Some landmarks in the proliferation of tools

6A huge number of tools already exist. In February 2012, the French Quantified-Self group listed 504, mostly developed in English-speaking countries, which it presented by theme (sport, health, sleep, weight), but also according to the medium (web application, smartphone application, specific device) [4]. The overall number of users of the different tools is difficult to estimate, as there are no statistics available. Some are completely confidential (Quantter, 42goals, Fitbit [5], etc.), while others are extremely popular with millions of athletes around the word (Nike+, Runkeeper, etc.). By the very way they are designed, these tools influence one’s ways of measuring one’s personal states or activities along several axes.

7Tools can be categorized according to their theme, from the most specialized, used to quantify a specific state or activity (such as sleep for Zeo or eating habits for DailyBurn), to the most general, which can record any activity (Daytum, Quantter, 42goals, Daily Deeds…). Between these poles, mid-spectrum tools can be used to quantify a particular type of activity and/or physical parameters: various sports with tools such as Runkeeper, Nike Running or Garmin Connect, healthy lifestyle (walking, sleep, weight) with Fitbit, etc. Tools can also be sorted by data recording mode. In some cases, data are recorded by a sensor (located on a specific tool, such as the Fitbit step recorder or the Wii, or embedded in a smartphone), and then transmitted to the interface. Conversely, in the declarative mode, data are entered manually, as is the case with all general tools (Quantter, Daytum, 42goals, etc.). Finally, some systems, such as Runkeeper, offer a “mix” of the two: some data are “captured”, others added manually.

8Almost all existing tools allow users to keep track of activities quantified in the past, over a variable period of time, thus allowing them to draw up series or to identify any regularity, progression or discontinuity. Some (sensor-based) tools also offer real-time interaction: a virtual coach (Runkeeper), a system of reminders (RescueTime, Sleep Cycle) or live encouragements in the form of “likes” (Nike+). More occasionally, tools allow a visualization of measurements according to objectives to be reached in the future, whether or not the date is fixed, for example in terms of weight (TargetWeight). Like the other differentiation axes mentioned here, this temporal orientation results in a more or less sophisticated representation of data. Visualization methods can be basic, to a greater or lesser degree, or on the contrary customizable and complex: a simple number measuring a deviation from one’s aim, a calendar representation of frequencies or duration of activities, or even data analysis graphs (curves, diagrams, pie charts, etc.). Most tools are game-like in some respects. Aside from the different forms of gamification (badges, feedback, challenges), the very principle of competition around numbers is generally presented as having a recreational interest. Tools thus offer several degrees and modes of publicity, sharing and interacting. The data are either private or immediately visible to a community of users, as in the case with “network tools” (interfaces built on the model of Facebook, where the user has a personal profile and where all users form a “community” in which posts, friendships and mutual reactions are exchanged). In most cases, the data’s degree of disclosure can be adjusted through setting options. If data are transferable, they can be broadly displayed by sharing them on several community/networks [6].

9First of all, what should be taken away from this overview of available tools is that they open up hardly any precise avenues for defining new fields of self-quantification. In fact, the most successful tools are based on health or sports measurements whose instruments have been developed for a long time now. Although it is, in a way, an important operation, all they actually do is to hand to the general public tools that until then were wielded by experts or professionals. They do not define new areas of the individual as quantifiable, but trivialize the practice of quantification for private use. A few tools allow their users to quantify previously unquantified activities or states, either by relying on sensors whose miniaturization and democratization make them available for everyday use (Fitbit’s step count, for instance), or by inviting them to construct new quantitative indicators to represent qualitative states (Moodscope for mood). In the latter case, the tools propose to extend, systematize and trivialize self-assessment techniques also used in the medical field, which consist in quantifying subjective variables (such as the pain indicator, for example). Finally, general-interest tools let each individual user decide which activities should be tracked, and therefore do not determine any specific activity as being particularly suitable for quantification.

