Un-Quantifiable

A little over four years ago I started down the path of quantified self, measuring sleep patterns and caloric intake. A year later I was fortunate enough to upgrade to a phone that included the ability to track steps and, with the help of other applications, other measurements such as heart rate were tracked multiple times a day. This stats collection has been a regular component of each and every day up until last night when I decided to bring it all to a stop, delete the data from the phone, and walk away from the practice. This isn't out of laziness or the added complexity of collecting data when a newborn is screaming like a banshee at 3:30am, but a niggling question that has remained unsatisfactorily answered since 2014: What problem am I trying to solve?

I love numbers and charts and turning data into actionable information. There isn't much that I'm particularly good at in life, but data processing is where I excel. One would think that analysis from all this self-observation would be sufficient reward in and of itself. However, looking at the statistics, this is what a person can surmise about me:

  • I'm getting older
  • I sleep better on a clear day after a bit of alcohol and a shower
  • I don't get very good sleep during the workweek
  • I consume more "unhealthy" calories during the workweek
  • My body weight alternates between 78.2kg and 93.8kg every sixteen months
  • My heart rate is a steady 52bpm at rest
  • I generally walk 7,200 steps on a working day, and half as many on weekends
  • Anxiety keeps me up at night

Just over four years of data shows all of this to be true … but did I really need to make tens of thousands of personal measurements across 1500+ days for these insights? Aside from the heart rate data, everything else is pretty much observable without fancy hardware or data-collecting software. More than this, half of these numbers are outright inaccurate due to the problems with the very tools I use to collect this data. So, if the data is unreliable, is any of it actionable?

First there's the problem with measuring caloric intake. No two people are alike, and my body likely processes calories different from your body. You and I could eat the very same quantities of a food, but absorb a different amount of energy, too. The calorie count on packaged foods are inaccurate as well, being an average number for a product, rather than what's actually in the package. Then there's the problem of accurately measuring foods that you prepare at home. If I have a banana that was grown on a farm on the western coast of the Philippines, does it have the same caloric value as a "standard banana"? Many people I've spoken to about this generally say "who cares? It's just a tiny percentage different!", but those tiny percentages add up very quickly. Then there's the other argument that calories mean little, and it's nutrition that counts.

So … calories are a poor measure of one's intake. Time for me to ditch it and not feel guilty for the occasional unhealthy snack that I enjoy for reasons that have nothing to do with fuelling the body.

How about measuring my heart rate? Over the last few years I've seen it get slower by about one beat per year. From what a lot of health sites online say1, this is completely normal as we age. What's more concerning to me isn't the speed of the heart beat, but the rhythm during times of acute stress. It's been really weird for quite some time, and I occasionally need to lean against a wall or sit down for a few minutes until things return to the regular pattern. My cholesterol is very low. My blood pressure is very high. Maybe these are the numbers I should be paying attention to.

So heart rate makes no sense. Okay, let's ditch that. But how about the number of steps taken every day?

Yesterday I managed to power-walk an extra 45 minutes above my regular route every day, and I climbed an extra dozen flights of stairs because the elevator at the office was incredibly busy. What did my phone report? I'd walked 2,000 steps less than the day before and climbed zero flights of stairs despite doing an average of 45 every work day for the last year. When I pace at home with a baby in my arms, that counts as zero steps. When I'm cooking in the kitchen and going back and forth between the stove, counter, and fridge, that's zero steps. When I get up at the office for a cup of coffee, trek to my locker for a coffee pouch, walk to the shared kitchen for the hot water, then walk back to my desk, that's fewer than two dozen steps despite the 50 meters travelled. The numbers reported by the phone2 are wholly inaccurate and untrustworthy, therefore cannot be used in any decision-making process. If I wanted generalizations, I'd measure distances and calculate steps based on the length of my stride.

No point using the pedometer function on the phone, then.

And then there's sleep. This was the very first measurement I started keeping track of because I was sleeping so very poorly in 2013. The reason turned out to be work and home-related stress, yet I continued to measure my sleep because the application3 doubled as an alarm clock and would measure my heart rate first thing in the morning. Simplicity was key, and this was a very simple way to collect data while unconscious. Looking at the stats, though, there is nothing unexpected. I still struggle with stress and anxiety. I still lay awake in bed until 2:30am two weeks every month. Do I need a constant reminder that my dog is infinitely superior at sleeping than I?

Perhaps it's time to ditch measuring the sleep and focus on actually sleeping?

Looking at the Future

A lot of people have become quite enamoured with the idea of quantifying their daily activities in order to lead a more conscientious lifestyle. I wouldn't go so far as to say the concept of Quantified Self allows people to lead a healthier lifestyle, because I don't believe this is necessarily true. Health is deeply personal, and the cold algorithms we currently use to "gamify" and otherwise motivate ourselves fail to take a great number of important details into consideration. Our psychological well-being intrinsically tied to our physiological well-being, yet so many applications that try to tie these two disparate metrics together do so in an arbitrary and ultimately inconsistent manner, which leads one to ask whether the information collected can actually be used to make better decisions.

Validity of data aside, the other problem that people are bound to face going forward is the matter of privacy. So much of the data we collect about ourselves is stored on 3rd-party servers by organizations who are eager to pay the bills. While we can see visualizations of our data, not every application gives us the ability to export our data in a usable format for analysis somewhere else. To make matters worse, some services will not simply give people their data but instead charge them for the privilege of downloading their own data in an unwieldy CSV format. We are paying companies to get our own data back. Even if this is explained in 100pt font in the Terms of Service when signing up for the service, the practice strikes me as terribly wrong. Then comes the question of whether that company sells your data to other companies — and you know they do — and what measures are in place to protect your privacy. We cannot go a week without hearing of a high-profile hack or data leak. It only makes sense to never share data with a health and fitness company that we wouldn't also share with the CEO of our employer or the attendant at the gas station down the street. This isn't always realistic, though.

Looking just a few years into the future, the entire Quantified Self industry looks set to undergo a huge revolution. One with millions (or billions) of people taking part in an attempt to lead healthier, happier lives. While this is certainly a noble goal, there should be a few pre-requisites beforehand. Any serious Quantified Self project should:

  • operate in complete isolation, free of corporate interference or surveillance
  • make use of incredibly accurate, dependable measurement tools
  • be personalized to take into account our lifestyle, genetic makeup, and heredity
  • take into account that people are not machines, and that we all like to do "less ideal" things from time to time

So while the last four years of data collection has not been a waste of time or energy, it has revealed that there is still a long way to go in our understanding of what needs to be measured as well as the appreciation of what makes us all different. In time these problems will likely be overcome as very smart people make very smart systems but, until such a time, we are ultimately unquantifiable.


  1. Yes, we should never read health sites online for all the incorrect and inaccurate information, but it's hard to justify making a doctor's appointment just to ask asinine questions

  2. an iPhone 6S for those who are interested

  3. I used SleepCycle exclusively for the entire time I measured my sleep