The self-perpetuation of the promise of care robots how doubtful application scenarios become promising

This article employs a narrative approach to explain the self-perpetuation of the promise of care robots. For two decades, care robots have been viewed as a promising digital innovation that will relieve care workers from service tasks and thus improve the situation in the care sector. Though they still show a rather poor performance in real-world applications, the expectations are as high as ever. To explain the self-perpetuation of the promise of care robots, we investigate sensemaking practices in the discourse about care robots and show how a predominant overarching narrative as well as project-related sensemaking practices play out.


Introduction
Care robots, defined as service robots 'that operate partly or fully autonomously performing care-related activities for people with physical and/or mental handicaps […] related to age and/or health-related restrictions' (Goeldner et al., 2015:115), have been considered to be a promising technology for the last 20 years (Tietze & McBride, 2020:5−13;Lee et al., 2018).This applies especially to collaborative care robots, on which this article focuses.Collaborative care robots are robots that work directly with care professionals in conducting care-related tasks. 1 Real-world applications that are actually working, however, are still far away for most collaborative work tasks (van Aerschot & Parviainen, 2020).Nevertheless, numerous care robot projects have been financed by funding agencies, conducted by robot developers and supported by care professionals and care organisations with reference to this claim.Consider, for instance, the application scenario of the beverage-serving care robot.Getting elderly people to drink enough to prevent dehydration is a major issue in elderly care.A robot able to take over the task of serving beverages thus would be truly helpful.This idea has existed since the beginning of care robotics.An early example from Germany is the second generation of the Care-O-bot that was employed from 2002 in test applications in care facilities to carry out this task (Compagna, Derpmann & Mauz, 2009:18−19;Graf, Hans & Schraft, 2004).Until today this application scenario has been a frequent topic of care robot projects (cf.for example, Graf, Parlitz & Hägele, 2009;Emeli, Wagner & Kemp, 2012;Chivarov et al., 2014;Juel et al., 2020).Participants of care robot projects have repeatedly managed to convince themselves and others that it makes sense to choose this application scenario, though its realisation is still in its infancy.Most of the other application scenarios of care robots are characterised by the same continuing discrepancy between promise and reality.This is not to say that there are no care-related robots at all that are already in use in care facilities.Some have been in use now in areas such as companionship, cognitive assistance, telepresence and monitoring (for robots as companions of elderly people see, for example, Lu et al., 2021).Recent studies show that, if successfully integrated into the workflows of care professionals, robots applied in these areas can have a positive effect on health outcomes and can support care professionals (for example, for the areas of telepresence and monitoring in critical care see Teng et al., 2022).However, these applications usually do not require the robot to be able to collaborate directly with care professionals and the robots do not have the degree of autonomy necessary for collaborative tasks (for example, the means for physically manipulating the environment).
Alongside industrial production, care work has become one of the most prominent domains of research, testing and application for collaborative robots (see, for example, Fischer, Krings, Moniz & Zimpelmann, 2017:11−12;Tietze & McBride, 2020:13−42).Collaborative robots are designed to work in direct physical interaction with humans without endangering them, a precondition for employing robots in elderly care as in other close contact services.While in industrial production highly controllable and standardisable work environments facilitate the employment of collaborative robots, the less controllable and standardisable work situations of close-contact services such as elderly care still constitute a major obstacle for robotic automation.The rise of care work as a main domain of application for collaborative robots thus needs explanation.What makes care robots such a promising technology for the care sector?
We suggest that this success of care robots can be explained as resulting from discursive practices of sensemaking.We argue that the sensemaking practices that are related to them result in a self-perpetuation of the promise of care robots which thus becomes decoupled to some extent from the actual progress of care robot development.It is not the discrepancy between promise and reality as such that distinguishes the development of care robots from other technologies that are considered as promising.On the contrary, this discrepancy is a constitutive part of the dynamics of developing new technology, as research on promising technologies has shown (cf.van Lente, 1993;Van de Ven, Polley, Garud & Venkatraman, 1999:31-32;Borup, Brown, Konrad & Lente, 2006).These dynamics usually either lead to closing the gap or to reconsidering the promise.In the case of care robotics, however, we find a self-perpetuation of the promise while the gap between promise and reality remains wide.
Our empirical analysis considers sensemaking practices at two different levels of the discourse about care robots.At the level of explaining, legitimising and justifying the need for developing care robots in general, we identify a powerful narrative we refer to as the nursing crisis narrative.According to findings from extensive literature on narratives, the sensemaking capacities of narratives result from their temporal structure.The plot at the core of a narrative presents a sequence of events as inevitable and thus provides a form of sensemaking that makes it possible to avoid complex and potentially controversial causal explanations (Vaara et al., 2016:498;Polletta, Chen, Gardner & Motes, 2011:111;Patriotta, 2003:351).As we argue in this article, the sensemaking capacity of the nursing crisis narrative contributes strongly to the self-perpetuation of the promise of care robots.
