it seems that the US approach to privacy is based upon a similar idea of the home, at least this is suggested by Warren and Brandeis’ considerations in their classic “The Right to Privacy” in which they discuss “the sacred precincts of private and domestic life” (War- ren and Brandeis 1890, 195). 38 Voice Assistants in Private Homes. Conceptual Considerations that may be reasonably called “public” or “private” as these terms are used in in- dustrialized times, it was nonetheless the function of the family “to strengthen the authority of the head of the household, without threatening the stability of his relationship with the community” (Ariès 1977, 228). What is more, just as in ancient times, the homestead featured a certain openness compared to the private sphere we have become accustomed to now: “The medieval household mixed up young and old, men and women, servants and masters, friends and family, intimates and strangers. It was open, almost like a café or pub, to the comings and goings of a multitude of diverse types of people, intent upon a bewildering variety of tasks concerned with business or pleasure” (Kumar 1997, 209). Leaving aside structural differences between ancient and medieval “oikos” (see Ochs 2022, 116), it is important to note that medieval family life was prac- ticed within the stratified social order of feudalism. Significantly, for nobles, the family was not positioned in dichotomous opposition to the polis (as in Greek antiquity) or the state (as it is to a certain degree in modernity), but was part of a competitive landscape with all the other families that ruled a partic- ular territorial dominion, always striving to expand their territory (Elias 1997, 95). As territories constantly changed hands, for a long time, medieval forms of rule remained decentralized – there was no overarching central power that could establish itself as a kind of quasi-public counterpart to some quasi-pri- vate familial sphere (Elias 1997, 28; Habermas 1990, 58; Ariès 1991, 7)2. Although sociological (e.g., Habermas 1990) and social history analyses of medieval privacy (e.g., Brandt 1997) disagree as to whether a specifically me- dieval type of privacy can be distinguished, the current state of research invites the conclusion that the development of the familial private sphere occurred as part of the processes of social differentiation that were observable in all areas of early modern society. The compartmentalization of social life (Shibtuani 1955, 567) had a lasting effect on the private sphere: Gradually, starting sometime in the early seventeenth century, this promis- cuous world was ordered and tidied up. Houses – upper-class houses to 2 This is not to say that medieval societies did not recognize any form of privacy at all. Shaw (1996), for example, identifies practices relating to property and to the body in medieval London that reference privacy both in semiotic (use of the word) and practical terms (distinct practices). Nonetheless, there are marked differences between ancient and modern ways of enacting privacy practices (Ochs 2022, 150). Carsten Ochs: The DataEconomy@Home 39 start with – began to reflect a marked degree of segregation of the status and functions of husband and wife, parents and children, masters and ser- vants, friends and family. Boundaries were more strictly drawn – in paths and hedges, bricks and mortar, as well as in social customs – between the private and intimate world of the home and family, and the public world of acquaintances, business associates, and strangers. Work and nonwork (‘living’) were rigidly separated. (Kumar 1997, 209) In the 18th century, the private sphere of the family once again comprised a closed realm, separate from public space and life (Sennett 2008, 18–19; 89–91). A range of oppositional counterparts distinguished themselves from the pri- vate sphere of the home and the family. First, the state evolved from an ab- solutist regime of surveillance (Elias 1997, 282) – “loath to accept the fact that there were certain areas of life beyond its sphere of control and influence” (Ariès 1977, 228) – into the public monopoly on violence and taxation that we are fa- miliar with today (Ochs 2022, 108). Second, and quite relevant for my argument here, the private sphere of the family became gradually separated from the realm of labor. The structural force driving this separation, as many scholars assert, was the sociotechnical drive towards industrialization. In the pre-in- dustrial economies of the Middle Ages, the whole “oikos” of the extended fam- ily’s homestead had been the site of economic reproduction (hence the term “economy” as derived from “oikos”), where economic and other social activi- ties consolidated as a family’s spatial-economic unit (Meier-Gräwe 2008, 116; Lundt 2008, 60–61). When the means and processes of production increasingly shifted to factories and sweatshops, this unit fell apart: the result was a “split between home and factory, a split between economic and other aspects of the parent-child relationship” in workers’ families (Smelser 1967, 31), while in bour- geois society in general, work was separated from the private realm and fam- ilies’ homes were conceived as a private sphere, shielded from labor (Burkart 2001, 403). As the spatial private sphere thus evolved in structural opposition to the state (representing public authority); to the private economy and working world; and also to “public life” in general, a gendering of the separated sphere occurred. The