“What do you want for dinner?” – need anticipation and the design of proactive technologies for the home
This paper examines ‘the routine shop’ as part of a project that is exploring automation and autonomy in the Internet of Things. In particular we explicate the ‘work’ involved in anticipating need using an ethnomethodological analysis that makes visible the mundane, ‘seen but unnoticed’ methodologies that household members accountably employ to organise list construction and accomplish calculation on the shop floor. We discuss and reflect on the challenges members’ methodologies pose for proactive systems that seek to support domestic grocery shopping, including the challenges of sensing, learning and predicting, and gearing autonomous agents into social practice within the home.
Key wordsEthnomethodology Domestic grocery shopping Proactive technology Automation Autonomous agents
In 2014 Kari Kuutti and Liam Bannon laid out the case for a new paradigm in HCI research that went beyond producing and analysing snapshots of interaction focused on the individual and the human-machine dyadic relationship (Kuutti and Bannon 2014). They called the proposed shift ‘the turn to practice’ and located it in ‘a decades-long process’ that has seen the social sciences come to focus on practice as ‘a fundamental unit of analysis’. Drawing on Nicolini (2013), Kuutti and Bannon identify six main schools of practice theory: praxeology (Reckwitz 2002), practice as the house of the social (Schatzki 1996), practice as tradition and community (Lave and Wenger 1991), practice as activity (Leontyev 1978), practice as discourse (Foucault 1969), and practice as accomplishment (Garfinkel 1967). This paper is rooted in the latter tradition. It thus focuses on how people do need anticipation and, following Garfinkel (ibid.), provides an analytic explication of the ways in which need anticipation is methodologically organised or ordered in-the-doing. Our interest in understanding the practical accomplishment of need anticipation is motivated by the emergence of ‘proactive’ technologies, which incorporate elements of automation, autonomy, and agency and seek to respond to human need. Nascent forms can already be found in the home and include products such as the NEST ‘learning thermostat’1 and Amazon’s ‘Dash’.2 Proactive technologies hold great social and economic promise (McKinsey Global 2015). Key to their uptake is the prediction of anticipated need. A smart thermostat must, for example, predict when and at what temperature to operate based on an understanding of user needs. However, research has shown that this may well be problematic in practice. Yang and Newman (2013) show, for example, that the NEST thermostat’s predictive features do not work well for co-habitants with differing comfort levels, a not uncommon situation (Devlin 2017), and ignore the ways in which temperature is collaboratively negotiated as a social practice.
The domestic environment is an important potential site for a broad range of ‘smart’ proactive technologies, yet the focus in HCI on ‘computational intelligence’ has typically focused on the efficacy of algorithmic prediction (e.g., Scott et al. 2011) and user acceptance (e.g., Forlizzi and DiSalvo 2006). In accordance with the paradigmatic shift elaborated by Kuutti and Bannon, we take a different tack and seek to understand the social nature of need anticipation as elaborated in practice, thus moving beyond a focus on the individual and human-machine interaction to study collaborative activity and the concerted use of artefacts within mundane organisational routines. We thus seek to understand how people, rather than machines, anticipate need in order that we might learn something of the challenges that confront proactive systems, especially in ‘multi-user’ settings.
We focus on domestic grocery shopping as it provides a ‘perspicuous setting’ (Crabtree et al. 2012), insofar as grocery shopping is essentially concerned with anticipating what is needed over some prospective period ahead, be it a few days, a week, etc. We are not the first to look at shopping practices (e.g., Brown 2004; Elms et al. 2016) or food consumption lifecycles (e.g., Ng et al. 2015), but we may well be the first to do so with an eye towards understanding the challenges human need anticipation raises for automated prediction in this particular context, and what the development of computational agency might therefore learn from human agency at work. Accordingly, our study attends to how household members anticipate need as a local matter and thus come to furnish themselves with the domestic foodstuffs and the other household goods they require over some prospective period of time. As made visible in the doing we explicate the mundane ‘methods’ or ‘methodologies’ (Garfinkel 1967) members accountably employ to anticipate need. These methodologies revolve around list construction in the home and calculation on the shop floor. They implicate a ‘seen but unnoticed’ ecology of anticipation and a distributed body of collaborative work and reasoning in their practical accomplishment. They also identify key challenges for the development of proactive technologies that seek to support domestic shopping practices. These include challenges for the physical sensing of products in the home, computational learning and the prediction of need, and the need for proactive systems to be designed in such a way so as to gear in with the social milieu, which drives the prospective anticipation of need.
2 Studying need anticipation
Participants in the study
Is in his fifties and lives with his wife and two children aged seven months and six years.
Is in his fifties and lives with his wife and two daughters aged 13 and 16.
Is in her fifties and lives with her husband and 22-year-old daughter.
Is in his fifties and lives with his wife.
Is in his twenties and lives with his partner.
Is in her twenties and lives with her partner.
Is in her thirties and lives with her partner.
Is a gentleman in his eighties living alone.
The study was approved by the University’s ethics committee, and informed consent was gained from all participants contributing to the research (not just the lead contact) prior to fieldwork. Studying need anticipation entailed undertaking direct observations and engaging in informal discussions to elaborate the witnessed details of shopping’s work. Observations and discussions were recorded on audio and video, producing 33 h of data, complemented with field notes and photographs.
Data was gathered during fieldwork sessions lasting between 2 and 3 h with each household and included visits to supermarkets (6 out of 8), local independent shops including butchers and bakers (1 out of 8), and online shops (2 out of 8). The specific duration of each study varied according to the particular ways in which each household anticipated need and spanned a range of activities implicated in shopping such as planning and preparation, finding specific products that will meet need online or on the shop floor, unpacking goods and placing them for use in the home, and cooking. Sequences of data were subsequently transcribed, blending audio and visual recordings to create rich descriptions of the naturally occurring and naturally accountable activities implicated in shopping’s accomplishment. These were then analysed to identify their socially organised (i.e., methodological) features, particularly those implicated in anticipating need.
The ‘in vivo’ work (Garfinkel 2002) of observation and discussion with participants was essential to understanding the intersubjective organisation of shopping’s work, and subsequently enabled the appreciation of an array of practical matters and practical methodologies involved in and reflexively organising need anticipation. We identified the rhythms and routines involved in shopping by employing the ‘horizontal and vertical slicing’ technique (Crabtree et al. 2012). This allowed us to map out the local order of shopping’s work in each household, which became apparent through our observations of shopping activities as they were taking place and when spoken about in discussion. The mapping exercise sensitised us to the details of shopping’s work and helped us tease out the ‘animal in the foliage’ (Garfinkel et al. 1981). That is, the methodological ways in which need was and is anticipated in the doing of domestic shopping across participating households in the face of the ‘disparate concerns’ (Crabtree and Rodden 2004) or the particular needs that marked out each home as unique.
“... in the inter-subjective organisation of human activities and put us onto real world topics that are relevant to the future technologies we wish to develop.”
Our analysis of the methodological ways in which domestic shopping is organised was particularly attentive to the life cycle of items from ‘shelf-to-shelf’. This included mapping out when items were used in the home, disposed of, arranged to be bought, found in a store, chosen, paid for, unpacked and placed back on a shelf (or fridge, cupboard, etc.). The temporal order of this cycle invited a closer inspection of how need was anticipated in the course of everyday life, in which products are embedded and used. This led to a focus upon two grossly observable features of need anticipation: shopping list construction in the home, and calculation on the shop floor. This is not to say that all a proactive system needs to do to support domestic shopping is calculate some variables and construct a list. As we shall see, while extremely mundane, list construction and calculation in domestic grocery shopping are complex methodological matters that raise significant challenges to automated need anticipation. They should be seen at this point then as real-world topics that provide us with ‘coat hangers’ (Garfinkel et al. 1988, unpublished manuscript) off which to hang our analysis and unpack our findings.
