Seth Roberts died on April 26. He collapsed while hiking near his Berkeley home. The cause of death was occlusive coronary artery disease and cardiomegaly.
Seth was one of the leading figures in the ancestral health and quantified self movements, and one of the editors of the Journal of Evolution and Health. It is fitting, therefore, that the day following the Ancestral Health Symposium, there will be a memorial for Seth at UC Berkeley. The memorial was the idea of Nassim Nicholas Taleb and will feature talks by Taleb, Tucker Max, Tim Ferriss, Gary Taubes, and John Durant.
Many of his friends have offered tributes, and I can second their praise. Saul Sternberg: “I would judge him to be the most interesting and original thinker I know.” Aaron Blaisdell: “His warm companionship, unique intellect, insatiable curiosity, and infinite creativity will be missed.”
Seth was a special person – one of a kind. Few scholars can match his creativity. Anyone who loves ideas quickly became his fan.
Seth and I discovered each other at the same time – at the first Ancestral Health Symposium at UCLA in 2011. We began reading each others’ blogs, and began an occasional correspondence. When Shou-Ching and I did a revised edition of our book for Scribner in 2012, he contributed this blurb:
“The sanest overview of what to eat I have ever seen. If you are going to read only one thing on the subject, read this.”
It was characteristic of Seth that he sent me a private message, “I want you to know I really mean it.” I’ve just now noticed, and am flattered, that PHD is the only ancestral community blog on Seth’s blogroll.
He was a fascinating conversation partner, but generally demanded more information than he gave. Andrew Gelman in his reminiscence of Seth wrote: “Seth was always interested in what people had to say. His conversational style was to ask question after question after question after question.”
When he and I were together, he always steered the conversation to a favorite topic: “Why is modern science so unproductive?” Biomedical science transitioned to its modern form about 1950, when chronic diseases like cardiovascular disease, cancer, and diabetes became the major killers. Yet, after 60 years of effort and, currently, $30 billion per year in federally funded biomedical research and $50 billion per year in pharmaceutical industry R&D, medicine has failed to cure any major disease, and has generally failed to elucidate causes. At AHS 2013 in Atlanta, Seth and I sat together at lunch and ended up staying an extra hour, missing talks, to discuss why that has been so.
I suppose he chose different topics of discussion with different people, and chose this topic with me because I brought a broad perspective to the issue, having been a physicist at MIT, Berkeley, and Harvard; a vicarious participant in my wife’s biomedical research career; and a practitioner of personal science to fix my own health issues. I had also had practical business experience and knowledge in economics, so I had some ideas about what institutions and cultures enable work to get done effectively.
Seth and I were both fans of personal science, and were optimistic that the ancestral health and quantified self movements could generate scientific knowledge. However, Seth trusted and relied upon the personal science approach more than, perhaps, anyone ever has. Above all, Seth treasured his own experiments.
This was both strength and weakness. It made him extraordinarily creative. From the food-flavor appetite retraining method of the Shangri-La Diet, to his tactics for improving sleep and mood, to his methods for improving reaction time, he repeatedly discovered new influences upon health.
He was also a careful experimentalist. I found his data trustworthy – it had the ring of truth, and often led me down fruitful pathways. For example, Seth’s “faces therapy” introduced me to the role of social interactions in entraining circadian rhythms. (See my blog posts here and here for more about faces therapy, including links to Seth’s blog.) Here is some of Seth’s data, as I present it in a talk at the Perfect Health Retreat:
Notice that when Seth stops looking at human faces, his mood is worsened only on the second day, not the first. And then when he resumes, his mood does not improve on the first day, but only on the second.
To detect such an effect, and reproduce it, is a mark of a skilled experimentalist. Mark Twain once said, “We should be careful to get out of an experience only the wisdom that is in it and stop there lest we be like the cat that sits down on a hot stove lid. She will never sit down on a hot stove lid again and that is well but also she will never sit down on a cold one anymore.” Now imagine if burns from the hot stove afflicted Twain’s cat only after a two day delay. She might never make the association, and continue sitting on hot stoves for years.