A new field at the interface of numbers and intimacy

10Our research is based on forty in-depth semi-structured interviews, on the analysis of the respondents’ own figures, and on an exploration of the exchange platforms they use when they share their personal data [7]. Interviewees were initially found among groups of people who turned out to be relatively scarcely equipped, but were in the habit of quantifying their weight, diet or physical activity. A quarter of our respondents recorded measurements very regularly, using a notebook or an Excel file. We also paid attention to several members of a network of heavily equipped runners, who were also bloggers and active participants of the Runnosphère association, which provides a platform for them to share their common passion. Finally, we interviewed Parisian members of Quantified Self who used, reviewed and promoted many personal monitoring systems. Though the age of our respondents varied widely, they were predominantly men, especially among the runners and supporters of Quantified Self, who were all very keen on new technologies. In terms of social groups, the interviewees were positioned on a continuum ranging from employees to managers, and included skilled workers. There were a few students and several freelancers. As for their professional fields, many respondents had jobs related to technological innovation in the field of digital, communication or computer science, and a few worked in fields related to sports and health.

11The activities quantified by our respondents can be sorted into three main blocks:

1. sports activities (specifically running, for 18 of our respondents)
2. activities related to health and lifestyle (weight, for 18 people, and more marginally various items such as sleep, food, mood, blood pressure, tinnitus, and migraines)
3. balancing work and leisure (some respondents measured either the time spent in front of their computer or television, or their productivity at work, their “Zen moments”, their “digital breaks”, etc.) [8].

12We have chosen to look at all these measurements as ways of producing figures about oneself or one’s activity. Due to their heterogeneity, they do not allow us to draw a profile of quantification enthusiasts. A few examples of the diversity of our sample include: the case of a 42-year-old Republican Guard, Erwan, who had been keeping a notebook of his swimming pool sessions for several years; that of a 56-year-old sophrology teacher Roselyne, who was worried about a recent tendency to gain weight, quantified her food and weight in Excel files, and had tried an application dedicated to food monitoring, called DailyBurn; Sylvain, a 32-year-old surgeon whose partner nick-named him “Mr Data”, and who was passionate about running and everything that allows him to follow and comment on the evolution of his performance: sophisticated watches, connected scales, internet platforms, specialized blogs and sites; or Henri, a 41-years-old consultant in a Web agency, who had been closely monitoring his health, mood, weight and nutrition for nearly ten years, and was reviewing all the new digital tools that could help one to adopt a healthy lifestyle.

13The approach that we propose in this article is inspired by two research fields that deal with related subjects but have not addressed the question of the personal uses of quantification, which is obviously minor in comparison with their main focui: works on the socio-history of quantification, on the one hand, and works on the anthropology of ordinary writings, on the other [9]. The self-measurement practices appear as being both very marginal and sometimes even as a blind spot. Note that domestic uses of quantification fall outside the scope of socio-historical work concerned with quantification (Porter, 1986; Desrosières, 2008), and are not either addressed in the anthropology of numbers, which describes the institutional uses of numbers and accounting, and their propagation from religious to economic spheres (Crump, 1998) but not their uses in daily life.

14There is also little evidence of personal statistics in approaches to ordinary writing practices, inspired by pioneering work on graphical reason (Goody, 1978) and carried out in France by researchers gathered around Daniel Fabre (Fabre, 1993, 1997). In our view, however, the analysis of numerical writings should, as Jean-Pierre Albert’s contribution suggests (Albert, 1993), be an integral part of these studies. In some of the functions of domestic writings that he analyzed, Albert identified numbers that were sometimes linked to the need to organize the household (“relieve memory”, “list and programme”, and “do the accounting”), and sometimes more closely linked to self-assertion (“being oneself”). It is thus the need to manage time that gives rise either to a posteriori notations on excursions and visits in retirees’ schedules, or to counting the last days before a break for students and teachers or military personnel (vacations, end of service). What is at play in noting and counting is “acting magically over time when it is painfully felt”, he writes. He is therefore not indifferent to the personal practices of measurement. It should also be noted that he insists on the continuity between ordinary accounting practices, the dissemination of material specifically for the exercise of balance sheets (accounting records) and the existence of an accounting reflexivity “waiting only to be extended to other aspects of life” (p. 9). However, as soon as he mentions the dimension of reflexivity and identity (“writing to see more clearly”), the question of quantification is sidelined: here, self-writing is classically approached as setting oneself in words, not in numbers.

