Adapted from a paper written in conjunction with Daniel Epstein for the CHI 2015 Workshop “Beyond Personal Informatics: Designing for Experiences of Data”. The original research publication, which includes additional content and references to many more relevant academic studies, can be read in its full form on Daniel’s website.
A Brief History of Wearable Technology
Wearable technology has been around for centuries. Step tracking traces back to Leonardo da Vinci, who designed a waist-worn mechanical contraption that responded to walking. Step tracking received widespread use with the manpo‑kei (万歩計, literally the “10,000 steps meter”), developed in Japan in 1965. These pedometers relied on mechanical methods and did not explicitly support people in reflecting on historical data.
The Rise of Personal Informatics
The term Personal Informatics was first coined by Ian Li and his colleagues in 2010 as “a class of applications to help people collect and reflect on personal information”. At the time, these applications were growing in popularity, in the form of both research prototypes and commercial products (such as the Nike+). Some self-tracking pioneers even took to designing hardware to record every aspect of their lives.
Ian and his collaborators surveyed predominantly young, educated, and technologically savvy people — a group that, at the time, was very representative of self-trackers. In the five years since the publication of their breakthrough paper, personal informatics has reached a tipping point. Self-tracking will soon be ubiquitous, as foreshadowed by the pervasiveness of self-tracking tools in modern smartphones and widespread general interest in tracking for health reasons.
Wearables in 2015
Wearable technology has now reached a critical mass, and is no longer strictly for technology-focused enthusiasts. Major corporations such as Apple, Garmin, Intel, Microsoft, and Samsung have all announced or released consumer wearable bands and watches that incorporate a variety of personal informatics features. Personal informatics has also spread to new tracking domains, with niche products like Vessyl and Hapi Fork. People commonly use multiple wearable devices over time, and they buy new devices (when they find new data they want to track and/or their devices break or are lost).
There are subsets of the personal informatics community that are predominantly young, educated, and technologically savvy (for example, the Quantified Self community). There will always be self-trackers who fit this profile, but they are not going to maintain their majority status.
These characteristics will soon stop describing the “average” self-tracker.
Personal informatics tools are seeing wide adoption, but critics argue that they do not meet the needs of important groups, including those who “need them most”. To support a broader audience, we must explore, understand, and design around a wider set of human experiences.
Designing for the Future
There are many demographic factors that should influence design in personal informatics. The following are some areas that offer opportunities for future research and design. Consider them to be examples of problems blocking widespread adoption of personal informatics tools, not a comprehensive list of the areas that we should explore.
Gender and Sex
Apple HealthKit is pitched as being able to track “all of your metrics you’re most interested in” but launched without the ability to track menstrual cycles. Hip worn sensors cannot be clipped to all outfits, such as dresses (the suggested workarounds of clipping the sensor to a bra strap or undergarment are not ideal, as they limit access to the display). Most fitness bands and smartwatches are large and bulky, ill-suited for small wrists.
Designing wearables for women cannot simply be an application of the infamous “shrink it and pink it” principle. To be truly inclusive, we must consider both the physical (ergonomics, reproductive health, sex-related disease profiles, etc.) and the emotional (social/cultural norms, varying personal gender expression, etc.).
Commercial wearables have only recently begun to acknowledge gender differences, with most devices taking into account only basic physical differences aligning with sex (wrist size, stride length, metabolism). Although a few wearables have been designed to appeal to women (such as Fitbit and Misfit’s designer lines), many devices remain large and masculine. Form factors will improve as hardware advances allow smaller devices, but the limited selection of gender-conforming options undoubtedly affects adoption.
Current wearable devices are aimed at young to middle-aged adults in their advertising, functionality, industrial design, and companion app design. There are significant potential benefits to explicitly targeting people outside of this age range. Childhood obesity and activity levels are linked to negative health effects later in life. Child‑appropriate wearables could provide both parents and medical professionals insight into a child’s activity levels, as well as noteworthy patterns affecting activity. Rich activity data coupled with clear recommendations could assist parents in monitoring their child’s health, as well as providing a more concrete way for older children (preteens and teenagers) to take ownership of their own health.
A few commercial pedometers have been marketed to children, such as the Pocket Pikachu, but these do little more than track and report steps taken. The HCI community has had some successes designing wearables for adolescents and teenagers, but these implementations are limited by the fact that they are research prototypes, not commercial products.