In addition to this narrative, we identified a number of sensemaking practices that are related more specifically to the circumstances and challenges of the particular projects of developing and implementing care robots.At the project level, the discrepancy that calls for sensemaking is the discrepancy between the application scenario (what the care robot under development will be able to do) and what the robot is actually able to do.We identified three sensemaking mechanisms for dealing with the gap between promise and reality at the project level: first, the rhetorical persuasiveness of prototype scenarios; second, the logic of the 'not yet'; third, the abstraction from real-world conditions.The remainder of this article is organised as follows: first, we describe the research design of the study and the empirical data on which this article is based.Then, we present our considerations and empirical findings on how the nursing crisis narrative contributes to the self-perpetuation of the promise of care robots.Afterwards, we describe the sensemaking mechanisms for dealing with discrepancies between promise and reality at the level of the particular projects of care robot development.The final section provides a summary and discussion of the results.

Research design and empirical data
The empirical data and findings in this article are from an ongoing research project on human-robot co-work2 in which we investigate the development of care robots and industrial collaborative robots (so-called cobots).One of the main research questions of the project is about how the development of collaborative robots is negotiated via sociotechnical scenarios, concrete pictures of a future situation in which the application of the new robots is envisioned as if it were already a reality.For reconstructing this process, which is either driven by technology supply or by societal demand, we distinguish three successive manifestations of these scenarios: first, the exploration of initial ideas about a new kind of human-robot co-work that takes place in the early development phase, typically in engineering laboratories; second, the more concrete, but still idealised, test environments for the imagined applications in settings suitable for controlled trials; and finally, attempts at realistic implementations for the actual workflow.In the sequence of the manifestations of these scenarios, the new technical functionalities become (more or less systematically) adapted to the requirements, needs and hopes of the human workers and their practices and vice versa.The same applies with respect to the wider context, such as organisational, legal or discursive structures.Because the sequence of the three types of manifestations of a scenario implies confronting the ideas built in the laboratories more and more with the − technical as social − constraints of present real-world applications, in which they have to fit, scenarios in our view can be seen as negotiation arenas for the social construction of technology (Schulz-Schaeffer & Meister, 2019).
Methodologically, our project follows a case-based approach (Yin 2018), where the unit of investigation is the scenario itself in its different manifestations.To reconstruct the scenarios and the respective negotiation processes, we draw on a broad range of empirical data: descriptions and especially videos of the envisioned scenarios at work, popular and scientific articles and relevant secondary publications.Moreover, as part of our case study-based approach, we conducted semi-structured interviews for each case with the most relevant protagonists: the visionaries, the engineers who did the technical specifications, the management of robotics companies, the mundane human professionals in the respective domains (for example, care workers), the officials responsible for domain applications (for example, care service and home management) and officials for public funding.
All interviews were transcribed word by word.Non-verbal utterances and pauses in the flow of speech of the interviewees were disregarded because our interview analysis focuses on the reconstruction of arguments.Therefore, the spoken language was slightly smoothed (Gläser & Laudel 2009:193).As a result of the pandemic, most interviews were conducted online, and at the time of writing this article, we started to complement these with on-site visits, which we are documenting in research diaries.
The analysis of the empirical data follows the approach of grounded theory (Strauss et al., 1996:50), where first hypotheses, drawn from the research question or from findings from our material, are iteratively developed and confronted with the empirical data, both on the level of the analysis of the single cases and on the level of the overall interpretation.Technically, we did this as a coding process using the content analysis software ATLAS.ti.
Our criteria for choosing cases for investigation were, first, that (the aim towards) a realisation in the application domain is evident, and second, that exemplars of all three manifestations of the scenarios are covered, whereas one case may cover more than one stage of manifestation.As this is ongoing work, we so far worked out case descriptions and interpretations for four cases for each domain under investigation (professional care work and industrial production).From our first interpretations of all cases of care robotics, we found strong evidence that the construction of the scenarios and the negotiation processes related to them cannot be understood without taking a shared discursive background structure − the nursing crisis narrative − into account.This article is a systematic elaboration of this empirical evidence.
For this article, we chose data from three cases within the domain of professional care.Two of the cases can be situated at an early development stage of a robot prototype, whereas the other case lies in between test and trial runs of a fully developed prototype and first, real-world applications.Thus, all three stages of manifestation mentioned above are represented in this sample to some extent.