15Personal uses of statistics are however linked to identity building and self-reporting. It is with this in mind that the use of personal numbers is currently becoming increasingly interesting for researchers, as a series of recent studies attests. The Journal of Knowledge Anthropology recently devoted a special issue to “confronting the traces of one’s activity” (Cahour and Licoppe, 2010). It is also with this in mind that Fanny Georges theorized “calculated identity”, based in particular on Facebook profile data (Georges, 2009). In relation to this work, our research is distinctive in its focus, namely the deliberate production of numbers about oneself, as pointed out in the introduction to this article. In this sense, our study is closer to those of Julien Soler and Pascale Trumpet (2010) or Marie-France NguyenVaillant (2010), who are working on the construction by individuals of numbers meaningful to them (e.g. logs for diabetes patients, or developing a threshold for epileptic seizure detection by patients’ relatives). What differentiates this study from these is the fact that they are anchored in the medical field, and that the quantification practices they examine are therefore supported by institutional practices from the outset.

Monitoring, routinization and performance: some ways of representing oneself in numbers

18When measurements are used for monitoring purposes, quantification is not a “bowl of cherries”, but a measure of risk. The notion of a threshold is central here: one checks that one does not reach a critical threshold; in fact one tries to move away from it (for example, a cholesterol level from which a drug treatment would be prescribed). Most of the time, this threshold is defined externally by standards (often medical), but sometimes it is more personal (“I can only bear myself when I’m under 58 kg”). The act of taking and reading the measurement can be painful, depressing and discouraging. Fabrice, a 51-year-old divorced technician, monitors his cholesterol levels and notes his diet in an Excel file:

19I fill in my meals on Mondays, Tuesdays, Wednesdays … and at the end it adds it all up and it says for example: “watch out there, you’re in the red” because I ate too much cheese, for example.

20The particularity of monitoring measurements is that they are not primarily oriented towards actions, although they can often help to guide them, and in this sense to generate other actions in return. In this monitoring logic, the parameter continues to be measured even during periods when there is little control over its evolution. For example, one continues to weigh oneself while an injury precludes any form of exercise, or when a professional schedule involves many meals in restaurants. The regularity of this self-monitoring is important, and recording is often based on a long-term commitment. Here, numbers are revealing and their reading can create some anxiety. Sharing, in the case of surveillance measurement, is not the obvious thing to do, primarily because of the nature of quantified phenomena (often medical or intimate parameters). In the online exchanges that our interviewees gave accounts of, such personal measurements were generally not published. Rather, they included advice, with a view to helping one another manage the problem. Thus, Fabrice often visited a cholesterol forum to ask specific questions, answer newcomers’ questions on the forum, or participate in debates that stirred the community of people concerned by cholesterol, but never to provide and comment on the data he produced about himself [10]. Monitoring logics can also apply to time budgets (not wanting to watch more than 6 hours of television per week, not wanting to exceed a certain Internet connection time daily). They generally appear in the accounts of all of our interviewees who track their financial expenses and want to “avoid being in the red”.

21Regularity (or routinization) measurements can take three forms: breaking with a “bad” habit (for example, quitting smoking, reducing alcohol consumption), developing a new habit (swimming every week) or staying consistent in a good practice (walking every day, having a balanced diet). It is common in sportspersons who do not set targets for performance improvement. Thus, Maiwen, a 30-year-old self-employed real estate rental manager, says:

22

For me, for sport, I’m not going to have a time goal, what matters to me is my regularity. I have to do this swimming once a week.

23The search for motivation is central. Unlike monitoring measurements, regularity measurements are carried out with the aim of serving an action: the measurement and its representation (for example, checking boxes on a calendar to record the completion of a daily routine) are sought for their incentive or even binding nature. The aim here is to install a routine (swim twice a week, write one page per day, etc.), so when the habit is acquired, the notation can stop. Amandine, a 36-year-old freelance Web project manager, started using the website Quantter to force herself to write a certain number of words a day. However, she quickly noticed that:

24

Now I write every day, except that I forget to “quantt” it, because I’ve reached my goal, so it no longer shows on Quantter.

25Here, third party gazes may also be sought to sustain the resolution, but the personal and often trivial nature of the action taken provides little matter for extensive interactions. The publication of personal data sometimes attracts friendly comments from friends and family, or from the other few people testing the same tool. Thus Henri, a 41-year-old consultant in a web agency, finds it pleasant to be encouraged to walk but complains that:

26

On Fit Bit, I try to find friends—well, that’s the exploration side—but I suspect people use it intermittently. I have three friends, and I’m still the one who walked the most. Because they have numbers that are weird, they have, you know, 3,000 steps a week, I’m thinking, “Man, he must have worn it one afternoon to see what it felt like!”