This doesn’t just apply to the young. The Digital Family Portrait is an example of an early personal informatics research tool that created a way for grown children to maintain peace-of-mind regarding elders without undermining their autonomy. The team found that passive tracking relieved much of the burden on individual caretakers, allowing them to devote more energy to maintaining relationships with their elderly relatives. While some general principles of designing for older adults are likely to translate directly to wearable device design (e.g., bigger screens, easy to press buttons), we haven’t yet integrated the unique motivations and needs of this demographic into our collective design consciousness.
The current trend in personal informatics is to present data in relatively raw, unprocessed form. The majority of commercial apps present predetermined metrics (steps taken, calories burned, minutes exercised) paired with abstract summary metrics (such as Fitbit’s Activity Score and Nike+’s Fuelpoints). Longitudinal data is presented in graph form, which assumes a certain level of graph literacy. Even those who are confident in their ability to read graphs are likely to be misinterpreting their data and acting on erroneous conclusions. Statistical insights, or lack thereof, should be clearly called out to help people make meaningful decisions. If wearables are to be useful to more people, the information that they collect must be analyzed and presented such that it is understandable and actionable to all of those people.
Geography, Context, and Environment
The standard pedometer goal of 10,000 steps per day might be wholly unrealistic on the streets of suburban Los Angeles, but achievable as a mere matter of routine by someone who commutes by walking in the dense urban downtown of Seoul. Devices should take geographic context into account (urban, suburban, rural) and recommend goals and supplementary activities that are appropriate. The benefits of wearable devices are undermined when they encourage risky behaviors or set users up to fail their goals.
Sensor suites should also be adapted to a person’s context and environment. Residents of dense urban centers may derive great value from information about air quality or noise pollution, but people in other environments may find that data meaningless. Outdoor physical activity is also significantly affected by weather, especially precipitation and extreme temperatures. To account for these limitations, fitness wearables need to effectively support indoor activities. This is starting to become more pervasive, (for example, Microsoft band supports gym workouts and most pedometers work effectively on treadmills), but it still remains difficult to track many indoor sports and (e.g. rock climbing, racketball, swimming).
Race, Culture, and Socioeconomic Status
Though products like Apple Watch are starting to introduce more variation, many current wearable devices are available in limited colors and styles. They largely adhere to a specific upper-middle class, tech‑friendly sensibility. These devices are identifiers for a specific subculture, and therefore exclusive of other subcultures. The industrial design of existing devices conforms to the values of an outspoken tech-forward subculture: sleek, minimalist, largely monochromatic, and LED-laden.
Are current wearables as fashionable to a teenager living in Harlem, an elderly couple in Beijing, or a schoolteacher in Paris? Style, fashion, and personal expression are highly cultural, and industrial design signals both who a device is for and who it is not for. Interface design sends similar signals: data reporting and incentives for behavior change in modern wearables target a narrow demographic and assume a specific set of cultural norms.
Going beyond questions of cultural fit, access and cost are ongoing issues. Wearable devices are generally expensive and assume a certain level of infrastructure. The majority of devices are designed to be paired with a smartphone, and companion apps are largely useless without consistent Internet access.
The costs associated with personal informatics devices are too high, for both the device itself and the supporting ecosystem. While companies have started exploring products for this market (e.g., Pivotal Living, which promised to launch a $12 wearable), their problematic launches indicate that considerable technical challenges remain.
For single adults, data sharing is largely focused on competition and casual socialization. Data reporting tools generally assume that self-trackers have complete autonomy over their data. These tools promote casual social sharing of data by making it easy to share short, compressed metrics. Showing a friend how many steps you’ve walked this week is simple, but reviewing and sharing detailed behavioral data and health metrics are largely unsolved design challenges.
Consider this from the perspective of a parent whose young children have wearable devices. Should a 10‑year-old child be wholly responsible for their data? Do parents have de facto access to a child’s comprehensive data sets? Through what interface? How does a parent or guardian manage a family’s collection of devices?
Answering the preceding questions is the first step in creating personal informatics tools that impact a far greater population, but that isn’t all that we need to do to successfully design for a broader audience. Inclusivity is not a checklist.
As this field develops, we must collectively answer these two questions:
How can we encourage adoption of personal informatics devices and tools?
How can we help everyone who adopts these tools benefit from them?
We’ve outlined a set of parameters that we believe are a good starting point, and shown some academic work that sets a good example for how to design in a more inclusive way. We haven’t exhaustively covered the important parameters in this space. Instead, we hope this sparks a discussion that will identify other key areas of opportunity (both those in which active research progress is being made and those that remain underexplored).
Identifying, exploring, and developing best practices for these design dimensions is not a task for any single researcher, design team, or project. It is instead a set of challenges to be considered across all facets of personal informatics research and design.