Seth Roberts on Personal Science, at Quantified Self

This post is part of my liveblogged account of a conference. Two disclaimers: Liveblogging is hard, and I often get things wrong. If I did, please feel free to correct me via email or in the comments and I’ll make changes when appropriate. Second, the opinions expressed in these sorts of posts are those of the speakers, rather than mine.

I’m here at the inaugural Quantified Self conference at the Computer History Museum in Mountain View, CA. While I’m doing less conference blogging these days than in years past, I thought I’d try reviving my practice for this event, because I’m intrigued by this emerging movement and what it might mean, not just for issues of health and body awareness, but for questions of tracking other aspects of our lives, like what we hear and read and who we talk with.

Journalist Gary Wolf is our co-host at the conference (along with Kevin Kelly), and one of the main conveners of a group of people who are trying to use data to better understand and optimize their behavior. Wolf tells us he’s particularly interested in habit formation and relapse – he shows us a graph of information on his personal meditation practice, and calls our attention to broad black lines: days where his well-established practice breaks down.

Gary explains that the people who are excited about quantified self (QS) at this point are generally people who aren’t scared by raw data… and the conference itself is intended to be somewhat rough and raw. With 400 attendees, there are over 100 projects being presented, which suggests that one out of four people – at minimum – has an active project to share. The schedule is optimized to let people connect to existing projects and start good conversations… even at the expense of actually attending all the sessions.

What is a quantified self? There’s lots of big data projects in the world, and not a lot of people are working on the self. The self is not a sideshow in the world of data – “it’s the main thing, it’s the center ring”. Computing, historically, is designed to help scientists and engineers solve math problems. These days, this isn’t how we think about computing – they’re about creativity, self expression, information. It’s personal computing. And we got here through advanced users, who teach technologies to do new things.

Advanced users are trying to solve problems that are important to everybody. Technology can connect us to people very different from us who are trying to make the same discoveries – how to eat, how to sleep, how to learn, how to work. By discovering goals we share in common, there’s the possibility that common ground on the quantified self leads us to empathy.

QS is a social business, a movement, a set of meetups around the world. In 2007, Wolf and Kelly developed a very simple protocol: what did you do, how did you do it and what did you learn? The goal today is to explore those three questions in as wide a way as possible, in the sessions and on thirty posters.

Seth Roberts is a major figure in the Bay Area QS community, and towards the general topic of trying to understand ourselves through self-experimentation. He’s an emeritus professor of psychology at UC Berkeley and now teaches at Tsinghua University. His topic, “Why Does Personal Science Matter?” starts with a personal story.

In graduate school, Roberts began to experiment on his acne. His dermatologist prescribed two medicines, and Roberts thought that one worked and the other didn’t. So he conducted an experiment and discovered he was wrong: only one worked, but it was the other one. But the doctor was wrong too: one of the medicines didn’t work. How had he discovered something in a few weeks that a trained and experienced doctor didn’t know? This question led him down a path of self experimentation on other health issues, and he kept making discoveries that experts seem to have missed.

Roberts tells us that medical science is in a state of stagnation. The main sources of disability – heart disease, diabetes, obesity, cancer – and the fields where Nobel Prizes are awarded, have very little overlap. This suggests that scientists have had trouble making progress on major diseases.

One sign of stagnation is our reliance on old treatments. See a doctor about obesity, and they’ll tell you to eat less and move more, which is 50 year old advice. Low carb diets were introduced as early as 1864, and low fat became popular in the 1960s. With depression, it’s the same story. SSRIs like Prozac were introduced in 1986, and cognitive behavioral therapy was introduced over half a century ago. Statins have made billions of dollars for drug companies, but there’s no clear benefit to them, and serious side effects. Autism is rising sharply over the past thirty years, and there’s little good thinking about the causes. The advice we get from experts of sleep is basically advice we could have gotten from the ancient Greeks – go to bed at a reasonable time, take a bath before bed. This sort of stagnation helps explain why Americans pay five times as much on healthcare than people in other rich companies, but we don’t live any longer.

Personal science is about using scientific methods to solve your own problems. You can gather data from yourself, or from others, and the Internet is making it vastly easier to share this sort of data and learn from it. And we’re getting great new tools through engineering, scholarship and from basic science.

The advent of the personal blood glucose meter is an example of how engineering can help. Engineer Richard Bernstein had poorly controlled diabetes, In 1969, he bought a glucometer designed for hospital emergency rooms and began to use it to track his glucose closely. His results were so good, in the 1980s, doctors began to adopt the devices and related methods. Glucose tracking is down a billion-dollar industry, the main standard of care.