27Third party gazes may also be shunned if the data are considered to be very personal and/or if social judgment is likely to be critical (e. g., alcohol consumption statistic).

28Finally, performance measurement is used to monitor the effectiveness of a specific effort, often in sports, though not only. In this case, it makes sense to record data only if the person can act upon it. The case of Paola, a 32-year-old head of a music label, is a perfect example of this. Having long been overweight, she created an Excel spreadsheet called “Paola’s Evolution”. Since 2005, she has recorded her weight every time she has started a diet (she had started quantifying her diet on a sheet of paper but gave up). Paola explains that she stops scoring during the Christmas holidays because she knows that her weight will not drop during that time:

29

In December, I don’t write it down, right, it’s useless, because December is December.

30In fact, on the file that she kindly let us have, only the sentences on weight reduction were written down.

31In performance measurements, numbers take on meaning in terms not of thresholds but of personally defined objectives, which can be re-evaluated as one’s performance progresses, and the measurement is made both to motivate oneself and to improve performance (speed, stamina, etc.). Manipulating figures that quantify efforts and results is pleasurable, as these indicators mark out one’s progression towards a goal. The number therefore becomes important, to the extent that the activity’s records sometimes tend to be confused with the activity itself (one no longer runs without measuring performance). Aurélien, a 31-year-old economist in a large company and a running enthusiast, considers himself to be addicted to the Runkeeper application. Referring to an episode in which he was unable to record his race data, he commented on the un-motivating nature of the event:

32

I felt like I ran for nothing… I ran and… I don’t even have a trace of my race.

33Measurements become objects of intense attention, giving rise to analyses and before/after comparisons (retrospective readings), and quantification takes on meaning in different temporal phases: the short time needed to prepare for a specific goal, the time of a season, the progression from one year to the next. This makes it easier to engage in exchanges on blogs or on social networking sites, be they dedicated or generic, around the sharing of experience and skills, on measurement methods and tools or on performance, sometimes in friendly terms (encouragement) and sometimes in a competitive mindset (challenges).

Figure 1

An example of a personal review in figures

An example of a personal review in figures

34As a conclusion to this presentation of these three usage rationales, it seems important to highlight that they strongly structure the relationship between people and the digital tools meant to make their task easier. It is within communities of enthusiasts gathered around a common practice (for example, running) that the uses of techniques are strongest. The measurement tools become the subject of discussion, in the same way and at the same time as the practice itself. Their use is part of the process of self-construction through which one identifies as an athlete. Thus, the Runnosphère’s respondents are heavily equipped (Garmin watches, Nike+ sensors, Runkeeper application on iPhone, etc.), and the use of these devices is an integral part of their sporting passion. Conversely, people who engage in quantification of their condition or of their domestic practices, who measure health parameters or the time they spend on certain activities, such as the household, are much less likely to be equipped and less likely to be passionate about the tools they use [11]. It should also be borne in mind that this typology describes practices, not individuals. Thus, the same person can use measurements in a performance logic for the follow-up of their running activity, and adopt a routine use to monitor their cycling activity. Finally, several respondents described various ways of moving from one logic to another. A number of trajectories are possible and one could, for instance, start enjoying playing with numbers and move from a routine logic to a performance logic, or, on the contrary, switch from a surveillance logic to a routine logic once the danger is averted. We know also that many people simply stop quantifying themselves. Our study is not extensive enough to provide an analysis of these trajectories here.

Is encrypted reflexivity a step back or a zoom in?

35Producing measurements about oneself may seem to reveal the growing reflexivity of the contemporary subject—a characteristic that is abundantly highlighted in the sociological literature on the individual (Giddens, 1991; Beck, 2002; Bauman, 2003). We discussed the reflective dimension of this practice in a previous article (Granjon, Nikolski and Pharabod, 2012). We then pointed out a sort of “limitation” of reflexive practice which, in light of Foucault’s care of the self theories, can be described as instrumental reflexivity rather than oriented towards self-knowledge (Foucault, 2005). We will come back to this issue here, looking at it from a slightly different angle. Our analysis of our field shows that the form of reflexivity at play in quantification practices has other distinctive features. On the one hand, it relies on a strong hybridization between objective and subjective affordances on oneself; and on the other hand, it implements a form of focalization of attention, akin to a “macroscopic” representation of oneself.

