My answer to the Quantifed Self Forum question Why do self-trackers build their own aggregator solutions?
To many of us addicted to self-tracking, the appeal is that we hope to find some non-obvious fact about ourselves that we can adjust to make for a happier, healthier life.
We are full of questions about ourselves: am I eating as well as I could be? am I getting enough sleep — and if too little, how can I sleep better? Is there something I should change about myself to reduce my suffering from [migraines | allergies | colds | indigestion | etc] ? Scientific, data-driven approaches work well in many other domains. Why not apply the same thinking to ourselves?
One rite of passage into the QS world is developing your own software solution. Just about all the serious self-trackers end up building their own tool eventually, whether it’s a complicated Excel spreadsheet or a full-blown web site. People often start informally: a pad of paper near the bed to keep track of dreams, a tower of empty Mountain Dew cans, notch marks in the bed. As ambition or necessity require, some will begin to make entries into a spreadsheet.
Eventually, anyone with a few software skills will end up writing their own data aggregator. This tendency is so common — and predictable — as to deserve its own law. (How about Wolf’s Law, after Quantified Self co-founder Gary Wolf, whose tireless organizing on behalf of the movement continue to birth new generations of self-tracking addicts?)
The reasons seem to range from (1) Missing features in another stand-alone solution, (2) lack of integration with a new input device, (3) privacy concerns related to opening your data to a third party, and (4) desire to try out some programming skills on a challenge of personal applicability.
Eventually, the tracking itself becomes its own end. What began as an attempt to find the factors that improve sleep turns into an ever-expanding series of new data types along with tools to contain and study them. Each additional sip of data leads to another, and another until the variety and quantity become overwhelming — which leads to an investment in additional analysis. It’s easy to forget why you began the self-tracking journey in the first place.
Then there’s sheer momentum. If you’ve collected 97 days of sleep data, do you really want to call it quits before you reach 100? In the absence of any non-obvious discoveries, the tedious frustration of trying to find actionable information beats the relatively minor effort of collecting just a little more data.
Meanwhile, no matter how much you collect or analyze, you’ll never match the incredible thoroughness of your fellow amateurs Gwern, Quantified Bob, and many others, not to mention the academics to whom this has been a full-time job for decades: Larry Smarr, Michael Synder, and on and on.
At some point you, or perhaps a concerned family member, will want to know why you’ve invested so much time and energy into this project, and you’ll need to offer some results. What can you say you’ve learned from the endeavor?
I’ve seen hundreds of Quantified Self testimonials, those short presentations of what-you-did, how-you-did-it, what-did-you-learn that are a hallmark of QS Meetups. I find all of them engaging, even inspiring, and offhand I can name several that were especially clever or original. But what can I name that is truly up to the high expectations we have when we invest so much effort?
I cough less on days when I don’t smoke.
I lose weight when I eat less.
Headaches are more likely on mornings following a night of heavy drinking.
Okay, I made up these examples, but you get my drift and you can certainly summarize the vast majority of QS experiments in similarly blunt fashion.
There are many systematic “experiments” that really do work. An elimination diet, for example, is an effective way to expose foods that cause trouble. But all the examples I can think of are, well, pretty obvious. You didn’t need a double-blind randomly controlled trial to tell if you are allergic to peanuts. You’ll know.
It’s nearly always the case that if an effect is real, it will be obvious. If you need fancy statistical tools to tell whether something works, chances are it’s not worth the trouble.