Response: The Data Driven Life

Summary:

Continuing with last week’s theme about personal informatics, this paper was heavily focused on learning about one’s self through personal tracking. The main argument of this paper was that in fields such as accounting or science, numbers mean everything. They allow us to see patterns and make predictions. Thus, the same reasoning can be applied to our personal lives as well by tracking numbers about what we do. Despite all the positives about personal tracking this paper lists, there were some drawbacks it mentioned briefly. Namely, that of crushing self-awareness due to “the numbers don't lie”. Overall however, the paper strongly implies that self-monitoring can have benefits in all areas of a persons’ life.

Response:

My initial response to the beginning of the paper was positive, as the arguments it made in regards to using numbers in other areas made sense. However, as it delved further into what it meant to do personal tracking, it became obvious to me this was not a simple task to do (and probably what its not as widely spread as the paper would have one to believe). Following this train of thought, I thought about last week’s paper 'Why Groupaware fails', and I think some of the lessons in that could apply here, particularly in regards to maintenance. Additionally, in tracking personal data, we ourselves become a collaborative system around our own data.

Response: Understanding my Data, Myself

Summary:

The purpose of this paper was to examine ways in which we view data about ourselves. The main result of asking this question was that as we grow individually, so does the data about ourselves and how we view it. In the paper, they describe this as a shift from the Discovery phase (where information is being discovered and there is an overall goal) to the Maintenance phase (where a goal in information seeking has been achieved and it is now a matter of maintaining it). Even with the phases, the paper also showed that our data is dynamic and we can switch in and out of phases constantly. This is shown by the study in which they noticed that different questions about our own data is asked at different times.

Response:

The examination of personal data in this paper was interesting, especially when framing it with ubicomp. However, while thinking about data itself, the question as to why would this need to be done becomes obvious to me. In my opinion, the visualization of the data itself becomes as important, if not more important than the phases themselves. Yes, collecting as much data as possible is important, but it is its visualization that can lead to shifts in the phases themselves. I would have liked to see a study ubicomp + visualization and the changes it can cause, perhaps to further support the discussion in this paper.

Response: Why Groupaware Fails

Summary:

The primary purpose of this paper was to highlight several major reasons (at least to the author), why groupaware software fails. The staple example it provides is that of an electronic calendar and the amount of work required for one group of users (those who schedule) vs. those who just view (those who simply view the schedule). From here, listed are problems such as additional work being required and design process failure due to non-realistic design (design is based upon users similar to designer and not real-world ones). Another important component of this paper was the distinctions made between Multi-user systems vs. Multi-user applications vs. Single-user Applications. With these distinctions, valuable points about organizational change, overall benefits and costs become apparent.

Response:

Overall, I thought this was a significant paper because it highlighted problems that STILL exist today. The staple electronic calendar problem is still a problem today. Although admittedly, Google Calendar has done a pretty good job for some of the problems.

To me however, the most important component of the paper were the distinctions between all the systems. Especially when framing the problems in that of a tabletop application, these problems are further enhanced in my view. In some regards, a tabletop application, with its inherently collaborative nature, can be all 3 of these at the same time, if not separately. Although I have not looked into it extensively, a more modern version of this paper could cover not just systems, but multi or single surface environments.

Response: Stage Model for Personal Informatics

Summary:

In this paper, the authors discuss about how to address problems that come across with people using Personal Informatics which are not fully understood or defined, and finding a way of setting up PI that patterns with what people typically do. To do so, they did a survey.

The survey focused on how difficult it is to collect personal data, what is the motivation for reflection, and what patterns were found during exploration of the information.

From the survey, they came out with different stages to PI: Preparation, Collection, Integration, Reflection, Action.

This staged-model improved on PI assessment; what current PI systems can improve on, and what future systems should have to be effective.

Reflection:

What I find interesting in this paper is that based on my personal experience with collecting personal data, the stages are quite real. However, some stages need not be linear. For example, when you are collecting data manually, you may actually skip integration and immediately do some reflection and action.

I believe each person will be able to learn something about their data in each stages so if we have PI systems that actually allow people to do each of the stages in a non-tedious way, people will get more from their personal data.

I also think that we should have a degree of automation and experience-sampling (manual) together in PI systems.

Response to Li's Stage Model for Personal Informatics

Problem: there is no comprehensive list of problems that users experience towards using personal informatics systems.

Motivation: memory is limited and PI allows a more true self reflection. But how can we step it up?

Approach:

They performed a survey:

· How difficult is it to collect personal information?

· What was the initial motivation to reflect on it?

· What patterns are found when exploring personal information?

Most important areas: Finance, journaling, exercise and health.

Reasons: curiosity, interest in data, discovery of new tools, suggestion and trigger event.

Stages:

· Preparation: motivation to collect personal information, what and how they will record it.

· Collection: observations and recording of the data.

· Integration: combining and putting together all the information.

· Reflection: observing the results of the data collected, short term or long term.

· Action: what they do after they see this data -> potential behavior changes.

Contributions:

(1) Problems across Personal Informatics tools.

(2) Model that improves diagnosis, assessment and prediction of PI systems.