36The implemented apparatuses are supposed to allow for a more rigorous reading of the person’s activities. Contrary to the discourse of the proponents of Quantified Self, our analysis of our interviews shows that this idea of quantification as a practice that reflects progress in reflexivity should be treated with caution. In most cases, the knowledge effect brought about by self-quantification is limited. It can be broken down into two effects: a revelation effect linked to the reading of the measurement result (e. g. a number of steps per day), and an analysis effect linked to data serialization and comparing variables. According to our interviewees, the revelation effects obtained through quantified data are limited. A few sensors can be used to measure one’s habits: number of steps per day, quality of sleep, household’s energy consumption, etc., but most of the time the measurement is presented as a simple diagnosis. It is meaningful only occasionally. For those who regularly measure themselves, the data generated is generally not surprising. Only Baptiste, a 24-years-old data mining engineer, remembered being surprised, during a holiday in Italy, by the fact that the number of steps he took during a day’s hiking in nature turned out to be smaller than the one generated by a day’s visit to the city. On a daily basis, the result of the measurement is often expected. Thus, Rémi, a 33-year-old jewellery designer, points out a form of triviality:

37

I would say I don’t really need that to listen to my body, to know if I feel good, to know how I’m running. You can feel if you can run a little faster. You also need to know yourself, to hear yourself.

38Furthermore, the analysis of the results is also limited. Interviewees perform few statistical operations on their data and the graphical representations provided by the tools they employ are largely underused. In most cases, readjustments of reflexivity and action are oriented by the mere serialization of raw figures (calendar representation). Even if it remains very occasional, data analysis is more in line with the logics of performance or monitoring than with routine logics: for example, it is a matter of going back in time to understand the occurrence of an injury or to take stock of a season.

39Rather than a knowledge-based approach, ordinary quantification is often justified by a desire for self-transparency: you “write down the numbers,” you “keep track” so you “don’t lie to yourself”. This claim mainly concerns routine measurements, but may also apply to performance and monitoring measurements. But here again, the practices are rarely in keeping with this displayed rigor and the whole game is about measuring oneself exhaustively. Even if one does not rig the figures that one builds for oneself, one chooses what to measure, and so one ends up holding back measurements when they become too depressing, and measures only positive things (e. g. mixing the kilometres of running with kilometres of walking, etc.).

40

My neurologist wanted me to count my migraines. Oh no! No, I don’t count them… that way, they don’t count!

41says Viviane, a 35-year-old sales representative for a brand of personal hygiene products. Behind a thin veil of sought-after transparency, respondents revealed a desire to construct a gaze upon them that is neither cold nor neutral, but benevolent and encouraging them to do well. It is thus a matter of taking care of themselves.

42In many approaches, [12] as well as in “common sense”, numbers are seen as radically opposed to subjectivity. Even the enthusiasts of the Quantified Self group are aware of this criticism that is often addressed to them: the culture of numbers is seen as the opposite of the culture of sense (in the dual sense of a quest for meaning and a quest for sensitive experiences). However, rather than an opposition, empirical findings shows a hybridization between these two logics, a strong entanglement between objectification movements and subjective support: the measurement of states or activities by instruments (sensors), or with recording tools (from the notebook to the connected platform), is articulated to the mobilization of subjective self-knowledge, rather than being opposed to it. Thus, we have seen athletes pay a lot of attention to their sensations and to the gap between the measurement and the feeling behind the measurement (notably pain). A few of them also keep track of their sensations (subjectively rated), over long periods if necessary. However, their interpretation of this data is mostly immediate, and involves self-knowledge (memorized markers of time, weight, etc.). The size of series recordings is generally of no interest in giving meaning to the measurement. In this respect, it is striking to note that even almost flat curves (frequent among Withings users or swimmers whose swimming times are always equal) are commented on: their content is defined far more by sensations than by numerical data. At the same time, the very meaning of a measurement is frequently expressed as an emotion. The effect of a measurement on the person taking it is, first and foremost, satisfaction or disappointment, rather than an orientation towards action: when reading a weight on one’s scale, one first says “it’s good” or “it sucks!” before saying, “I have to eat less.” It is this strongly emotional aspect of numbers that explains some of the behaviours that we have observed, such as erasing or not recording data when one knows that it disturbs a series (noting calories when one has “strayed”), or explaining to readers of a running blog how to read a number given by a sensor. For example:

43

There, the time is not right because, in fact, because I forgot to stop the timer when I answered a phone call.