As the origin and development of discursive structures seem to be important for understanding the processes we are interested in, we have in addition analysed research articles about care robot projects.We focussed on literature about the technical development of robots where care professionals had participated in this development process.To create the sample of literature to be included, we took a two-step approach.First, we explored the first ten result pages of the German and English language versions of Google Scholar using the combined search terms 'work ' , 'care' and 'robot' and 'work' , 'nursing' and 'robot' (in German: ' Arbeit' , 'Pflege' and 'Roboter').For every article we found, we checked the abstract and excluded all articles that did not address the technical development of a robot meant for collaborating with care professionals.The remaining articles were either technical descriptions of a developed care robot or overview articles.Second, we searched the overview articles themselves for references to other or earlier robot developments, to include early examples of development in our sample.In this way, we created a sample of 27 research articles.The oldest publication in our sample dated from 1992 (Hashino 1992), the most recent ones in our sample were from 2020 (for example, Juel et al., 2020).These were reviewed for justifications or reasons given for the development of the robot in question.

The narrative of the self-perpetuation of the promise of care robots
In our empirical data, we find a recurring line of argument for explaining and justifying the need for developing care robots.Drawing on the literature on narrative sensemaking, we interpret this as a narrative and refer to it as the nursing crisis narrative.There is some literature about the influence of sensemaking practices and strategies on how and why care robots are developed.This literature, however, does not employ the concept of narrative.It has been observed that governmental support for care robotics and the roboticists' design practices have promoted a one-sided cultural conception of care work, which has been criticised as technocratic, paternalistic and lacking substantial input from elderly persons (Neven, 2010;Hsu et al., 2020;Wright, 2020).Furthermore, the rhetorical character of the discourse about care robots has been addressed, for instance, in how 'the use of the term "assistance" is consciously linked to traditional metaphors of the care setting, including welfare, companionship and empathetic support, in order to increase the social acceptability of these new technologies' (Krings & Weinberger, 2018:2).It has been argued that this discourse has impeded alternative concepts of 'care centred value-sensitive design approaches' (Van Wynsberghe, 2013:407).The potentially unique case of care robotics in Japan has attracted special attention.Here, it has been discussed if and how a specific combination of an aggressive governmental innovation policy and a broader cultural affinity to robotic devices is the reason that Japan represents a 'robotopia' (Wagner, 2014;Sabanović, 2014;Ishiguro, 2017).Though these and other approaches address the discourse about care robots from the perspective of sense-making, none of them relate the observed sensemaking strategies and practices to the specifics of narratives.
Narratives are 'chronological accounts that outline the state of an issue, what has led to it and how things will or should play out' (Boswell, 2013:620).They 'offer a particular point of view on a situation through the use of a plot' (Bartel & Garud, 2009:108), having a beginning, middle and end (Jones, McBeth & Shanahan, 2014:6).This 'temporal aspect of narratives […] distinguishes them from other forms of discourse' (Vaara et al., 2016:498).In contrast to other forms of sensemaking, with narratives it is the plot which provides the 'causal links between events' (Polletta, Chen, Gardner & Motes, 2011:111;cf. Polkinghorne, 1988).Presenting and perceiving the plot as an inevitable course of events, as a plausible description of a development or as commonsense knowledge is the way in which narratives are deployed to select and institutionalise meanings and to legitimise and justify courses of action (Patriotta, 2003:351;Mintrom, Rublee, Bonotti & Zech, 2021).

The nursing crisis narrative
In a nutshell, the nursing crisis narrative goes as follows: due to the demographics of ageing societies, an increasing percentage of the population is made up of elderly people.Accordingly, the number of people in need of care is constantly growing.At the same time, there is a shortage of nurses, which will become only more severe in the future.Considering the increasing shortage of human work in the care sector, care robots should be developed and deployed to fill the gap.This narrative is very prominent in the discourse on care robots.It occurs in 22 of the 27 research articles about care robot development projects included in our sample and also in many of the interviews.
Whenever the narrative is referred to in the literature or in the interviews, this is done quite similarly.Typically, the narrative starts with describing the current situation as characterised by an ongoing demographic change towards an 'elderly society' with an 'ageing population': 'many societies are facing a demographic shift.In 2015 8,5% of the global population was aged 65 or above.This number is projected to increase to 17% by 2050' (Juel et al., 2020:1; cf. Lee & Naguib, 2020:6).
According to the narrative, the demographic change is of particular importance for the topic of elderly care.It turns elderly care into a critical social issue and a major challenge: 'this rapid transition to an elderly society makes elderly care an ever more critical social issue worldwide' (Lee & Naguib, 2020: 6;cf. Carros et al., 2020:1).This is for two reasons.One reason is the rising demand for care work: 'inherited [from the demographic shift], an increase in multi-morbidity is expected which causes prolonged, complex and transverse patient care' .(Juel et al., 2020:1) The other reason is a shortage of care workers: 'as the society's aging rapidly advances, the shortage of workers and care workers has become a major issue' (Yamamoto et al., 2018:1;cf. Carros et al., 2020:1).