male homo eoconomicus was deemed to belong “naturally” to pub- lic life in all its varieties, while females were considered domina privata (Meier- 40 Voice Assistants in Private Homes. Conceptual Considerations Gräwe 2008, 117)3. At the same time, there was a shrinking of the family, which in the 19th century increasingly came to play the role of a “bulwark against the buffets of a rapidly changing world” (Kumar 1997, 222). With the transition to capitalism inducing massive transformations that unsettled established expectations and practices, actors retreated into the idealized private sphere of the familial homestead, which came to be seen as a refuge from the vagaries of public social and economic life (Sennett 2008, 20)4. Over the course of the 20th century, the shrinking of the “staff” operating in the spatial private sphere continued: The twentieth century has seen the decline and disintegration of the fam- ily as a community, as a collectivity expressing the common purposes of its members. Individualism’s progress, interrupted and held in check in various ways, has continued apace. It has now invaded the family as well as other sectors of society. In the end it’s individualism, not the family that has tri- umphed. (Kumar 1997, 222) Whether or not one agrees with the idea that the family is in a process of dis- solution (the patchwork character of many families rather suggests a de-nat- uralization of the form called ‘family’), most will accept that the private sphere nowadays can be occupied by different constellations such as single persons, familial groupings, or flatmates. But whoever the actors are that claim the pri- vacy of their homes, the closed-shop character of the private sphere as a realm distinct from the working world, from the attention of public authorities, and from uninvited listeners representing the economy or the general public, re- mains a widespread normative expectation5. 3 The picture drawn here is an accurate, yet simplified one, as empirical reality is always more messy than historical analysis suggests. For detailed and at the same time con- troversial accounts of the gendering of public and private spheres in industrial society see Hausen (1976); Pleck (1976); and Lundt (2008). Please note that despite the ways in which these researchers’ views differ, they largely agree on what counts for the argu- ment of this chapter: the spatial private sphere (of the family) began to separate from that of work in the 17th century and gradually became a distinct realm. 4 At the same time, the private sphere of the family became the site of gendered violence, especially against women and children (Müller 2008); the 20th-century “women’s movement” therefore re-politicized the private in order to render patriarchal violence accessible to public intervention (Lundt 2008, 51). 5 The phenomenon of the ‘home office’ in the course of the COVID-19 pandemic tem- porarily blurred the boundaries between the private home and the working world. Carsten Ochs: The DataEconomy@Home 41 So, here we have our first explanation for the head-shaking of people who feel disturbed by the introduction into the home of listening devices that are deemed at least potentially capable of transmitting recorded audio to an un- known audience: such persons are uneasy about the unsettling of the closed shop that they still expect the private sphere of their homes to encapsulate. 3. Information Control: Privacy in the 20th Century AI-equipped smart speakers and the infrastructures they form part of disturb people’s entrenched expectations concerning the exclusivity of the private home; its separation from the economy, from the realm of work, and from external observation in general. A further aspect that normative attitudes towards smart speakers relate to are issues of privacy and data protection. What is called “information privacy” in social theory (e.g., Rössler 2001, 45) usually goes under the name of “data protection” in regulation. Smart speakers seem to affect this idea of privacy/data protection, because as human agents we are visible in almost every interaction with technological platforms. We are always being tracked, quantified, analyzed and commod- ified. But in contrast to user visibility, the precise details about the phases of birth, life and death of networked devices are obscured. With emerging devices like the Echo relying on a centralized AI infrastructure far from view, even more of the detail falls into the shadows. (Crawford and Joler 2018, 12) This may well be true, but why is it at all noteworthy that we “are always be- ing tracked, quantified, analyzed and commodified”? Couched in social theory terms: why should information privacy (to be distinguished from the private sphere) be an issue at all? What is the meaning of “information privacy” in the first place? And, how did information privacy become an entrenched practice in contemporary digital society’s genealogical forerunner – 20th century Euro- pean modernity? To answer these questions, I will begin by offering a general sociological characterization of 20th century high modernity, before focusing on the issue of self-constitution and privacy. However, I will not discuss here whether these developments have had structural con- sequences for people’s normative expectations concerning the privacy of their homes. 