3 Anticipating need through list construction
Our participants used two types of lists: physical and digital. David, Charles, Tina, Mark and Harry used physical (paper) lists. Liana and Brenda used digital lists, including lists on smartphones and lists provided by online grocery stores. George used neither. Given their predominance in our study, we first treat the assembly of physical lists, which were compiled over an extended period of time before shopping (e.g., over the week) and completed just prior to the shop or, alternatively, constructed ‘here and now’ on the day of the shop itself as needed.
It is notable too that physical lists were situated amongst the arrangement of cupboard space, shelves, and work surfaces to provide for ready access and use. Thus, physical lists were placed in sites where a high throughput of goods takes place, typically in or near to the kitchen ‘work triangle’ (Lange 2012) where food items in particular flow through the home at speed, which in turn occasions frequent entries being made on the list. This traffic shapes the very placement of lists and is accompanied by the assemblage of resources: paper, pads, pens, and vouchers, which are placed to enable and to reflect the potential anticipation of need. Physical lists are situated then within a physical ecology of anticipation.
Our studies also revealed that physical lists were constructed on the fly, and unique in that no two lists in the same home were the same week after week. There may be recurrent items on a list – e.g., milk, bread, cereal, etc. – but every list is different. It was also noticeable that entries on lists are ‘specifically vague’, which is to say that items are not described in detail, but only in terms sufficient for the list writer to identify the items required. Thus, and for example, someone might write ‘beans’. To the outsider what is needed may be opaque, but to the list writer this clearly specifies the kind, brand and quantity of beans needed. The clarity of the matter is provided by the list writer and reader’s membership competence. As a member of this home they thus know what ‘beans’ means here.
Lists were at one and the same time recognisable to us as outsiders and yet largely unintelligible when it came to specifying just what was anticipated. Item entries and annotations are evidently ‘indexical’ (Garfinkel 1967) to, and thus get their definite sense from, the local ecology of anticipation and the cohort that produced them. Lists are thus ‘procedurally encounterable’ (Wieder 1999) and acquire their sense from the practical activities and reasoning that provide for their construction. We find, for example, that annotations support the accomplishment of shopping, marking out where items are located in a store and the order of shopping. It is towards unpacking the indexicality and procedural encounterability of list construction that we turn next, focusing particularly on how lists are both incidentally and intentionally assembled and how need is mundanely anticipated in the course of these accomplishments.
3.1 The incidental anticipation of need
Here, the addition of an item to the list is occasioned in preparing a meal. When getting the thyme out from the cupboard, just at the point when the process of cooking occasions its addition, Tina notices that it is running low. This triggers a series of actions that acknowledge and register prospective need, before returning to cooking the bolognaise sauce. As the example makes clear, need is contingently anticipated in the doing of mundane domestic activities, often at the point an item is taken from a shelf or cupboard or when it is disposed of, and turns upon seeing and noticing the numbers, quantities and/or amounts of remaining items. It is also the case that the recurrent nature of many mundane activities in the home (e.g. making tea or coffee or toast, etc.) provides ongoingly for incidental monitoring and drives a local life cycle of need anticipation.
Tina: Ah we’re running out of thyme (shakes thyme pot, picks up pen and adds it to the list on the fridge)
Where the use of items is bound to the undertaking of particular activities, in this case the changing of the baby’s nappy, the practical task of making sure that those items are restocked is also routinely carried out by the individual who largely uses them. Anticipating incidental need is, then, bound to specific individuals who are responsible for doing certain ‘jobs’.
David: Well my wife uses them
David: predominantly, so she’s the one that notices when they’re empty.
In addition to demonstrating that mutual monitoring is not fool proof, the example also demonstrates that it turns upon agreed measures of what constitutes ‘running low’.
David: We usually keep spare toothpaste in stock but my wife forgot to put it on the list so this time I bought three.
In addition to being differentially distributed then, the local life cycle of need anticipation is also managed for some members of the home, and is organised with respect to the local social and moral order. Thus, and for example, managing his six-year old daughter’s intake of biscuits and what it is appropriate for her to have is, for David, an accountable feature of parenting, made visible through the work of assembling the list.
Fieldworker: Custard creams (reading from the list)?
David: Now I don't usually buy custard creams but my daughter [aged 6], er, her friend, when she goes to her house, they always have custard creams.
Fieldworker: OK right.
David: So she said Daddy can we have some of those custardy biscuits and I said yeah course we can. We usually buy something else for her, so she’ll get those instead.
3.2 The intentional anticipation of need
Consulting household members as to what is needed routinely occurred as the doing of the household shop drew closer, e.g., the night before, the morning of, or just before the shop. As the above example makes visible, the intentional anticipation of need is routinely articulated through proposing candidate items (e.g., bagels) whose need status is determined through practices of looking-and-checking (e.g., looking in the bread bin, where it is found that what is actually needed is bread).
Charles: What kind of things do we need so I can get a sense of what we are out of?
Julie: OK, Bread.
Julie: There’s bagels there (in the bread bin) Fran.
Charles: (Gets up to check bread bin) We need bread.
Julie: Coffee maybe. Actually, you don't have coffee do you
Charles: No? Well, actually I need coffee, thanks for that.
Julie: because Sarah Johnson I think had the last of it yesterday when she came round.
Charles: Oh did she now.
Clarification may be required from specific members as above as to just what is wanted, and also to what has been previously written on the list (e.g., when an illegible or unfamiliar entry has been made). Discussion of food categories may also occasion members being called to account for not eating certain items and wanting other items instead (as again can be seen above).
Charles: What cereal?
Julie: Girls what cereal?
Charles: Not the chocolate ones.
Julie: Fran have you stopped eating malt wheats?
Fran: They’ve done something to the malt wheats, I don't know what they done but they’ve done something.
Fieldworker: They’ve changed have they?
Fran: They’ve done something and its not me being picky, they’ve done something.
Julie: Right so you don't want malt wheats. What do you want?
Fran: Rice Krispies.
As Amy and Mark make visible, planning with reference to established routines turns on building variability into them. After all, most people do not want to eat the same things day after day.
Amy: Do you want umm four days worth of roasted veg and rice, or shall we do like two, three days and maybe do something else again?
Mark: Umm, well is there anything in there that you could have for lunch the next day?
Amy: No. Not really.
Mark: Do you want to do pasta instead one day?
In the course of planning, the implications of persons availability on what might take place and when is taken into explicit account, and shapes what goes on the list ‘here and now’. Wrapped up in this are considerations as to just who making a shared meal might fall to, with an individual’s competences driving just what goes on the list. Charles, for example, is not as accomplished a cook as Fran’s mum, which results in them discussing ‘easy’ meals to make on Friday. Just who has to provide for the routine’s accomplishment (e.g., eating a shared dinner) may even occasion alternative options (such as a takeaway). It is also the case that guests and visitors are factored into determinations of what is needed when considering the impact of schedules on the intentional anticipation of need.
Charles: Mum’s away Friday Fran, what do you want for dinner then?