The narrative does not provide substantial arguments why the current shortage of care workers will necessarily turn into an even worse situation in the future and why it should not be possible to train and employ enough care workers for the future demand.Rather, it relies on the intuitive plausibility of the temporal logic of the plot: an increasing percentage of elderly people is equated with an increasing demand for more care work and an increasing shortage of care workers.This is where the robots enter the stage and are presented as a solution to the problem: 'in this situation, it is natural to consider how robotic technologies, in particular elderly care robots, can contribute to a solution for the above social challenge' .(Lee & Naguib, 2020:6) 'Given the issue of an ageing population and shortage of medical and nursing staff in many countries, this naturally leads to attempts to use robotics and automation addressing this problem' (Miseikis et al., 2020:5339).
The narrative portrays care robots as the most obvious ('natural') path for dealing with the diagnosed social challenge.In fact, this is far from being obvious.Even if it turns out to be impossible to recruit enough care workers to meet future demand, there are obviously other technological or non-technological ways of improving the situation.Again, the narrative provides no substantial arguments but relies on the intuitive plausibility of the storyline (for a more detailed account of the reasons for this assessment see below).

How the narrative is used to justify care robot projects
In our empirical data, we see three different ways in which the nursing crisis narrative is used to justify care robot projects.One way is simply to refer to the narrative: The US population is aging at an alarming rate.At present, 12.5% of the US population is of age 65 or older.It is widely recognized that this ratio will increase as the baby-boomer generation moves into retirement age.Meanwhile, the nation faces a significant shortage of nursing professionals.[…] This acute need provides significant opportunities for robotics and AI researchers to develop assistive technology that can improve the quality of life of our aging population, and help nurses become more effective in their activities.The Nursebot project was conceived in response to this challenge.[…] The overall goal of the project is to develop mobile robotic assistants that can assist nurses and elderly people in their daily activities.(Pineau et al., 2003: 271−72) So, the idea is to counteract this nursing overload that is currently taking place, or this nursing shortage, a little bit [...] And we have dealt with the question of how this can be counteracted accordingly.[…] we have also played a bit with the future perspective of how a robot can later perhaps carry out wound care itself, for example, if it is precise enough to practically apply these bandages itself. 3Engineer, IC09 #00:54:44-3#) The second and third ways elaborate on the basic version of the narrative by adding particular characteristics of care work that cause particular additional challenges.The care robot project is then presented as a way to deal with these challenges.
The second way refers to the characteristic of care work that it includes physical demanding labour.The argument goes as follows: in the situation of the nursing crisis, the load of physical labour becomes too much for care workers.Thus, the particular project develops a robot that takes over tasks that are physically challenging in order to reduce the nurses' physical workload.The following quotation from an interview provides an example (for another example see Mukai et al., 2010:5996): We are short of nurses.The ones we do have are under too much physical strain, which of course doesn't make the job any more attractive.And this is the motivation.We want to develop a tool, and in this case, we have opted for collaborative robotics, which takes over various posture and repositioning tasks on the patient.Yes, that is the initial goal of the project.That is, in very rough summary, we want to reduce the physical burden on the nursing staff via the assistive device.(Engineer, IC12 #00:03:24-0#) The third way refers to the characteristic of care work as a combination of providing care-related services ('Versorgungsarbeit') and emotional work ('Sorgearbeit').As Strauss and co-authors have put it, care work consists of a 'complex interplay between the standard occupational assignment of tasks and the subtle weaving in and out of occupationally sentimental work' (Strauss, Fagerhaugh, Suczek & Wiener, 1982:270).Care work is thus always about synchronising the provision of services with personal affection and affective balance (Remmers, 2018:167).Due to the nursing crisis, according to this line of argument, these two aspects of care work are increasingly becoming out of balance because the growing workload does not leave enough time for emotional work.The particular care robot project, then, is introduced as a way to regain balance by relieving the care workers from service work tasks.The following quotations from interviews illustrate this line of argument: Otherwise, one of the service staff would have to deliver these drinks.[The care robot] takes that already partly away from them and then this person has the opportunity to do something else.[…] This is of course the ideal: One has more time for these value-adding tasks and these annoying routine things are done by

How the nursing crisis narrative contributes to the self-perpetuation of care robotics as a field of digital innovation
The fact that research articles on care robot development refer so frequently to the nursing crisis narrative shows that it is a well-established line of argument within the discourse about the situation of elderly care and its future.The earliest reference we found is from 1998 (Schraft & Schmierer, 1998), indicating that this well-established narrative is also quite persistent. 4The narrative is usually presented in the introductory passages of these articles.It is thus positioned at a point where authors typically explain why their researches are important, which problems and research questions they address and how they contribute to solving them.One might argue that for many of these articles, it is by imitation that the narrative occurs in their introductory sections and that their authors simply repeat an argument they know from literature as being often used without thinking more about it.This is probably true.But this is exactly how narratives work: by distributing meanings that are taken for granted.Our interviews with participants in care robot projects show that the narrative is no less prominent when they talk about their projects.Together, these findings suggest that the nursing crisis narrative is a major argumentative means for explaining, legitimising and justifying the need for care robots and the importance of the related R&D.