44How should we understand this production of measurements which ultimately seems to be anchored in subjectivity, memory, emotions and sensations that words would more easily convey? What do the figures add to feelings? In shifting from a daily notation of words (dietary diet) to a daily notation of numbers (the calories ingested), one creates an oriented series that shows a trend, be it minimal. Where there previously was nothing to see, a kind of evolution is “created” (where today is better or worse than yesterday). In terms of daily variations, curves never appear to be totally flat. Even with the knowledge that these minimal variations do not have any “meaning”, one draws from them a sense of change, of existence, of control over one’s future, even without learning or changing anything in the long or even medium run. This is particularly evident in interviews about weight-related measurements. Many people report that they cannot avoid looking at their weight too often, even though they know that variations “don’t mean anything”. Similarly in running, Laura, a 30-year-old editorial secretary, confesses that since she has her running watch:

45

It’s a bit of an obsession[…] I try not to look at it for every kilometre […] I’ll look anyway. It’s already good if I can keep from looking at my speed in real time, which is… in any case, I know it’s distorted.

46Therefore, statistics may be desired not for their “objectifying” power, but on the contrary for their “subjectifying” aspect, their ability to produce emotion at the cost of a loss of insight. In the same logic, switching frequently from one measurement tool to another, as some members of the Quantified Self do, can appear as a way of renewing one’s gaze on oneself, of seeing oneself differently. Foremost, to quantify oneself is to change the lens through which one looks at oneself.

47The reflexive dimension of this practice thus hides an almost inverse effect: rather than a step back from oneself, many quantifications are part of a movement of focusing attention, a kind of “zooming in”, on a particular aspect of oneself, which precludes any broader perspective. And only sometimes is another change of focal distance grafted onto this change of focus: the change in perception then brings about a change of behaviour, activities or state.

A performative tool for the project-individual

49By examining the peculiarity of a form of reflexivity that is mediated by numbers, we have seen how measurements can inscribe a person in a dynamic representation, in a movement [13]. This notation’s performativity can also be a source of what one respondent called “chain strength”, that is, measurements are thought to have an effect on an individual’s action, as an incentive-type constraint. Alex, a 22-year-old computer science PhD student, summarized the working principle of the 42Goals interface he created:

I’d like to try to make all my days productive. So then, if for instance I already have four productive days in the week, it makes me want to be productive for the fifth to continue the chain. For users, this is a motivation.

Self-quantification and its normative underpinnings

50Using numbers to describe a phenomenon is not a neutral act. In L’Argument statistique (I), Alain Desrosières (2008) argues that quantification (statistics, accounting, etc.), in “allowing” for certain phenomena to be measured, makes it “natural” to measure and understand them in terms of numbers, and thus produces a new relationship with the social world, new representations of “quantifiable” objects, and new ways of acting. The tools proposed today to increase the number of metrics applicable to oneself, as we saw in the first part of this text, hardly designate any new object as being quantifiable. On the other hand, they trivialize and generalize state or activity measurements that were previously taken within an institutional framework. Does its privatization make the measurement of activities or personal status lose all references to collectively defined criteria, thresholds or objectives?

51In the quantification practices that we have observed, individuals determine precisely, through key figures, what they consider normal and according to which they will adjust their actions. However, these values vary in scope and status, depending on the type of quantification. Thus, standards are very important for monitoring measurements: the thresholds that frame them are usually drawn from outside sources and users take them into account, sometimes as they are, sometimes by adapting them (taking into account dominant medical/aesthetic standards—e.g. cholesterol levels, body mass index—BMI, “ideal” weight). This adjustment to a standard is less apparent in routinization and performance measurements. In the latter, collective benchmarks very often exist, and records and levels to be attained in a given discipline are often in the hands of institutions, such as sports federations, for instance. But they are not always actualized in amateur practices. If runners’ equipment is in line with that of professionals, if sophisticated watches with a chronometer, an altimeter, a cardio-frequency meter, etc., create references previously absent from solitary practices, amateur athletes will mobilize them both, to seek relative and absolute marks. A thread on a running enthusiasts’ forum shows the ambivalence between the privatization of benchmarks and the search for an alignment with collective reference values:

52

At what point can you say that you’re very good on 10 km? Everything’s relative but I find that around 33’, it’s starting to become a good time. Some people will say that everyone has their own reference, okay, but that doesn’t stop them from setting limits. Just to qualify for the French 10 km road championships in the senior class, you have to do less than 34’, which seems like a pretty good bar to me [14].