Notwithstanding established routines and preferences, there is a strongly negotiated character to planning and need anticipation, which turns upon manifold variations and contingencies including item turnover, the weather, the season, who is available, when, who carries responsibility for cooking, and who else may be in attendance, all of which drives just what is put on the list ‘here and now’. One final matter is also key to the intentional anticipation of need. It was observable that list construction was also done with an eye to cost, with the candidacy of items being established as commensurate with what could be afforded. It was the case for some of our participants that the party who paid for the shopping had a direct impact on the anticipation of need. Thus we find that while people may co-habit, they may have independent financial arrangements and different levels of income to the effect that one party does not attend especially to cost whereas the other does. This may result in fewer, or smaller, or lesser amounts of things being added to the list, or them not being added at all. The anticipation of need is, then, not only wrapped up in the economics of the household but also in the economics of the interpersonal relationships within it.
3.3 Anticipating need online
As noted above, not all of our participants constructed physical lists in the course of need anticipation. Brenda and Liana used digital lists, including smart phones (typically for a handful of items required off the cuff) and those provided by online stores. Online shopping is possessed of many of the methodological features found in anticipating need during physical list construction. Thus we found participants looking and checking, consulting one another, planning with reference to preferences, routines, schedules and scheduled events (e.g., birthdays, parties, and guests), and cost. However, in the digital world the methodological construction of the list reflexively provides for the completion of the shop, i.e., in anticipating what is needed online participants are not only assembling a list but actually doing the shopping.
Liana’s anticipation of the need for coconut milk is occasioned by her looking at her ‘favourites’ list, i.e., a list of previously bought items provided by a digital shop. As with so many other items, the potential need for coconut milk was not encountered incidentally, but intentionally in checking the favourites list, which sometimes prompts looking and checking elsewhere in the home.
Liana: See like coconut milk, I don't know if we’ve got some and like, sometimes I just get it anyway and sometimes I get up and check.
As Brenda and Frank make visible, shop furnished categories or self-formulated ones (e.g., ‘Christmas things’) are leveraged to surf and browse digital shops to discover candidate items and find just which amongst them will satisfy their needs.
Brenda: Right, what else?
Frank: Just write Christmas things.
Brenda: (Types ‘Christmas things’ in search bar; the category Christmas is also furnished by the website).
Clearly there are various pathways whereby need is anticipated in the course of online shopping, but in each case the exact number, size, quantity, weight, amount or brand needed is determined at the interface in light of available choices on just this occasion of shopping. Choosing inevitably involves comparing and making selections from amongst multiple available items. Comparing and selecting turns upon striking a balance between a broad range of practical concerns to do with item selection – e.g., best deal, the look of an item, preferences, fit with meal plans, etc. – and the cost of the shop as a whole. Choosing items thus turns upon calculation, which is a complex matter that also plays out, and in richer ways, on the shop floor.
Liana: I don't know if I’ve got cheese actually (gets up from sofa and goes to the fridge, finds they have not got much cheese; searches for cheese deals and adds one to the online list).
4 Calculation on the shop floor
Calculation often turns, as it does for George, on ‘sensing’ items in order to choose the most suitable to meet his anticipated needs. The extract shows how sensing the specific physical conditions of the potatoes underpins this. Thus, choosing from a container of potatoes is a calculable matter demonstrated in seeking out potatoes that do not have blemishes. Of course, this particular kind of sensing is contextually bound to choosing potatoes and reveals what it is that people know about the items they are buying, e.g., that potatoes are not sensed in the same way as avocadoes, which are squeezed in various ways in order to determine their ripeness.
George: Get a few jacket potatoes, those ones look alright.
Fieldworker: What are you looking for here?
George: Dents. Blemishes. Anything. That's why I often pick my own because if they are in a pack you can’t do anything about it.
In this case the sensing of the bread is tied to its prospective use in making toast, along with Charles’ personal preferences on the matter. It was observable too how the prospective use of items was tied ‘here and now’ on the shop floor to the places our participants anticipated using them. Thus, and for example, tinned sardines in BBQ sauce were bought instead of sardines in oil as there is ‘less waste to deal with’ on the building site where one participant worked.
Charles: I want something that feels springy, and I look at the size of it - I hate thin bread because when you toast it, it turns really dry, you know?
In the above extract Harry takes more than a cursory glance at the peaches. A close reading of the label on the tin allows him to get a detailed sense of its contents, which is key to deciding whether ‘just this’ item in hand will meet anticipated need. This determination turns upon Harry’s knowledge of buying ‘this kind’ of product in ‘these (budget) shops’ and displays his anticipation of the features it may have that he may or may not want. It is a common feature of calculation on the shop floor that a label is appealed to in order to make a choice when a degree of specificity about an item’s features may not be possible to determine by other means. Thus, we saw labels being read to find out about contents, nutritional value, quantity, how long something takes to cook, and various other product features that turned upon, and served, a disparate range of preferences, interests and activities.
Harry: Usually those just say ‘in light syrup’, but then cinnamon, yuck. Perhaps that's why it’s cheap because not many people would buy that, with cinnamon. So with these shops, you’ve got to look beyond them a little bit.
Fieldworker: You’ve got to be careful?
Harry: Yes. In fact I do read labels a lot.
Cleary ‘the (use by) dates’ on products is a key calculation in anticipating need on the shop floor and determining just which items to buy. Cost labels were also routinely checked and were found on items themselves and on the shelves. These were used in comparing prices, which included those currently on display and those previously seen, as described by George while heading towards the cakes.
Fieldworker: So you check the dates on each one.
Amy: I do. You don't, so much (looking at Mark).
Mark: I normally do because I get a bollocking if I don't.
Checking the price allows George to calculate whether or not ‘here and now’ is a good time to buy cake. The calculation turns upon his local knowledge of costs, which is drawn upon to make sure he buys items when the price is low. George, like many of our participants, routinely kept a stock of certain items in order to prevent him from being forced to buy when the price is higher. Thus, choice is often calculated based upon shop specific cycles of costs, sales and offers. Our participants also considered the prospective costs of the same or similar items in other shops being visited that day while at the shelf face.
George: Cakes will be the next thing. We don't eat a lot of cake. Sally might have some. Again, I’ll check. If the price is good I’ll have them, if not I’ll wait another week. It’s not a problem.
Here Tina’s anticipation of need is calculated upon seeing tubs of Flora (a butter-like spread) on sale, a discovery that makes her ‘excited’ as saving money is not an inconsequential matter. We found that many of our participants kept their ‘eyes peeled’ for bargains as they went about shopping, and certain sections in certain stores would be routinely visited in order to see ‘what they had on offer’, all of which means that our participants routinely bought more items than were on the list. We found too that where items were serendipitously discovered but had not been tried before, then our participants employed a distinct calculative strategy as seen here when Harry notices a stack of tinned tomatoes on sale.
Tina: Ah, Flora. This is where I get excited. You see (picks up two large tubs and puts them in the basket).
Discoveries of unknown items, as in the case above, would often involve buying a small amount of the item to try out. This enabled participants to manage the risk that accompanies spending money on items whose ability to meet anticipated need has yet to be determined. Thus methods for anticipating need are routinely bound up with contingency plans to make sure that risk is minimised.
Harry: Well I haven’t heard of this brand before, but they’re very cheap.
Fieldworker: Yes, that is very cheap isn’t it.
Harry: So I might buy two and try ‘em, see if they’re any good. It says that it’s just tomato so that's more or less the same – salt – more or less the same as the usual ones. But its an unknown brand to me. But at twenty-five pence I can’t go wrong.
Fieldworker: You would usually buy four?
Harry: Yes I would usually buy packs, but these are twenty-five pence in the singular (puts two in basket).