The nursing crisis narrative sequences the events in such a way that developing and employing care robots emerges as the most obvious next step.Within the narrative, this next step follows 'naturally' from the sequence of events.Upon closer examination, however, it is highly debatable if the cause-and-effect relationships are as inevitable as implied in the narrative.Let us re-examine the argument step by step: 4 Schaft & Schmierer (1998) is an article from a handbook of service robotics.Since handbooks usually refer to an already established state of the art, the nursing crisis narrative is probably even older.The first part of the narrative (without care robots as the solution) can already be found in Hashino 1992:97−98.
First, the narrative argues that an increasing percentage of elderly people leads to a growing number of people in need of care.The data indeed show a growing number of people in need of care.The German Federal Health Monitoring System reports an increase of 76.5% (from 2,338,252 to 4,127,605 people) between 2009 and 2019. 5 During the same period, however, the number of people in Germany aged 65 and over grew from 16.9 million to 18.1 million, an increase of only 7%, and the number of people aged 85 years and over grew from 1.9 million to 2.4 million, an increase of 26% but also far from the 76.5% increase of people in need of care. 6Thus, these numbers indicate that the aging society is not the only and maybe not even the most important reason for the increasing number of people in need of care. 7 Second, the narrative argues that the already existing shortage of nurses will become more severe in the future.The data indeed show that the number of care workers does not keep pace with the increase of people in need of care.While the latter increased by 76.5%, the number of professional care workers in Germany increased only by 36.8% between 2009 and 2019. 8By extrapolating from these past developments into the future, the narrative arrives at the diagnosis of an increasing shortage of nurses.Arguments by extrapolation, however, assume that the factors that have been responsible for past developments will also be at work in the future.Arguments by extrapolation treat all these factors as taken for granted.Consequently, there is no need to take them into consideration as causes of the development at issue that one would want to understand in order to influence its future dynamics.According to Krings et al. (2014), the view that demographic change necessarily causes a shortage of care work results from this lack of interest in many other factors, which, in combination with the demographic changes, constitute the reasons for the shortage of care work.They argue that this view implies that everything else remains the same while the demographics change (Krings et al., 2014:74−75).This ignores the way that the need for professional care and the means to meet this demand results from a complex interplay of heterogeneous factors, which, for instance, influence the relationship between residential care and home care, the relationship between professional care and care by relatives, the number of people willing to work in elderly care, or the healthiness and the related need for care of elderly people (cf.Krings et al., 2014:73−77).
Third, the narrative suggests that care robots should be developed and deployed as a way to deal with the increasing shortage of human work in the care sector.Even if it turns out that without technological innovation there is no way to balance the demand and supply of care work, the question that remains to be answered is which kind of technology would be best suited to support care workers.There are at least two reasons why care robots are not the natural technological solution the narrative presents them as.For many of the tasks for which care robots have been developed during the last two decades, technological solutions that are technologically less advanced, easier to develop and less expensive are available.For instance, for the task of raising patients from a lying to a sitting position, it makes more sense to add the necessary functionalities to hospital beds and wheelchairs than to develop care robots such as the RIBA II of the RIKEN-TRI Collaboration Centre for Human-Interactive Robot Research (http://rtc.nagoya.riken.jp/RIBA/index-e.html).Another reason is that after two decades of research on care robot development, there are still very few care robots that actually relieve care workers from service work (van Aerschot & Parviainen, 2020).There is little evidence that this will change substantially in the next few years.This is also reflected by most of our interviewees who project a feasible robotics solution in the care sector into the next 15 to 20 years.The current shortage of human work in the care sector, however, calls for solutions that are available as early as possible.If technological innovation is expected to contribute soon to improving the current situation in the care sector, this is certainly not innovation in the field of robotics.
A closer look at the cause-effect relations implied in the nursing crisis narrative thus shows that developing care robots is by no means a 'natural' way to deal with the current situation of the care sector.Nevertheless, robot developers are continuously successful in acquiring project funds for care robot projects and in winning care facilities as project partners for these projects.We believe that the nursing crisis narrative contributes to this success in at least two ways: first, by its rhetorical persuasiveness; and second, by its capacity to align the divergent views and interests of several groups of actors relevant to furthering the development of care robots.
