1.2 Quantifying the anecdotal

If the results of a treatment are not recorded, we still have reference values. People still rely on word of mouth — anecdotes — when looking for new treatments. But those reference values are anecdotal. You regularly hear stories of the form “I tried X and it worked for me”. Hear enough of those stories and you may want to try it yourself. But how many of those stories constitutes “enough” to try for yourself?

What if there were a common way for everyone who tries X to record their results quantitatively?

That’s the idea behind symptom tracking, and it’s a nice start. Some companies try to add fancy additional features on top, like using machine learning to try to guess better than you can alone about the various correlations found within your data. Many companies go this direction — gather enough data, either from yourself or from others, so that we can predict the causes for various states. Again, that’s interesting and it’s a nice start, but it’s limited.

What you really want — and the key, original idea behind Personal Science — is to let you take that quantitative data and compare it to others: others like you, people who you consider to be just like you except for such-and-such symptom.

Now, in some cases, a symptom tracking or quantified self product will let you see yourself compared to an aggregated summary of all other users. Fitbit might let you compare yourself to all those of your sex or age, for example, or maybe those in your geography. This is a good start.

But what if you could choose your own subset of users with whom you want to compare yourself? Because only you know which type of person you identify with, or to which type of health condition you want to belong, Personal Science lets you analyze and study the data as a whole.

That’s why it’s personal – it’s about the one, unique data point that is you – and why it’s science – democratize the quantitative tools of science to let you understand your condition, for yourself.


The term “personal science” was first popularized by the late Seth Roberts, an Emeritus Professor in the Psychology at University of California, Berkeley. His best-selling book1 and popular blog2 insisted that much of modern science is too complicated for its own good, that interesting and practical results are often best achieved through personal experimentation. Through multiple examples from his own self-experiments, he used his own data to show non-obvious treatments for better sleep (skip breakfast), lower depression (faces in the morning), and many other situations.3


Most of the examples in this book are based on over 600 near-daily samples I took of my own microbiome over a three year period. Inspired by an experiment conducted at MIT4, during most of that time I also carefully tracked the food I ate, my sleep, and other variables like activity and location. Most of my near-daily samples were of my gut, but I also regularly tested my skin, nose, and mouth. Since I’m generally healthy, I didn’t have a specific goal in mind other than to try to understand better what these microbes are doing, so many of my tests were taken while undergoing simple experiments, like eating a specific type of food or traveling to a new place. While not necessarily up to the rigorous standards of a formal scientific trial, these “n of 1” studies on myself helped me discover several new interesting facts about my own microbiome, many of which appear to contradict other published studies. In addition, hundreds of people sent me their own test results, letting me compare many different microbiomes. And of course, I also followed the latest developments in scientific publications and the general press as I eagerly tried to learn more.

This book tells you what I learned.


  1. Roberts (2007)

  2. His blog, active until his death in 2014, is actively discussed on a Facebook Community: https://sethroberts.net/2016/01/13/seth-roberts-community-on-facebook/

  3. Roberts (2004)

  4. David, Materna, et al. (2014)