This extract makes it perspicuous how items are bought, or not, with regard to how they may be received and treated by other household members, in this case by Harriet one of Charles’ teenage daughters. All of our participants routinely undertook such calculations, which are bound to the social and moral order of domestic life and the potential impact of particular purchases on that order. Calculation is also done with respect to the preferences of others, as seen below when Mark finds himself with a different dessert to Amy who has chosen a more ‘chocolaty’ one.
Charles: Now I am in a quandary (looking at yogurt) because I can’t remember what they eat.
Fieldworker: Do they change the type of yogurt they want often?
Charles: Well Harriet (daughter) demands these ones, and she thinks it’s hers and won’t share.
Charles: And if she were here looking at it with me now I would say no you’re not getting this one to yourself.
In this it can be seen that the prospective consumption of food is a calculative matter that shapes the anticipation of need ‘here and now’ on the shop floor not only in the face of manifold product choice but also in the face of the choices made by other members of the home. It is not only the qualities or properties of ‘things on shelves’ that matter then, their prospective relationship to people and the events they engage in is important too.
Mark: Happy with that?
Amy: Yeah, yeah.
Mark: I feel a bit left out though. I feel as though I should probably get the same, so I don't get food envy (puts dessert back on the shelf and picks up the same one as Amy).
To this we would add that the ecology of the home is a key ingredient in calculating just what will meet anticipated need. Choosing just which items to buy in the face of manifold choice is inevitably bound to what will fit in the home. It was routinely the case then that our participants brought ecological considerations to bear when choosing items, particularly with respect to the size or the amount of items required to meet anticipated need. Thus, calculation was done ‘here and now’ on the shop floor with respect to the unique design and layout of space in each of our participants homes, for when they return home with their purchases each has to be situated in its place within the ecology of anticipation, e.g., in a cupboard, drawer or on a shelf, ready for retrieval and use.
5 So what?
“The motivation … is that ‘practice’ is seen as a means for providing social theorizing with an ‘ontological’ (or transcendental) foundation, i.e., a foundation prior to building a framework bottom up from actual empirical studies. So far, without luck, for the concept of ‘practice’ is notoriously unfit for doing that kind of work. As a result, the term ‘practice’ is being used in confusing and confused ways: as another word for ‘activity’, ‘culture’, ‘tradition’, ‘paradigm’, ‘embodied action’, ‘knowing how’, and so forth.” (ibid.)
“ … the Modern concept of ‘practice’ … focus[es] on the ways in which the competent actor in his or her action … [takes] particular conditions into account while committed to and guided by … general principles [plans, rules, procedures, etc.]. When studying a practice we are [then] focusing on how … practitioners determine the nature of a situation, how they select effective and efficient techniques (materials, implements, bodily postures, methods, etc.), determine deviations from what has been assumed in … rule formulations, deal with routine troubles, recover from breakdowns, etc. (ibid., our emphasis)
They are, but it is not that ethno-methods, or members’ methodologies, ‘make practices possible’; it’s not as if there are ethno-methods on the one hand, and practices on the other, with the former having a causal relationship to the latter. Rather, it is that ‘ethno-methods’ and ‘practices’ are identical. These are but different names for the same naturally occurring and naturally accountable phenomenon, ‘born in obscurity’ as Schmidt (2014) puts it or ‘seen but unnoticed’ in ethnomethodological parlance and left unexamined by social theory (Hutchinson et al. 2008). The notion of ‘ethno-methods’ was coined by an ‘incommensurable, asymmetrically alternate technology of social analysis’ (Garfinkel and Wieder 1992), which is fundamentally incompatible with a technology of social analysis that proceeds by theorising practice and thus abstracting from what it is that people actually do in order to build (typically explanatory) frameworks from actual empirical studies (Lynch 1997). This alternate technology of social analysis (Garfinkel 1988) was subsequently taken up in CSCW, where it came to be referred to as ‘studies of work’ or ‘work practice’, a concept as Button and Harper (1996) point out,
“For ethnomethodology practices are locally produced by using a selection of certain ‘ethno-methods’ of interaction that make practices, and thus all social activity, possible. These ethno-methods are the central object of ethnomethodological studies.”
“ … that needs to be grounded in analytic explication of work rather than in theoretical abstractions … … …. capable of revealing the methods through which the situated contingencies of activities and interactions are contextually managed as orderly courses of work. (our emphasis)
The turn towards proactive systems has given us motive and we find, when we look, that something as commonplace and ‘banal’ as constructing a shopping list and handling the manifold contingencies of choice on the shop floor in anticipating need has a rich ‘methodic’ or methodological character. We can see in the first instance that physical lists and resources for their construction are situated in an ecology of anticipation, amongst high density traffic in the flow of fast moving consumer goods. This is no accident, but a methodological feature of list construction found across the homes participating in our study, which enables need to be anticipated incidentally on the fly and to be intentionally provided for as a preface to or during the actual shop. Both incidental and intentional need anticipation are shot through with methodological characteristics, as is calculation on the shop floor, and we summarise each in turn as a preface to considering the implications of these methodological matters for the design of proactive technologies to support domestic shopping.
“ … practices of such unquestioned efficacy and banality that no motive ordinarily exists, either in commonplace settings or professional inquiries to make an issue of their methodic character.”
5.1 The methodological character of incidental need anticipation
Our studies make it perspicuous that need is often anticipated incidentally in the course of doing routine activities in the home, within which items or products are used. It is, then, a methodological feature of incidental anticipation that need is contingently encountered in the making of meals, bathing, cleaning, washing, etc., as items run low or run out. It is not simply that items run low or run out that matters, however. It is that in being encountered the prospective need to replace them is anticipated. The contingent anticipation of need is not a ‘one shot deal’ either, something simply encountered in the moment of seeing that items are running low or running out, but is methodologically provided for as an ongoing matter through incidental monitoring. Thus members actively ‘look out’ for items running out or running low over the course of their use. A further methodological feature of incidental monitoring is that it is socially distributed amongst household members and bound to the undertaking of particular activities by particular members. In some cases, with respect to particular categories of product, especially those that members do not want to run out of, incidental monitoring is also bound to multiple members and turns, methodologically, on agreed measures as to what constitutes ‘running low’. It is a methodological feature of need anticipation too that the entitlement to add items to a list is differentially distributed, with young members of the home being excluded from direct item entry (and this would also appear to extend to visitors and guests). Thus the anticipation of need is, for certain members of the home, a managed matter, though we note younger members have their own methods for getting items onto the list.
5.2 The methodological character of intentional need anticipation
It is perspicuous too that need is also methodologically anticipated in intentional ways as a preface to the actual occasion of shopping. Thus members are consulted to identify items required on ‘just this’ occasion of shopping. Consultation turns, as a matter of method, on members making what is needed accountable to the person who will add items to the list, here and now, and is done methodologically through members a) proposing candidate items and b) through looking-and-checking. Looking-and-checking is socially distributed and bound to individual responsibilities and personal needs; it is spatially distributed and requires as a matter of method that members search the home to anticipate need, including checking cohorts of products (e.g., bathroom items); and it is temporally distributed spanning both incidental and intentional need anticipation. Methodologically, the job of consultation also provides for the clarification of anticipated need, including previously inscribed articulations of need and the candidate status of prospective articulations. Need is also anticipated through planning, which is possessed of various methodological characteristics. Thus, need is planned for with reference to established preferences (individual and collective), cycles of item usage, and seasonal cycles where the need for specific items changes periodically. Need is also planned for with reference to established routines and with respect to the need to build variety into them; with explicit reference to sharing, particularly with respect to food items and specific occasions of eating together during the prospective period ahead (including all manner of social events; and with explicit reference to members schedules to determine their availability, not only in terms of who may be present at a shared event but who the job of making food might fall to. The availability of members implicates their culinary competence, and provides methodological grounds upon which to anticipate what might therefore be needed. It is also the case that the anticipation of need is as a matter of method planned for with reference to cost, which is not reducible simply to money but to who is paying to satisfy need on ‘just this’ occasion.