Why the two-day delay? The obvious answer is that circadian rhythms in the body decay in 24 to 48 hours. And that is indeed the case. Here is another slide from that same talk at the Retreat:
The amplitude of the gene expression cycle in circadian genes is cut in half every day without stimulus. Within a few days, amplitudes are too small to fulfill clock functions.
The moral: We need to stimulate our circadian rhythms daily, or at worst every other day.
Reading Seth was like that. Almost every finding Seth made, sent me to the literature, and led to new discoveries. But the conclusions I drew from his experiments were often different from his own.
For Seth the lesson of his faces therapy was that he needed to look at human faces in the morning, and avoid them in the evening. A friend of Seth’s, Ellen Rosenthal, wrote, “It was actually quite hard to see him face to face. He had discovered that exposure to human company (he usually described it in terms only of the human face) was a harmful practice if carried on in the evening.”
For me, the life-changing lesson was that I needed to exercise – another circadian time-giver – every single day.
There was rarely a single, unequivocal interpretation of his experiments. If the Shangri-La Diet worked by breaking flavor-calorie associations in the brain, was it better to retrain the brain by eating flavorless calories, or flavorful non-calories? Or perhaps the diet needed to be more diverse and variable, and so less prone to triggering Pavlovian eating. Experiments to clarify what was going on were never arranged, so far as I know.
The weaknesses of Seth’s approach to science show up best, I think, in how he ate. Although he considered ours the “sanest” diet book, he didn’t eat our diet. He prized his own experimental results above all else. If an experiment persuaded him that eating something would improve his health, he ate it.
To my mind, this led him on a somewhat fanciful peregrination through dietary parameter space. His approach risked two pitfalls:
- In any complex optimization problem, a sequence of small improvements does not lead to a global optimum, for the same reason that climbing ever upwards does not lead to the top of the world’s highest peak. Odds are, it leads to the top of a small hillock from which there is nowhere to go. Without a larger view of the landscape, enabling one to find the vicinity of the highest peak, there is no way to guarantee a good outcome through a series of experiments.
- Even a local optimum – the top of our small hillock – may be unreachable if one cannot accurately assess health. Seth’s experiments typically lasted days to weeks. But most modern health problems take 60 years to develop. So there was no way for Seth to directly appraise whether his diet would generate good health or poor health; he had to appraise its effect on readily measurable, quick-adjusting biomarkers. But how do we know those biomarkers are reliable indicators of overall health?
It is because of these two problems that our book, Perfect Health Diet, rejected experimental approaches to dietary science, and relied upon novel approaches grounded in evolutionary biology, and molecular and cellular biology.
But Seth was wedded to experimentation as a scientific methodology. This worked well as long as he was using sleep quality as a biomarker, since sleep quality is close to 100% correlated with health. He entered riskier ground, I think, when he selected reaction time as a biomarker to optimize. I doubt this has a simple relationship to health; I suspect one can improve reaction time while damaging health. And when optimizing this biomarker led him to consume large amounts of butter on top of large amounts of flaxseed oil, I think he should have recalled the arguments of our book, and been more persuaded by them than he was.
We never had a chance to discuss this issue. There were always topics other than diet to interest us when we were together.
Ironically, had he lived we might have discussed this by now. Seth and Bryan Davis of the Ask Bryan Podcast were launching a podcast devoted to stories of personal science, and had selected me to be their first guest. The day after Seth’s death, Bryan wrote, “Did you record that interview? I hope so. I sincerely hope so. Perhaps the final recorded words from Seth?” Alas, it will never be.
Seth had as much influence on this blog as any one. A Google search of this blog for his name yields 741 hits, more than almost any one else.
Seth was – and this is the highest praise I can give – a true scientist. He loved the truth, and worked ardently to discover it. He was creative and insightful. His death is a grievous loss.


















Recent Comments