The ketogenic (high-fat) diet for tempering childhood epilepsy is the result of personal exploration through scholarship. Movie producer Jim Abrahams had a son who suffered from debilitating epilepsy. He dug deep into the literature and discovered that a diet was developed in the 1920s to mimic the affects of starvation, which scholars had discovered appeared to counter epilepsy. The diet, updated to be usable today, now helps 2/3rds of the children who try it. 1/3rd who adopt it become seizure free.

Dennis Mangan, a blood bank technician, was worried that his mother was suffering from restless leg syndrome. Many people do – a study in Italy showed that 11% of Italians 50-89 years old had the condition. Reading alternative medical sites, he saw advice about taking high dose niacin. His mother tried it and her thirty-year problem went away almost overnight. The information wasn’t on an “authoritative” site – it took work to discover it.

Much of Roberts’s work falls into the category of “basic science”, discovering new cause-effect relationships by running experiments on himself, relying on the large sets of data he’s collected through past self-monitoring. During a visit to Paris, he found he had mysteriously lost his appetite. His theory of weight control suggested it was due to the soft drinks he’d been drinking due to the heat. He started an experiment with sugar water, and lost thirty pounds in three months due to loss of appetite, shedding weight so quickly that colleagues began to express concerns for his health. He turned his observations into a book called “The Shangri-La Diet”.

He shows us a ten year data set from Alex Chernavksy, who monitored his weight and his struggles to find a workable diet. He lost weight through a low-carb diet, but regained some of it. He shed weight through long walks, but found the time commitments too constraining. Eventually, he Shangri-La Diet (drinking flaxseed oil between meals) helped him shed and lose weight. Roberts’s point is not the superiority of this diet – it’s the utility of having ten-year long data sets, which have features not found in data from clinical trials. The data isn’t artificial – it’s the experiences of a person trying his best to find workable approaches to a problem and comparing results.

After solving his acne, Roberts tried to focus on improving his sleep. One experiment involved watching videotape of human faces shortly after he’d woken up in the morning. He tried a simple experiment and was startled by how good he felt the following morning after watching these faces. Using experiments and tracking his mood, he concluded that a circadian oscillator controls mood and the faces “push” this oscillator, causing it to oscillate (raising mood during the day, lowering it at night).

Now he’s trying to optimize his sleep through a particular form of standing. “I discovered I can sleep much better when I stand nine hours a day.” But that’s not always practical. So instead, he stands on one bent leg, to exhaustion, six times a day.

Most recently, he’s been experimenting with expanding his cognitive abilities. He tracks his performance by solving arithmetic problems within a fixed time period – he’s got a long baseline of this data. After eating some pork from a boutique farmer, he found himself wondering what to do with a part of the animal which was virtually pure fat. Finally, he brought himself to eat it… and discovered he slept very well and was full of energy the next day. So he began eating pork fat everyday, though he found this harder to do in San Francisco than in Beijing. So he tried butter instead, and discovered that the butter had an effect the pork fat did not: His ability to perform arithmetic quickly improved sharply.

After presenting these results at a QS meetup, colleagues were interested in replicating the effects. 40 people participated in an experiment where they ate extra butter or coconut oil for a week, or made no dietary changes. The butter eaters saw an average 5% increase in their speed compared to the other two groups. A cardiologist at the meeting advised Roberts that he might be getting smarter, but was surely killing himself. “Fortunately, I happened to have a heart scan, which I’d had done for self-tracking reasons.” The scan measures the plaque burden on the heart, and the scores very rarely improve. But by adding large amounts of saturated fat to the diet, Roberts’s scores dropped from 38 to 29 in a year.

Useful discovery, Roberts tells us, is a function of material resources, knowledge, time, freedom and the desire to be useful. Freedom is the ability to try lots of different paths seeking a solution. Desire to be useful is the impulse to find a practical, usable change in behavior, exactly what sufferers of chronic diseases are searching for. Professional scientists have lots of resources and knowledge, Roberts tells us. But they’re very low on freedom, as they don’t want colleagues thinking they’re nuts. They don’t have enough time – grants are two or three years, not ten or twenty. And practical knowledge often has less prestige in the academy than “pure” science.

Personal scientists have increasing material resources, because sensors and tracking technologies are getting cheaper. Knowledge is improving, in no small part due to the internet. And they’ve got all the time in the world, plenty of motivation towards practical knowledge, and are often less constrained by the need to try the conventional.

Rituals, Roberts warns, can interfere with understanding. Rain dances don’t actually cause rain, but they’re very reassuring for those who participate in and witness them. Randomized double-blind placebo-controlled clinical trials are done mostly because they’re reassuring. He urges us not to waste our time on rituals. “Personal scientists are more likely to make useful discoveries than professional scientists.” We’re the first wave, he says, and we’re luck to be part of this because in discovery, it’s the first wave that gets the glory.