5.3 The methodological character of anticipation on the shop floor
Need anticipation does not stop with the completion of the list. This may close the matter ‘here and now’ on the occasion preparatory to shopping, but it reopens on the shop floor in the face of the manifold contingencies occasioned by choice. Thus, need anticipation continues on the shop floor and is methodologically organised through situated practices of calculation. These practices include sensing items through look, touch and/or smell (just which sense or senses apply turns on the kind of item being sensed), and occasionally doing so in collaboration with shop staff, to determine the suitability, condition and quality of candidate items and to make particular selections from amongst them. Product labels and information posted on shelves are also drawn on methodically to determine the suitability, condition and quality of candidate items, with members drawing contingently upon various categories of information (nutritional value, sell by dates, prices, offers, etc.) to make particular selections from amongst the range of choice. Determining an item’s suitability is not only concerned with various categories of product-related information, but methodologically tied to the shopper’s retrospective and prospective knowledge of cost cycles and projected reductions for particular kinds of items in and across particular stores, which is to say that shoppers know what constitutes a ‘good deal’ and keep an eye open for their (re)occurrence. As a matter of method shoppers also keep their ‘eyes peeled’ for serendipitous discoveries by scanning shelves for offers or reduced-price labels on items. It is notable too that novel serendipitous discoveries are as a rule purchased in small quantities to in order to determine their suitability. Calculation is also methodologically bound to the prospective use of items and choices are thus made with respect to the occasion of product use, the choices made by those they are shopping with, the recipients they are buying products for, and the social implications of making particular choices (e.g., the negative impact of certain choices on certain members behaviour and the need to avoid such situations). Finally we note that need anticipation is done methodically with respect to the available storage space in the home.
6 Implications for design
Our studies make it perspicuous that anticipating need even in the case of such a ‘banal’ commonplace activity as domestic shopping is a complicated business whose accomplishment turns on and is reflexively organised through a rich assemblage of members methods. We make no claim to have uncovered or described all this assemblage might consist of, only that it exists, is there for anyone in ‘western’ cultures at least to see even though it typically goes unnoticed, is taken for granted, and ignored. Nonetheless this methodological assemblage of practices is consequential as the anticipation of need turns upon its unquestioned efficacy and we see no reason why it won’t continue to do so as we turn to the proactive technologies and autonomous systems. Here we wish to consider the challenges the methodological organisation of need anticipation raises for the design of proactive systems that aim to support domestic shopping practices. In particular, we reflect on how proactive systems might be designed to fit in with the physical ecology of anticipation in which the incidental as well as the intentional anticipation of need is situated. This raises design challenges around sensing, and includes informational as well as physical properties. Furthermore, in effectively bringing the shop floor into the home, proactive systems will, as digital systems currently do, rub up against the range of calculative practices surfaced by our study that are bound up with need anticipation. This raises challenges around learning and prediction. Finally, the collaborative and social nature of need anticipation inevitably situates proactive systems in a division of labour, which we also discuss below.
6.1 Sensing in the physical ecology of anticipation
Making proactive systems fit in with the ecology of anticipation would appear to lend itself well to the application of Internet of Things (IoT) technologies. The potential ‘connectedness’ of the physical fabric of the home and products within it may support the fine-grained monitoring and looking-and-checking implicated in incidental and intentional need anticipation. A connected future in which ‘smart products’ are coupled with a local sensing infrastructure to monitor product lifecycles from cupboard shelf to waste bin would be a first key step towards proactive need anticipation (Cognizant 2015). Existing work has designed and studied sensor-based interventions at various stages of this lifecycle, many of which rely on self-reporting (see Comber et al. 2014 for an overview). For example, Ng et al. (2015) used wearable cameras to study the overall food consumption lifecycle, while Hupfeld and Rodden took a photo diary approach (Hupfeld and Rodden 2012). A significant body of work has focused on understanding food waste management and design to support its avoidance (Comber and Thieme 2013; Ganglbauer et al. 2013, 2015). In this space, interventions aimed to increase awareness and reduce waste have relied on cameras placed in rubbish bins (Thieme et al. 2012), and the fridge (Ganglbauer et al. 2012), and have also explored automated recognition through additional colour coding of items in the fridge (Farr-Wharton et al. 2014).
Other work draws on continuous monitoring of kitchen spaces. It includes research that has instrumented a kitchen worktop with weight scales and overhead cameras to estimate nutritional value using a Wizard-of-Oz approach (Chen et al. 2010), motion-triggered cameras over the stove (Clear et al. 2013), and measuring the contents of containers using a sonic sensor (Fan and Truong 2015). Amutha et al. (2012) demonstrated that combining load sensing with UHF RFID technology allows for a more detailed view of product usage. Their monitoring of item stock thresholds via the ‘smart kitchen cabinet’ has also demonstrated that automatic shopping lists can be crudely constructed. More generally, load sensing has been suggested as a robust way to gather ‘contextual information’ in everyday environments (Schmidt et al. 2002). However, while load sensing may be promising to monitor products and anticipate need, the challenges of implementing load sensing are not just technical and practical (i.e., fitting load sensors in the right place). Our study also instructs us that monitoring turns as a matter of method on locally agreed measures as to just what constitutes ‘running low’, which trigger the anticipation of need. Thus, it is not simply that items run low or run out that matters, what matters is how those measures are locally established and sustained. Sensing alone then, does not provide reliable measures to anticipate need, and interaction mechanisms are thus required to enable users to create and manage situationally relevant monitoring arrangements.
Insofar as the digital brings the shop floor into the home it will also be necessary to marry sensing/monitoring to product information and thus enable calculation in the face of manifold choice. The availability of detailed product information is an important part of need anticipation, as our study has shown for example in practices of appealing to and checking labels. In order to support the anticipation of need appropriately, digital alternatives need to make at least the same information available as physical labels do. Related work has explored the recognition of product information through OCR of scanned shopping receipts (Mankoff et al. 2002), and Reitberger et al. sought to exploit existing food databases to retrieve product information (Reitberger et al. 2014). However, the findings from Reitberger et al.’s study confirm our own investigations that truly open and comprehensive product databases are lacking. While some supermarket-specific open APIs exist in the UK, for example, data is generally kept in proprietary systems behind paywalls, making open innovation and development difficult, despite initiatives such as the Open Data Institutes’ Food Data challenge.3 This seems to us to create a particularly thorny barrier to the development of proactive systems in this context, for as our study makes it plain to see, it cannot be assumed that an item will be replaced with the same or even at all. Detailed information is required to enable effective need anticipation.