53Many of the people we met were looking for peers with whom to exchange numerical data on online platforms, who would be roughly on the same level as themselves, and with whom they could compare their progress. In reading other people’s data, they try to see if they share the same degree of commitment to the practice, if they run in the same position, if they have the same goals—10 km, half-marathon, marathon, trails—, if they spend about the same amount of time training, if they are in their age and weight range, etc. Here, statistics are used mainly as matching points for runners. This comparison makes it possible to base the personal path of progression on collective values, without however reducing the personal data to a broader scale that would rank riders from the least good to the best [15]. It should be noted that common reference values, numerical standards, do not necessarily provide positive support for performance practices. They can also be contested and performance measurements can thus be diverted from conventional uses. This is the case of anorexic patients who, far from monitoring their body weight, use their scale to measure performance in losing weight when they are already below the lower end of the BMI range (Darmon, 2003).

54In routinization measurements, reference values do not necessarily exist and the individualization of numbers, thresholds and even units of measurement is greater. Quantification tool developers take note of this proliferation and customization of reference values, or even of units of measurement. An Internet user thus notes on a website that aggregates accounts of Wii Fit users:

55

Otherwise, other than new exercises, what’s interesting is that objectives vary. One can aim to lose so much weight, do so many hours of this or that a day, or burn the calorie equivalent of a fruit or of a cake. Right now I’m going through a glass of coke every day.

56While this customization of measurement units satisfies some of our respondents’ expectations—those who want the tools to match their quantification needs as closely as possible—it overrides the principle of equivalence, and thus makes comparison impossible (kilos and glasses of soda cannot be directly compared). It is thus apparent that sharing and discussing statistics is above all common among, and significant for, patients or athletes—users who share universally accepted metrics in the frameworks of monitoring or performance measurements. In order to share numbers, people need to share the same standards, thresholds and numerical targets, and often the same equipment, so that each party’s figures are reliable. It is thus revealing that the personal data most viewed and commented on by runners on their peers’ blogs—which are the most important to them—are the official figures, measured in a race situation. In routinization measurements, individuals do not mobilize absolute or precise relative references. These quantifications are individual responses to a general injunction to take oneself in hand. Voluntary scoring in various fields of activity (healthy lifestyle, personal activities) introduces almost professional management methods, but the assessments thus made do not resonate with standards or averages, which are, as such, non-existent.

Towards a standardization of private activities?

57When associated with online sharing, routinization quantifications lead to a focus on trivial, insignificant things, which are given credit on the sole basis that they are supposed to say something authentic about the individual. They confer visibility on ordinary and common behaviours (swimming twice a week, losing 500g, working well on a given day at the office, etc.) related to the most informal areas of daily life (Javeau, 2006). Contrary to sports blogs, where there are more technical and sometimes more conflicting interactions, exchanges of personal data on social networks are characterized by their very limited and consensual nature (“great!”, “bravo!”, “way to go!”, etc.). It must be remembered that this valuation of trivial things is only very marginal, since most of our interviewees did not find it useful to “pollute others” with figures that were of interest to them only. But it points to a specific issue that we would like to address in this last part: that of the articulation between numerical reflexivity and the standardization of private activities. This was an underlying question throughout our research because the practices of measuring personal activity were reminiscent of management techniques peculiar to the professional space: on the tail-end of measurement practice is the evaluation of tasks, which depend, in turn, on their standardization [16].