(3) Recommendations about how to improve existing systems and build effective personal information systems.

Thoughts:

I actually found a lot of value in the different stages, mainly because I hadn’t thought of them before. These are all things that I did when I collected my data for my infovis class. The interesting is that I don’t think we’re necessarily aware that we do these stages, or at least I never noticed when I did them. The interesting question is “what data should be collected manually vs electronically?” and to what extent can we automate it? What stage do we actually learn more from? I actually think we do still gain insight from just collecting or integrating the data, and this happened to me in one of my visualizations.

Summary Response Week 4: Personal Informatics

CPSC 601.25 Week 4 Response Part 2

Papers

In this response I will discuss two papers relating to personal informatics, both of which present models which describe how users interact with the system.

Personal Informatics

In the past ten years a new class of application has arisen, personal informatics. As it has become easier to record data, the desire to process and analyze it has increased. The authors of both papers consider personal informatics to be a practice, not a tool, and if considered this way the practice is not old. Certain people, such as Ben Franklin, had the patience to record information pertaining to themselves for years and years and on. Technology changes the equation in two ways, making it easier (or even automatic) to record the data and improving the tools for analyzing it.

What? Why?

The underlying question behind personal informatics seems to be 'why?'. The stage based model paper discusses it only briefly, but suggests that reflection is the key. In a later work, the understand my data paper, they suggest that behavioral change is a key motivating factor. While creating models and workflows and interviewing users can shed some light on the tools people are using and their goals in using them, the cost benefit equation of personal informatics has not been completely decided. Does recording meals help people who are struggling to lose weight? Are people who struggle with conscientiousness, willpower or self control really likely to be diligent plodding collectors of data? It seems at first glance that a person who can't force themselves to abstain from pie cannot force themselves to record either. If the purpose of PI is to change behavior, then it can be evaluated like any other method in this role. So what then is PI for?

Self Experimentation

When the paper authors discuss reflection they make no real distinction between a user casually reflecting over events in a journal, old pictures and emails and those doing experiments and investigations. But the later is far more interesting then the former. For a user trying to determine why they can't sleep, a tool which tracks their sleep can provide a baseline while they experiment with other variables, such as caffeine, lighting and exercise.

Seeking Utility

While users do follow certain 'models' when they use PI tools, the models themselves might not be that important. Certainly users begin using a PI tool by evaluating it, recording data with it, and then reflecting on it ... but this is common sense and generalizing this doesn't solve a real problem. And categorizing the general types of PI tools as well is also interesting, but it doesn't solve the real utility. Perhaps users enjoy recording and visualizing data about themselves in it self but probably not most people. To extend PI applications into the mainstream they need to provide a more clear utility to actual users. PI researchers should push users to show what utility they really get out of these tools and derive ways to make them more useful.

A Staged-Based Model of Personal Informatics Systems and Understanding My Data, Myself

Summaries

In this post I'm covering from both of the two papers because they integrate quite closely, one covering more macro level questions relating to why people collect information about themselves and one covering more micro level questions relating to how people collect information about themselves.

In these papers the authors, Li, Day and Forlizzi query groups of people who practice "self tracking". In their discussions with these people they look for patterns in their motivations and their practices to see what commonalities exist. At the high level in "Understanding My Data, Myself" they look at the motivations for collecting this information and group the questions that inspire self tracking into 6 types, Status, History, Goals, Discrepancies, Context and Factors and further group these types of questions into two phases, maintenance and discovery.

In "A Stage-Based Model of Personal Informatics Systems" the same authors tackle lower level questions about how people collect self tracking information. They identify five stages preparation, collection, integration, reflection and action and look at the barriers to each of these stages. Overall they determine that barriers in each stage can impact those stages that follow, that the process is iterative and people will return to earlier stages that each stage can be either user or system driven which impacts both the users motivation and engagement and that the stages can either focus on uni-faceted data from a single source or  multi-faceted data from many sources.

The results and recommendations boil down to three main points, collect as much data as possible as early as possible and as unobtrusively as possible, use the systems to reduce the work for the person using the system but don't limit their options and ensure that the system is responsive to the needs of the person using the system.

Thoughts

The authors mention the bias in the people selected to interview, that they are all selected from self-tracking enthusiasts and wonder what the impact would be to see this applied to the general public. In some cases this would be interesting (although I wonder if the general public's response would be: "Why would you bother?" to the majority of their questions), however I think the larger flaw in their studies was the relative newness of the bulk of their participants. Many of their interviewees had only self tracked for a few weeks or months and only a few had long multi-year experiences.

On the one hand I don't know that this matters so much because the bulk of the "interesting" aspects of these systems are important in the Discovery phase. However I don't know that they really cover what is important to the people doing this type of activity for really long periods of time, especially in cases where health is not the motivating factor.

It is interesting to see in contrast to the article on "Beyond Total Capture" the pull towards total capture that the authors here have. It makes sense that in order to more intelligently support users of the system having data ready to reflection earlier in the process is a strength, but they don't really address the privacy and inconvenience aspects of this collection. It would be interesting to address (possibly with their more public study) the comfort level people have with self tracking particular types of information.