6.2 Learning and predicting calculative practice
Our study makes perspicuous the situated and contingent array of calculative practices through which choices are made, which inhibit simplistic ‘if-this-then-that’ style actuation. For example, that something was consumed last week, and even many weeks and months before that, does not mean that it will be required next week (or whatever the prospective period may be). Nonetheless, a designer may posit that many of the resources drawn on in calculation may usefully be drawn upon by proactive systems: preferred products, ingredients, nutrition, use by dates, cost, offers, storage and product size, stock levels, frequency of use, recipients, etc., once sensed and connected to a user model are all computable. It would appear then on the face of it that machine learning techniques such as preference learning (e.g., Fürnkranz and Eyke 2010), pattern recognition, and the prediction of use cycles presents itself as a promising solution to the problem of need anticipation in proactive systems. There are challenges of course, particularly gearing algorithms into fine-grained local patterns (Crabtree and Tolmie 2016) rather than large scale clusters of preference, including deliberate variability of routines, seasonal variation, and individual’s schedules, but these are not, in principle at least, technically insurmountable. Further limitations include extensive ‘bootstrapping’ and training of proactive systems to learn local preferences and to handle the shortcomings occasioned by decisions outside of the learned behaviours, which can lead to undesirable outcomes (e.g., Yang and Newman 2013).
However, the contingent nature of calculation explicated in our study suggests it may be unwise to attempt to automate decision making entirely. Instead, our findings indicate that in most cases we should design proactive systems so that users can inspect, influence, and take control of machine decisions on demand. Research has demonstrated interactive approaches to intelligent systems that make predictions and act on the user’s behalf - Costanza et al. (2014) present a washing machine agent that charges a battery based on time slots the user provides to the agent through a booking interface, for example, and Alan et al. (2016) present a Tariff Agent that helps users switch energy tariffs while offering user-definable levels of autonomy. This work shows how proactive system decision-making can be designed to involve the user at key moments.
6.3 Situating proactive technologies in the division of labour
While it may be possible to build ‘in situ forms of interaction’ into proactive systems to support monitoring, looking-and-checking, consultation, proposing candidates, clarification, and planning, it will not be possible to build in all the competencies involved in anticipating need. There is, then, need to build the human into the proactive loop.
“Most importantly (and problematically) for the project of designing autonomous systems, plans or any other form of prescriptive specification presuppose competencies and in situ forms of interaction that they can never fully specify.”
Thus we suggest that the efficacy of proactive systems will turn on their ability to assume an accountable role within the division of labour and social organisation of need anticipation. We are not the first to introduce the idea of roles into the design of proactive systems (e.g., Woolridge 2009). However, an important challenge here is to move beyond modelling the unique responsibilities, permissions, activities, and protocols required of computational agents to support need anticipation (Woolridge et al. 2000). It is also key that the users of proactive systems can configure the particular role proactive systems play in anticipating need in ‘this’ home. It is important then to enable users to assign specific responsibilities to proactive systems, be it with respect to anticipating need for particular activities for example, or monitoring specific products with respect to agreed upon measures, or constraining proactive need anticipation to the making of suggestions or proposal of candidates, whatever and whichever as users see fit. Key too then is the need to build interaction mechanisms into proactive systems that enable users to mesh the actions of such systems in with the distributed practices that provide for the efficacious anticipation of need (Schmidt and Bannon 1992). For even if the technological challenges of instrumenting the home with sensing, and learning and predicting calculative practice can be overcome, the essential problem of anticipating need on just ‘this’ occasion ‘here and now’, with all the contingencies that accompany it and the unspoken competencies involved, still have to be addressed and their implications resolved.4
Proactive technologies that sense, predict, and respond to changes in their environment are finding their way into everyday life. However, research has shown that they may fail to anticipate need appropriately in multi-user contexts. This paper presents the results of an ethnomethodological study of domestic grocery shopping, a commonplace activity implicating multiple parties that is essentially concerned with anticipating need, in a bid to understand key challenges confronting proactive systems in such settings. Our study reveals that the anticipation of need is situated in a distinct ecology of anticipation and articulated through an array of differentially distributed practices implicated in incidental and intentional need anticipation in the home, and calculation on the shop floor. If proactive systems are to anticipate need with a similar degree of efficacy as human beings - and that they perform efficaciously seems to us a key requirement, otherwise it is hard to see what need we would have of them - then they will need to fit into the ecology of anticipation and support the incidental and intentional anticipation of need.
Furthermore, in effectively bringing the shop floor into the home, proactive systems will need to support an array of calculative practices that need anticipation turns upon in the face of the manifold contingencies of choice. The tangible limitations of autonomous reasoning also make it important that proactive systems occupy an accountable place, as household members occupy an accountable place, in the social milieu that continuously drives need anticipation. Proactive systems must, then, become a part of the division of labour implicated in need anticipation too, rather than behave as agents operating in their own right. These issues raise challenges for physical sensing of products in the home, computational learning and the prediction of need, and the need for proactive systems to be designed in such a way so as to mesh with the social division of labour in the home. This includes tying sensing to locally agreed measures of monitoring and the product information required to determine just what is needed; gearing algorithms into fine-grained local patterns of preference and the variability of routines, seasonal variation, and individual’s schedules; and enabling the users of proactive systems to configure the role proactive systems play in anticipating need including the ability to inspect, influence, and take control of machine decisions on demand.
The issue was raised in discussion of this paper how proactive technologies like NEST or Amazon’s Dash “deal (or not) with the acquired understanding of anticipation”? The answer is they don’t, which is why they are experienced as problematic in practice, especially in multi-user settings as Yang and Newman (2013) unpack. It is also why we have undertaken this study, to try and understand what proactive systems need to get to grips with if they are to deal effectively with need anticipation in multi-user settings. In short, they will need to get to grips with and gear into the methodological ways in which members anticipate need in their respective application domains. What we have considered here is what that might amount to with respect to the design of proactive systems supporting domestic grocery shopping.
This work was supported by the Engineering and Physical Sciences Research Council [grant numbers EP/N014243/1, EP/M001636/1]. Data supporting this publication is not openly available as our ethics approval does not allow for the release of transcripts to third parties.
- Alan, Alper; Enrico Costanza; Sarvapali Ramchurn; Joel Fischer; Tom Rodden; and Nicholas Jennings (2016). Tariff agent: Interacting with a future smart energy system at home. ACM Transactions on Computer-Human Interaction, vol. 23, no. 4, August 2016, Article 25.Google Scholar
- Amutha, Pal, Karuppiah; Chidambaram Sethukkarasi; and Raja Pitchiah (2012). Smart kitchen cabinet for aware home. In SMART ‘12. Proceedings of the 1 st International Conference on Smart Systems, Devices and Technologies, 27 May – 1 June 2012. Wilmington: IARIA, pp. 9–14.Google Scholar
- Bittner, Egon (1965). The concept of organization. Social Research, vol. 32, Autumn 1965, pp. 239–255.Google Scholar
- Brown, Barry (2004). The Order of Service: The Practical Management Of Customer Interaction. Sociological Research Online, vol. 9, no. 4, pp. 1–22.Google Scholar
- Button, Graham; and Richard Harper (1996). The Relevance of ‘Work-practice’ for Design. Computer Supported Cooperative Work (CSCW), vol. 4, pp. 263–280.Google Scholar
- Chen, Jen-Hao; Peggy Pei-Yu Chi; Hao-Hua Chu; Cheryl Chia-Hui Chen; and Polly Huang (2010). A Smart Kitchen for nutrition-aware Cooking. IEEE Pervasive Computing, vol. 9, no. 4, October 2010, pp. 58–65.Google Scholar
- Clear, Adrian; Mike Hazas; Janine Morley; Adrian Friday; and Oliver Bates (2013). Domestic food and sustainable design: A Study of University Student Cooking and Its Impacts. In CHI ‘13. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 27 April - 2 May 2013. New York: ACM Press, pp. 2447–2456.Google Scholar
- Cognizant (2015). The Rise of the Smart Product Economy. https://www.cognizant.com/whitepapers/the-rise-of-the-smart-product-economy-codex1249.pdf. Accessed on 9 February 2018.