The persuasiveness of the nursing crisis narrative emerges from its temporal structure as a story that tacitly implies the underlying cause-effect relationships and thus establishes them as taken-for-granted knowledge.Using the term 'natural' to characterise care robotics as a solution is but one of the rhetorical means of presenting the narrative's sequencing of events as self-evident.Though the narrative has existed for more than 20 years, the temporal sequence of events it consists of has remained unchanged.Whether the narrative has been referred to in the past or whether it is presented today, it always starts with the demographic change, continues with the increasing shortage of care work and ends with care robotics as a solution.In this way, the promise of care robotics becomes continually renewed and rejuvenated as a promise for the near future of care work while the past failures to fulfil this promise vanish from the picture.Thereby, the nursing crisis narrative helps to perpetuate the promise of care robotics and contributes to the self-perpetuation of care robotics as a field of digital innovation.
Concerning the narrative's capacity to align heterogeneous actors, we refer to an argument developed by Benjamin Lipp (2019).Drawing on the Foucauldian concept of dispositif (or 'apparatus'), Lipp conceptualises what we call the nursing crisis narrative as part of an 'apparatus of innovation' .He argues that 'this apparatus of innovation works towards interfacing robotics and care within various milieus' (Lipp, 2019:78) by 'construct[ing] and respond [ing] to the urgency to interconnect technology and society, assistive devices and demographic change, business and the state, elderly people and the market in ever more ubiquitous and intimate ways' (Lipp, 2019: 85).
To be more specific, the nursing crisis narrative aligns the views and interests of robot developers, state agencies and care workers in the following way.For developers of robots, the narrative promises the opportunity to conduct R&D in a domain of application that is technologically demanding and at the same time of high societal value (by contributing to solving a societal crisis), meets a real demand and is in accordance with the professional expertise of the care workers in the domain of application.For state agencies involved in innovation policy, the narrative backs up a re-orientation of funding schemes for care robotics research that requires the inclusion of the expertise of care professionals.It thus supports a policy strategy to overcome the 'disconnection between development and implementation' (Wright, 2020:10) in care robot development.For care workers, the narrative connects the employment of care robots with the promise of restricting automation to service work tasks, thus relieving the care workers from those parts of their work that are purely instrumental while giving them more time for what many of them view as the most important and valuable part of their jobs: emotional work.By aligning views and interests that are important to these relevant groups of actors (for a critical account of this alignment see Maibaum, Bischof, Hergesell & Lipp, 2021), the narrative gives all of them a reason to stick to it even though there is still little success in actually employing care robots in the field of care work.This is, in our opinion, a second mechanism by which the nursing crisis narrative contributes to the self-perpetuation of care robotics as a field of digital innovation.

Self-perpetuation of robot-based problem-solving through making sense of doubtful application scenarios
The persuasiveness of the nursing crisis narrative helps to explain why care robot development is viewed as promising in spite of all the other technical and non-technical options for improving the situation of elderly care that could be realised more rapidly and more easily.It also helps to explain why certain kinds of application scenarios for care robots (applications where the robots take over service work tasks and physically demanding work tasks) prevail.The nursing crisis narrative creates a discursive reality in which care robots remain generally promising.This way, the narrative provides a supporting background for particular care robot projects, as shown above.The sensemaking provided by the nursing crisis narrative is, however, not specific enough to provide orientation with respect to the particular ways that problems are perceived in individual projects.Thus, we need to find additional mechanisms for explaining, legitimising and justifying the development of care robots in these contexts.
We identified three mechanisms of this kind.Before discussing them, we will describe how the use of care robots as a technology for supporting care workers is questionable and thus in need of sensemaking at the level of particular development projects.This can best be demonstrated by cases in which there is close cooperation between technology developers and care professionals, which is the usual constellation in care robot development projects, and by cases in which there is additionally a test application of the robot under development in a care facility.Consider, for instance, the beverage-serving robot project mentioned earlier.The care manager of the participating care facility describes the application scenario as follows: The robot has a specially designed tray on its back, where four special cups fit in.

It was then given another position at a table where our residents often sit [in addition to the position of the kitchen where it collects the beverages]. It approaches the table, measures it with its cameras […], and then positions these
four cups on the table and says: 'Have a drink.Drinking is healthy'.(Care manager, IC04 #00:06:28-6#) From this description, the robot appears to have all the capabilities to take over this task.In fact, however, the opposite is true.The robot that had been on trial for almost one and a half years in this care facility at the time of the interview was unable to fetch the beverages in the kitchen by itself.It also had limited capabilities for dealing with deviations from its implemented course of action (such as when the table is moved from its predefined position or when people are standing in its way).Therefore, it remained necessary for a nurse to accompany and support the robot during the task of beverage serving with the consequence that 'in that time, I would have brought three times as many beverages to the residents myself.It's really not relieving yet' (Care manager, IC04 #00:13:12−3#).