42 Voice Assistants in Private Homes. Conceptual Considerations According to Andreas Reckwitz (2006, 275), the early decades of the 20th century marked the end of bourgeois cultural rule. The period witnessed a mas- sive expansion of space–time relations, enabled by innovations in technolo- gies of transport, communication, media, and production (Berger, Berger, and Kellner 1975; Beniger 1986). At the same time, social life came to be increasingly structured by large organizations, such as unions, associations, people’s par- ties, huge corporations etc. – an observation that has led sociological analysis to characterize, roughly speaking, the first half of the 20th century as “Orga- nized Modernity” (Wagner 1998). Nazi barbarism, totalitarianism, and the two industrialized world wars of the “short 20th century” (Hobsbawm 1994) could not have taken place without Organized Modernity’s capacity to assemble peo- ple by sociotechnical means at a huge scale; and to construct for them collective identities based on the sometimes violent and lethal exclusion of “othered” (i.e., purposefully generated) “outsiders” (Bauman 1989; Wagner 1998, 68–69; Arendt 1975). After World War II, European post-war societies passed into what has been called “Reflexive” or “Second Modernity” (Beck 1986; Beck, Giddens, and Lash 1994), within which self-constitution became an ever more individualized process that was to be realized by neo-liberalism’s structurally “released” – and also isolated – actors themselves. The shifting logic of self-constitution mirrors the transition from Orga- nized to Second Modernity. The beginning of the short 20th century witnessed the appearance of “organization man”, a social figure who tended to follow a career largely predetermined by organizational environments (Reckwitz 2006). A typical trajectory of “organization man” would lead him through organizations that aim to provide their members with a “corporate identity” (Whyte 2002). In such settings, organizations strive to fix their members’ identities (Mönkeberg 2014), because stable – or rather stabilized – identities can be easily integrated into large organizations and formalized sequences of operation (e.g., production under Taylorism). However, while organizations demanded stable identities, the mass media (radio, TV) and urbanization began to make it plain for all to see that “[m]ost people live more or less compartmentalized lives, shifting from one social world to another as they participate in a succession of transactions” (Shibutani 1955, 567). For 20th century subjects, it came to be taken for granted that “[d]ifferent sectors of their everyday life relate them to vastly and often severely discrepant worlds of meaning and experience” (Berger, Berger, and Kellner 1975, 63). Whereas in the 19th century, everybody had implicitly known that they lived “compartmental- ized lives”, radio and television rendered visible this compartmentalization Carsten Ochs: The DataEconomy@Home 43 of life by putting the pluralism of social worlds on display simultaneously (Berger, Berger, and Kellner 1975, 64 ff.; Goffman 1959). Now, everybody knew that everybody knows that everybody lives compartmentalized lives. As a result, the idea of the self as an undivided coherent whole, which de- fined early modernity’s notion of the individual, begins to seem increasingly unsustainable. Sociologists monitor closely how actors moved in everyday life and over the life course through different social worlds and organizational contexts that offer contradicting rules and roles. Pierre Bourdieu (1987) elabo- rates in a virtuoso manner how people in European post-war societies came to terms with the different social worlds and areas they passed through, how they continually adapted themselves and developed further instead of self-consti- tuting as a static self with some singular once-and-for-all core identity. In 20th century high modernity, processes of self-constitution were obliged to incor- porate frequent changes of subjectification schemes as well as organizations’ identity fixations. The mechanism that allows people to reconcile continuous change with the constancy of corporate identity is the career mode (Luhmann 1997, 742). Facilitating the organizational channeling (fixation) of develop- mental trajectories (movement) through society, it became subjectification’s key mechanism. Giddens (1991) accounts for 20th century self-constitution with the concept of the “reflexive project of the self”, while Goffman sheds light on the informational aspects of practicing such a self. The project-self is habitually bound to play contradictory roles, for “[i]n each [social] world there are special norms of conduct, a set of values, a special prestige ladder, charac- teristic career lines, and a common outlook toward life – a Weltanschauung” (Shibutani 1955, 567). Given the potential contradictions between contexts, it becomes imperative for individual project-selves to separate the audiences associated with different roles from one another, and to hide internal incon- sistencies. Individuals are obliged to establish “audience segregation”, and to do this, the project-self takes measures to control which audiences have access to which elements of their personal information (Goffman 1959). Hence, over the course of the 20th century in Euro-American society, boundaries came to be drawn between different types of information. As long as these boundaries were not crossed, “contextual integrity” (Nissenbaum 2010) remained intact. In the 1980s, a conflict arose in Germany that led to the practice of individ- ual information control becoming a case of legal