- Comber, Rob; and Anja Thieme (2013). Designing Beyond Habit: Opening Space for Improved Recycling and Food Waste Behaviors through Processes of Persuasion, Social Influence and Aversive Affect. Personal and Ubiquitous Computing, vol.17, no. 6, August 2013, pp. 1197–1210.Google Scholar
- Comber, Rob; Jaz Hee-Jeong Choi; Jettie Hoonhout; and Kenton O’Hara (2014). Designing for Human–Food Interaction: An Introduction to the Special Issue on ‘Food and Interaction Design’. International Journal of Human-Computer Studies, vol. 72, no. 2, February 2014, pp. 181–184.Google Scholar
- Costanza, Enrico; Joel Fischer; James Colley; Tom Rodden; Sarvapali Ramchurn; and Nicholas Jennings (2014). Doing the Laundry with Agents: A Field Trial of a Future Smart Energy System in the Home. In CHI ‘14. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 26 April – 1 May 2014. New York: ACM Press, pp. 813–822.Google Scholar
- Crabtree, Andy; and Tom Rodden (2004). Domestic Routines and Design for the Home. Computer Supported Cooperative Work (CSCW), vol. 13, no. 2, April 2004, pp. 191–220.Google Scholar
- Crabtree, Andy; and Peter Tolmie. (2016). A Day in the Life of Things in the Home. In CSCW ‘16. Proceedings of the 19 th ACM Conference on Computer Supported Cooperative Work and Social Computing, San Francisco, United States, 27 February – 2 march 2016. New York: ACM Press, pp. 1738–1750.Google Scholar
- Crabtree, Andy; Mark Rouncefield; and Peter Tolmie (2012). Doing Design Ethnography. London: Springer.Google Scholar
- Crabtree, Andy; Peter Tolmie; and Will Knight (2017). Repacking ‘privacy’ for a networked world. Computer Supported Cooperative Work (CSCW), vol. 26, nos. 4–6, May 2017, pp. 453–488.Google Scholar
- Devlin, Hannah (2017). Why Women Secretly Turn Up the Heating, The Guardian, 11 October 2017. https://www.theguardian.com/science/shortcuts/2017/oct/11/why-women-sewcretly-turn-up-the-heating. Accessed 9 February 2018.
- Elms, Jonathan; Ronan de Kervenoael; and Alan Hallsworth (2016). Internet or Store? An Ethnographic Study of Consumers’ Internet and Store-Based Grocery Shopping Practices. Journal of Retailing and Consumer Services, vol. 2, pp. 234–243.Google Scholar
- Fan, Mingming; and Khai Truong (2015). SoQr: Sonically Quantifying the Content Level Inside Containers. In UbiComp ‘15. Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing, Osaka, Japan, 7–11 September 2015. New York: ACM Press, pp. 3–14.Google Scholar
- Farr-Wharton, Geremy; Jaz Hee-jeong Choi; and Marcus Foth (2014). Technicolouring the Fridge: Reducing Food Waste through Uses of Colour-coding and Cameras. In MUM ‘14. Proceedings of the 13 th International Conference on Mobile and Ubiquitous Multimedia, Melbourne, Australia, 25–28 November 2014. New York: ACM press, pp. 48–57.Google Scholar
- Fischer, Joel; Andy Crabtree; Tom Rodden; James Colley; Enrico Costanza; Michael Jewell; and Sarvapali Ramchurn (2016). “Just whack it on until it gets hot” - Working with IoT Data in the Home. In CHI ‘16. Proceedings of the CHI Conference on Human Factors in Computing Systems, San Hose, United States, 7–12 May 2016. New York: ACM Press, pp. 5933–5944,Google Scholar
- Fischer, Joel; Andy Crabtree; James Colley; Tom Rodden; and Enrico Costanza (2017). Data Work: How Energy Advisors and Clients Make IoT Data Accountable. Computer Supported Cooperative Work (CSCW), vol. 26, nos. 4–6, June 2017, pp. 597–626.Google Scholar
- Forlizzi, Jodi; and Carl DiSalvo (2006). Service Robots in the Domestic Environment: A Study of the Roomba Vacuum in the Home. In HRI ‘06. Proceedings of the 1 st Annual Conference on Human-Robot Interaction, Utah, United States, 2–3 March 2006. New York: ACM Press, pp. 258–265.Google Scholar
- Foucault, Michel (1969). The Archaeology of Knowledge (trans. A. M. Sheridan smith). London: Routledge.Google Scholar
- Fürnkranz, Johannes; and Hüllermeier Eyke (2010). Preference Learning: An Introduction. In J. Fürnkranz; and E. Hüllermeier (eds.): Preference Learning, Heidelberg: Springer-Verlag, pp. 1–17.Google Scholar
- Ganglbauer, Eva; Geraldine Fitzpatrick; and Georg Molzer (2012). Creating Visibility: Understanding the Design Space for Food Waste. In MUM ‘12. Proceedings of the 11 th International Conference on Mobile and Ubiquitous Multimedia, Ulm, Germany, 4–6 December 2012. New York: ACM Press, Article 1.Google Scholar
- Ganglbauer, Eva; Geraldine Fitzpatrick; and Rob Comber (2013). Negotiating Food Waste: Using a practice Lens to Inform Design. ACM Transactions on Computer-Human Interaction, vol. 20, no. 2, May 2013, Article 11.Google Scholar
- Ganglbauer, Eva; Geraldine Fitzpatrick; and Florian Güldenpfennig. (2015). Why and what did we throw out? Probing on reflection through the food waste diary. In CHI ‘15. Proceedings of the 33 rd Annual ACM Conference on Human Factors in Computing Systems. Seoul, Republic of Korea, 12–23 April 2015, New York: ACM Press, pp. 1105–1114.Google Scholar
- Garfinkel, Harold (1967). Studies in Ethnomethodology. Englewood cliffs: Prentice-Hall.Google Scholar
- Garfinkel, Harold (1988). Evidence for Locally Produced, Naturally Accountable Phenomena of Order, Logic, Reason, Meaning, Method, etc. in and as of the Essential Quiddity of Immortal Ordinary Society, (I of IV): An Announcement of Studies. Sociological Theory, vol. 6, no. 1, Spring 1988, pp. 103–109.Google Scholar
- Garfinkel, Harold (2002). Ethnomethodology’s Program: Working Out Durkheim’s Aphorism. Lanham: Rowman & Littlefield.Google Scholar
- Garfinkel, Harold; and D. Lawrence Wieder (1992). Two Incommensurable, Asymmetrically Alternate Technologies of Social Analysis. In G. Watson; and S.M. Seiler (eds.) Text in context: Contributions to Ethnomethodology. London: Sage, pp. 175–206.Google Scholar
- Garfinkel, Harold; Michael Lynch; and Eric Livingston (1981). The Work of a Discovering Science Construed with Materials from the Optically Discovered Pulsar. Philosophy of the Social Sciences, vol. 11, no. 2, June 1981, pp. 131–158.Google Scholar
- Garfinkel, Harold; Eric Livingston; Michael Lynch; Douglas Macbeth; and Albert Robillard (1988). (unpub. manu.). Respecifying the Natural Sciences as Discovering Sciences of Practical Action, I & II: Doing so Ethnographically by Administering a Schedule of Contingencies in Discussions with Laboratory Scientists and by Hanging Around their Laboratories. Google Scholar
- Hughes, John; Val King; Tom Rodden; and Hans Andersen (1994). Moving Out of the Control Room. In CSCW ‘94. Proceedings of the 1994 ACM Conference on Computer Supported Cooperative Work, North Carolina, United States, 22–26 October 1994. New York: ACM Press, pp.429–439Google Scholar
- Hupfeld, Annika; and Tom Rodden. (2012). Laying the Table for HCI: Uncovering Ecologies of Domestic Food Consumption in CHI ‘12. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, Texas, United States, 5–10 May 2012. New York: ACM Press, pp. 119–128.Google Scholar
- Hutchinson, Phil; Rupert Read; and Wes Sharrock (2008). There Is No Such Thing as a Social Science. London: Routledge.Google Scholar
- Kuutti, Kari; and Liam Bannon. (2014). The Turn to Practice in HCI : Towards a research agenda. In CHI 14. Proceedings of the 32 nd annual conference on human factors in computing systems. Toronto, Canada, 26 April-1 May 2014, New York: ACM Press pp. 3543–3552.Google Scholar
- Lange, Alexandra (2012). The Woman Who Invented the Kitchen. Slate, 25 October 2012. http://www.slate.com/articles/life/design/2012/10/lillian_gilbreth_s_kitchen_practical_how_it_reinvented_the_modern_kitchen.html. Accessed on 9 February 2018.