The care robot projects in our empirical sample and most of the cases we know from the literature are similar in that the complexities of the work situations demand too much of the care robots, which consequently do not prove to be of much help in the test applications and trial runs.Nevertheless, we found in our empirical research mostly positive views of the participants in care robot projects, not only from the engineers but from the care professionals as well.We believe that the following three mechanisms play a part in reinterpreting care robot application scenarios as promising that should rather be viewed as doubtful, judged from the evidence the projects' participants actually gather.
The first mechanism refers to the rhetorical persuasiveness of prototype scenarios.Prototype scenarios, in our definition, are more or less comprehensive physical implementations of the components that make up the application scenario within more or less realistic laboratory settings where the technical prototype represents the main technological components of the innovation and the testbed within which the prototype's usefulness is demonstrated represents the situation of application (Schulz-Schaeffer & Meister, 2017: 198, 201;Schulz-Schaeffer & Meister, 2019: 40, 44).Prototype scenarios create the impression that they represent a solution that already exists and works.While this is true to some extent, it often remains unacknowledged by observers that this is only the case within the controlled environment of the laboratory setting, while the corresponding real-world applications may still be far away.Creating the impression of an already existing solution constitutes the rhetorical persuasiveness of prototype scenarios (Schulz-Schaeffer & Meister, 2017: 211−212).
The care professionals who participate in care robot development projects usually get to know prototypes of the robots to be developed ahead of the project start, either through on-site presentations or, more often, through demonstration videos on the homepages of the robot developers or on YouTube.Mostly, these are demonstrations of prototype scenarios.Statements from our interviewees, such as the following, confirm the persuasive power of these demonstrations.
There are videos on YouTube, for example, in which [the robot] mixes a drink or picks up an object from the floor, that kind of thing.[…] In retrospect, I would say that we were perhaps a bit naïve in our approach, in that we simply assumed that this was a robot that could already do a lot of things.And then it turned out that this was not yet the case.That it can't take over things in care as easily as we had imagined.(Nursing scientist, IC07 #00:17:13-2#) When the participating care professionals realise how far from a real-world application these demonstrations actually are, they are already engaged in and committed to the development projects.Thus, they are more likely to react by lowering their expectations than withdrawing their support.In all of our empirical data, the application scenarios that had been built based on such lower expectations were of little help to the care workers, rather to the contrary.
In this situation, we observed two additional mechanisms that serve to deal with the discrepancies between the promises of care robotics as represented by the prototype scenarios and the reality of the project work.The first of these, which we refer to as the logic of the 'not yet' , interprets the ongoing projects as the groundwork to be done first before the really promising applications will come into reach.It becomes acceptable to the participants that the current application scenarios are unsatisfying because they are considered necessary as steps leading to a promising future.The following statement refers to an application scenario in which the robot is supposed to lift and hold the injured leg or arm of the patient while the nurse tends to the wound.The interviewee informs us that it is the holding, rather than the lifting, that is demanding for nurses when performing this task, but that there exists supporting material for this purpose.Nevertheless, he argues in favour of this application scenario as follows: But you have to start with something, and that's where we decided on, for example, leg or arm lifting as basic things.That is, we break the whole thing into small pieces, into small parts, and start with a small part until that works properly and then build up further.When you're in the process, it's a completely logical process.When I think back to the expectations I had in the beginning, to what we can do now that's a downward movement […] Because the basics really have to be taken care of first.In that sense, I understand it, I also agree.(Nurse, IC11 #00:34:38-0#) The third mechanism we identified that helps to keep faith in care robots is abstraction from real world conditions.It is based on a strategy that one of our interviewees summarises as follows: 'this would actually be part of the use case but is currently not quite considered by us' (Engineer, IC08 #00:18:26-3#).In every care robot project, we investigated or found in the literature, there are several components or features that are deliberately not developed though they are part of the application scenario without which there is no working real-world application.The just-mentioned robot for lifting and holding injured legs and arms, for instance, was developed without the capability to avoid touching the wounded parts, which is a very obvious precondition for its actual use.The beverage-serving robot mentioned earlier lacks the capability to fetch the beverages on its own; without this ability, it is of no help.Another robot from our sample was supposed to support wound documentation by taking pictures, but it is too big for the available space at the hospital bed.
In all of these cases, it is almost a matter of design that the care robots are not able to keep up with the demands of their real-world use.Usually, the decision not to develop important components and features is interpreted and communicated within the projects as a 'not yet' , and often as a 'not yet' that exceeds the duration of the project.These decisions and their interpretation thus reinforce the logic of the 'not yet' .But there is more to them.They also play a part in redefining the situation of the care professionals and care facilities participating in the development projects as a research setting rather than as a situation of application.As one of our interviewees put it: The abstractions from real-world conditions are making it obvious to the participants that they are situated in a laboratory setting rather than a real-world setting when participating in these projects, even when the robots are already being trialled at the care facilities.