58In the case of routinization measurements, the introduction of digital tools is not neutral in this respect. In switching from personal notes in a notebook or even an Excel file to the use of a formatted tool to record one’s personal activities, the issue is the introduction of standards where previously no better practices had been defined (how much should I walk per day? how many pages should I write per day? do I have to count pages or time spent writing? how many minutes should I devote to relaxation?). Any standardization work requires a restrictive definition of the activity to be standardized: professional tasks are carried out under working conditions, on workstations, by qualified persons using skills sheets, and so on. Likewise, sports authorities set the conditions for competitions so as to make competition possible. In everyday life, objectives are not listed and levelled out, situations are not stabilized, equivalencies are not constructed, and contexts are not neutralized in such a way as to make magnitudes commensurable [17]. This is true to such an extent that measurements made opportunistically, because the tools exist, can hardly acquire meaning. This is at least what the interviews with people interested in the Quantified Self movement have shown: few people manage to stabilize their use of a tool beyond discovering it. And if they can find a use for themselves, the sharing possibilities (emulation, comparison) offered by the tools turn out to be difficult to exploit. Some tools allow users to compare their data with that of other users on the same website. For instance, Fitbit ranks users’ daily physical activity level among other users of their gender, age group and BMI. Even presented in this way, according to the respondents, these statistics are completely useless. People do not record the same things:

59

Some people record every step they take all the time, others track themselves only very occasionally, others record themselves only when they run, so it’s lame, it means nothing!

60commented Nicolas, a 35-year-old engineer who monitored the time he spent on various activities (reading, television) according to thresholds he had set for himself. Contrary to quantifications introduced to govern, studied by Alain Desrosières, the promoters of personal quantifications work in the reverse order: whereas in the former context, “to quantify is to agree first and then to measure” (Desrosières, 2008), in the latter context, to quantify is to measure first and perhaps to agree later, because for the time being, we are far from witnessing the emergence of a standardization of private activities. Although they are encouraged by the promoters of the tools and the proponents of Quantified Self, discussions among users are rare, and it does not seem like “alignments” of users’ practices should be strongly expected any time soon. For our respondents, there are no better practices in personal quantification than those that are personally convenient to them and which they sometimes use in complete confidentiality. Thus, some people use tools that oblige them to make reported data public, but code the activities measured in such a way that they alone understand what they are about. In this way tinnitus, migraines or alcohol consumption are hidden from the gaze of others.

61The handful of tool designers we met with were also embarrassed to raise the subject of best practices to be promoted. As if to better avoid the question, the founder of a general interest website for declarative tracing of users’ activities, consumption habits, etc., claimed to use his tool to see whether he or his colleague drank the most beers, just for fun. The embarrassment of the designer of a tool to support healthy lifestyle practices is also revealing. When we asked her to explain the algorithm used to represent a better or worse day of physical activity as measured by the sensors on a smart phone (in the form of a more or less blooming flower), she claimed to be incompetent. When we insisted, she said:

62

What is a good number of steps per day, though? What transparency will you have in relation to the users, to tell them?

63And to answer her question:

64

It’s not our role, we’re not going to communicate on this, it’s a bit complicated to say that it takes so much or so much, especially since the specialists don’t agree, so we’re even less legitimate than them to say that.

65—which does not stop tools from being produced, and algorithms from self-indulgently containing standards. In the world of personal quantification, no one seems to want to ask the question or discuss the construction of new reference values to guide daily actions, but some already boast of having raised averages. Women who use the Withings scale weigh themselves on average 7 times a month, and pick up more weight at Christmas than at Thanksgiving [18]; and a run using Nike equipment lasts on average 35 minutes [19].

66This sensitive point on the connection between quantification and standardization can shed light on a precise aspect of the fieldwork carried out with Runnosphère bloggers, the only ones who truly communicate on their practices. Analyzing the interviews and reading the comments posted on the runners’ blogs, we were struck by the success of a two-person running story, mixing data captured on the joggers’ bodies and an account of their sensations during the training, complete with photographs. The attention and success of an unusual detail of their run, an encounter with couples of toads, shows that their experience defied any form of standardization: it remained unique and non repeatable, even though it was measured and timed in such a way as to fit into an otherwise quite rational training programme. These forms of self-narration on the runners’ blogs, which eliminate any risk of making the singular experience disappear behind the standardized practice, attract attention. They are the ones that all interviewees say they prefer to read. On the other hand, it has to be said that the elaboration of such an original written narrative is a rare form of content, which is peculiar to blogging, and that much more minimalist narratives of oneself are proliferating on social networks. The singularization of experiences is conveyed by images (a photograph of oneself) far more than by words. Moreover, the use of thumbnails showing data collected by automatically formatted sensors—sometimes with a very short commentary in the form of a status update—is becoming more and more widespread (Figure 2).

Figure 2

Examples of standardized personal statistics posted on Facebook

Examples of standardized personal statistics posted on Facebook