- Lave, Jean; and Etienne Wenger (1991). Situated Learning: Legitimate Peripheral Participation. Cambridge: Cambridge University Press.Google Scholar
- Leontyev, Aleksi Nikolaevich (1978). Activity, Consciousness, and Personality (trans. M. J. Hall). Englewood Cliffs: Prentice-Hall.Google Scholar
- Lynch, Michael (1997). Theorizing practice. Human Studies. Vol. 20, no. 3, July 1997, pp. 335–344.Google Scholar
- Mankoff, Jennifer; Gary Hsieh; Ho Chak Hung; Sharon Lee; and Elizabeth Nitao (2002). Using Low-cost Sensing to Support Nutritional Awareness. In UbiComp ‘02. Proceedings of the 4 th International Conference on Ubiquitous Computing, Goteborg, Sweden, 29 September – 1 October 2002. Heidelberg: Springer-Verlag, pp. 371–378.Google Scholar
- McKinsey Global (2015). The Internet of Things: Mapping the Value Beyond the Hype. http://www.mckinsey.com/~/media/McKinsey/Business%20Functions/McKinsey%20Digital/Our%20Insights/The%20Internet%20of%20Things%20The%20value%20of%20digitizing%20the%20physical%20world/The-Internet-of-things-Mapping-the-value-beyond-the-hype.ashx. Accessed 9 February 2018.
- Ng, Kher Hui; Victoria Shipp; Richard Mortier; Steve Benford; Martin Flintham; and Tom Rodden (2015). Understanding Food Consumption Lifecycles Using Wearable Cameras. Personal and Ubiquitous Computing, vol. 19, no. 7, October 2015, pp. 1183–1195.Google Scholar
- Nicolini, Davide (2013). Practice Theory, Work, and Organization: An Introduction. Oxford: Oxford University Press.Google Scholar
- Reckwitz, Andreas (2002). Toward a Theory of Social Practices: A Development in Culturalist Theorizing. European Journal of Social Theory, vol. 5, no. 2, 1 May 2002, pp. 243–263.Google Scholar
- Reitberger, Wolfgang; Wolfgang Spreicer; and Geraldine Fitzpatrick (2014). Nutriflect: Reflecting Collective Shopping Behavior and Nutrition. In CHI ‘14. Proceedings of the CHI Conference on Human Factors in Computing Systems, Toronto, Canada, 26 April −1 May 2014. New York: ACM Press, pp. 3309–3318.Google Scholar
- Schatzki, Theodore R. (1996). Social Practices: A Wittgensteinian Approach to Human Activity and the Social. Cambridge: Cambridge University Press.Google Scholar
- Schmidt, Kjeld (2014). The concept of ‘practice’: What’s the point? In C. Rossitto, et al. (eds.): COOP 2014: Proceedings of the 11th International Conference on the Design of Cooperative Systems, 27–30 May 2014, Nice, France. London: Springer, pp. 427–444.Google Scholar
- Schmidt, Kjeld; and Liam Bannon (1992). Taking CSCW Seriously: Supporting Articulation work. Computer Supported Cooperative Work (CSCW), vol. 1, nos. 1–2, March 1992, pp. 7–40.Google Scholar
- Schmidt, Albrecht; Martin Strohbach; Kristof van Laerhoven; Adrian Friday; and Hans-Werner Gellersen (2002). Context Acquisition based on Load Sensing. In UbiComp ‘02. Proceedings of the 4 th International Conference on Ubiquitous Computing, Goteborg, Sweden, 29 September – 1 October 2002. Heidelberg: Springer-Verlag, pp. 333–350.Google Scholar
- Scott, James; Bernheim Brush; John Krumm; Brian Meyers; Michael Hazas; Stephen Hodges; and Nicolas Villar (2011). PreHeat: Controlling Home Heating Using Occupancy Prediction. In UbiComp ‘11. Proceedings of the 13 th International Conference on Ubiquitous Computing, Beijing, China, September 17–21, 2011. New York: ACM press, pp. 281–290.Google Scholar
- Suchman, Lucy; and Jutta Weber (2016). Human Machine Autonomies. In N. Bhuta; S. Beck; R. Geiβ; H. Liu; and C. Kreβ (eds.) Autonomous Weapons Systems: Law, Ethics, Policy. Cambridge: Cambridge University Press, pp. 75–102.Google Scholar
- Thieme, Anja; Rob Comber; Julia Miebach; Jack Weeden; Nicole Kraemer; Shaun Lawson; and Patrick Olivier (2012). “We’ve bin watching you” - Designing for Reflection and Social Persuasion to promote sustainable lifestyles. In CHI ‘12. Proceedings of the CHI Conference on Human Factors in Computing Systems, Toronto, Canada, 5–10 May 2014. New York: ACM Press, pp. 2337–2346.Google Scholar
- Tolmie, Peter; and Mark Rouncefield (2011). Organizational Acumen. In M. Rouncefield and Tolmie. P (eds.) Ethnomethodology at Work. Aldershot: Ashgate Publishing Ltd, pp. 37–56Google Scholar
- Wieder, Lawrence D. (1999). Ethnomethodology, Conversation, Analysis, Microanalysis, and the Ethnography of Speaking (EM-CA-MA-ES): Resonances and Basic Issues. Research on Language and Social Interaction, vol. 32, nos. 1–2, pp. 163–71.Google Scholar
- Woolridge, Michael (2009). An Introduction to MultiAgent Systems. Chichester: John Wiley & Sons Ltd.Google Scholar
- Woolridge, Michael; Nick Jennings; and David Kinny (2000). The Gaia Methodology for Agent-Oriented Analysis and Design. Autonomous Agents and Multi-Agent Systems, vol. 3, no. 3, September 2000, pp. 285–312.Google Scholar
- Yang, Rayoung; and Mark Newman (2013). Learning from a Learning Thermostat: Lessons for Intelligent Systems for the Home. In CHI ‘13. Proceedings of the CHI Conference on Human Factors in Computing Systems, Paris, France, 27 April - 2 May 2013. New York: ACM Press, pp. 93–102.Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.