Discussion and conclusion
Care robotics is not the only new technology that has over decades maintained its status as a promising technology even though the related technology development has not kept up with the promises made.Perhaps the most prominent other example is Artificial Intelligence (AI) for which the same applied between the 1950s and the 1990s.The self-perpetuation of the promise of AI can be explained by what Andreas Breiter (1995) has described as 'hourglass-shaped development dynamics' .According to Breiter, AI development in this period was characterised by a succession of scientific approaches (such as semantic information processing, knowledge-based systems, or artificial neural networks) each of which was predominant for some years before being superseded by another approach.The promise of AI was kept alive because when one approach struggled with keeping it, the next promising approach was already ready to take over.With care robotics, however, there is no similar succession of approaches that might explain the self-perpetuation of the technology's promise.Rather, as we have argued in this paper, it is a powerful narrative in combination with several projectrelated sensemaking mechanisms that helps to explain how its promoters succeed in keeping the promise of care robots alive.
To understand how the narrative and the project-related sensemaking mechanisms create the discursive reality of care robots as a promising technology, we have focused on their temporal structure.In agreement with the literature, we found that it is the temporal sequence of the plot of the nursing crisis narrative that creates the causal links between the events and thus reinforces the persuasiveness of the problem description (the increasing deficit of human labour in care work) and the solution (care robots).In addition, we can see that the way the nursing crisis narrative is employed makes it possible to deal with two contrasting threats to the credibility of the promise of care robots.It would be a threat to its credibility if care robots were expected to be of actual use only in the distant future but also if there was an expectation that the technology had already had the chance to prove its usefulness in the care sector (and had failed).As argued above, the way the narrative has, over the last two decades, been told as a story that unfolds in the present and leads to a solution in the near future (today's demographic change leads to the situation in which care robots are a natural solution) helps to cope with both credibility threats.
In a different but not unrelated way, the three project-related sensemaking mechanisms we identified are also about establishing a temporal order of events that helps to maintain the credibility of the promise of care robots.Here, the threat to its credibility results from the rather poor performance that current care robots still show.Their actual experiences with the robots' performance cause the participants of care robot projects to view the promise of care robots as addressing a more distant future than the one implied in the narrative.But, at the same time, the three discursive mechanisms allow them to cling to the promise despite their experiences, for several reasons.First, it is already observable today in prototype scenarios that care robots − at least in principle if not in practice − fulfil care-related service work tasks.Second, the 'not yet' implies a 'now' in the future with respect to which the current care robot projects make sense, even if their application scenarios do not.And, third, because it is obvious from the necessary abstractions from real-world conditions in the past and current care robot projects that these robots cannot already work as envisioned for the future.
Our findings suggest that the existence of a prevalent and very coherent narrative that fits with the sensemaking practices at the project level plays a major part in the selfperpetuation of the promise of care robots.For enhancing the scope of our findings, it might be promising to investigate the existence and role of similar narrative structures in related domains of technology development.To mention two examples: for the development of industrial cobots, our data point to the existence of a narrative about more flexibility of work processes but this narrative seems to be more complex and less coherent than the nursing crisis narrative.It might be interesting to examine the structuring role of this rather different type of narrative.In the field of ambient assisted living, we find evidence of a very coherent narrative about enabling seniors to stay independent and live in their own homes.It would be interesting to examine if and how this narrative influences the concrete application scenarios in a way similar to the nursing crisis narrative.

[
the robot].(Nursing scientist, IC02 #00:59:06-4#)Everything was based on the premise[…]  that nurses are very overburdened with tasks.So, the question was then in which tasks they could be relieved or where the greatest potential for relief is seen.Where can nurses be best supported?And the greatest need was identified with respect to transport and travel times[…],and preparation of any materials.[…] That's kind of what takes up the most time and at the same time also prevents the nurses from being in personal interactionwith the patients and building a relationship with the patients.So ultimately all activities that actually take them away from the patients.(Engineer, IC08 #00:12:42-8#)

I
think we have now reached the point where everyone realises that what we are doing is basic research.It is very, very far away from the market.That has to be said clearly.We are certainly 15 or 20 years away from the market with what we are currently doing.And after that was made clear, I think it's okay.(Engineer, IC12, #00:18:52-0#) assume you've watched those YouTube videos.There it [the robot] goes to the kitchen cabinet and almost gets the coffee cup out of there, escorts any residents who are leaning on his arm.It's all just not like that.[It's] like if you look at the Persil commercial.Even that just doesn't wash everything bright white and that's kind of how you feel there.(Care manager, IC04 #00:55:44-0#)