dispute: the right to informa- tional self-determination. At the time, “new social movements” were evolving, addressing issues such as women’s rights, environmental protection, discrimi- nation, etc. (Beck 1986). Extending the objectives of German social movements 44 Voice Assistants in Private Homes. Conceptual Considerations beyond labor issues, these movements contributed to a generally politicized atmosphere, marked by the Cold War and accompanying controversies. It was in this tense political atmosphere that the German government announced its intention to conduct a census (Berlinghoff 2013). Fueled by the politicized Zeitgeist, a large-scale controversy erupted. Before long, advocates of data protection who were worried about government surveillance had filed a suit to the German Federal Constitutional Court. Crucially, the conflict unfolded against the backdrop of the computerization of administration and heated debate about data protection (Frohman 2013). The Constitutional Court’s response was sensitive to this and explicitly pointed out the potential dangers of the networking of data across informational contexts. It argued that, as citizens, people might feel pressurized to hide their political commitments if they knew they were being monitored from a central point of observation. For this reason, the court ruled, information about persons’ political activities must remain private (BVerfG 1983). The verdict of this Volkszählungsurteil asserted that any German citizen has the general right to control who knows what about them, at what point in time and for what purpose – because if they did not, they might not be able to en- gage freely in self-development, and in the processes of self-constitution. This is ultimately a legalistic articulation of the view that any individual actor, in or- der to self-constitute as a Giddensian “project-self” (Giddens 1991), or to follow a Luhmannian “career” (Luhmann 1989), must be able to regulate what infor- mation concerning their person is accessible to actors from the various social contexts and worlds that that individual passes through. Arguing along simi- lar lines of reasoning, the court translated the everyday practice of information control into the right to information self-determination (Rössler 2010, 45). Individual information control became the dominant privacy practice of the 20th century because it allowed the project-self to deal with the contradic- tion between corporate identity fixation and ongoing personal development. The court mobilized this practice and turned it into a legally guaranteed right when the practice appeared to be coming under threat from a novel type of emergent public enabled by digital networking – which was already discussed in the data protection discourse of the 1980s, although the internet at that time was but a far cry from being part of digital everyday practices (Steinmüller 1988). Even so, a technological innovation that facilitated the flow of informa- tion across borders was already on the horizon, threatening to disrupt “con- textual integrity” (Nissenbaum 2010). Nevertheless, the right to control who has access to one’s own personal information still forms the basis of current Carsten Ochs: The DataEconomy@Home 45 data protection law, and Amazon’s Echo and Alexa operate in a techno-legal environment that is still largely informed by the idea of individual informa- tion control. This raises the question of whether these technologies contribute to the border-crossing of information flows, and, if so, what the consequences are in terms of social structuration. Perhaps those who shake their heads at the thought of Alexa implicitly assume that there will indeed be consequences? Let’s render this assumption explicit. 4. Digital Self-Constitution and Machine Learning@Home Having gained some clarity regarding the different conceptualizations of the private that seem to be somehow affected by the integration of smart speak- ers and AI assistants into private homes, we can now move on to consider the functionality of these technical apparatuses, i.e., the purposes they serve and operations they perform once they have been installed in people’s homes. From the perspective of Echo/Alexa users, smart speakers are there to increase au- tomation and convenience. At least, that is Amazon’s great promise. Describing a 2017 promotional video advertising the Echo, Kate Crawford and Vladen Joler observe: The video … explains that the Echo will connect to Alexa (the artificial intel- ligence agent) in order to ‘play music, call friends and family, control smart home devices, and more.’ … The shiny design options maintain a kind of blankness: nothing will alert the owner to the vast network that subtends and drives its interactive capacities. The promotional video simply states that the range of things you can ask Alexa to do is always expanding. (Craw- ford and Joler 2018, 3) As the authors go on to point out, firstly, the smart speaker itself appears to be just an “‘ear’ in the home” but is actually far more than that: “a disembod- ied listening agent that never shows its deep connections to remote systems” (Crawford and Joler 2018, 5), by means of which the private home of the Alexa user is connected to an extensive infrastructure that is inaccessible to the user. Second, the device seems to have been designed to remain unnoticed, and is notably unrevealing of its connection to the external infrastructure. And third, the number of tasks Alexa can fulfill is pro