Last year I gave a 20-minute talk at EG (EG is short for Entertainment Gathering) titled “You Had Me at Bacon” about my self-experimentation. I described some of the things I’ve discovered by self-experimentation. Then I tried to say why it had been successful — why I had managed to discover such useful stuff. My conclusion is that my success came from the combination of four things: 1. Self-experimentation. Much faster, more flexible than ordinary research. 2. The Stone Age = good idea. I used the idea that our bodies were shaped to work well under Stone-Age conditions to choose what experiments to do. 3. Subject-matter knowledge. My knowledge of psychology, experimental design, and data analysis helped a lot. My weight-control theory, for example, was based on ideas from animal learning. 4. Freedom. I could do and say what I wanted. Most scientists cannot. They fear career damage. The combination of these four things is why my work was effective.
After my talk, a few people asked: Were you serious? No doubt you’ve heard Arthur Clarke’s maxim that “any sufficiently advanced technology is indistinguishable from magic.” Let me propose a related idea: Any sufficiently advanced science is indistinguishable from a joke.
Seth, you should write-up some kind of “how to” guide for other people who want pursue this type of self-experimentation. I know that much of the pertinent information is already on your blog, but it’s scattered among many different posts.
I think that for most people, statistical analysis will prove to be the most-difficult part of the process. I wonder what kind of concise guidelines can be created for people who don’t have time for a full-blown education in data analysis.
I agree with Alex.
I kind of have the question, “how does one start self experimentation?” But I feel the answer is, “start recording data.” Still, overcoming the initial inertia is challenging, and perhaps guidance would be helpful.
Good talk. I think the “were you serious” reactions were from the cognitive dissonance people experienced when they thought you were advocating animal fat as a healthful endeavor. I mean you may as well have been in a church telling people that God doesn’t exist. It’s easy to see why they would have a hard time reconciling the dogma that fat is bad with the concept that animal fat might be good. And the next slide showing the saturated fat vs country comparison would have served to only create more cognitive dissonance. The concept that self-experimentation can actually be a legitimate source of empirical study already goes against the grain–we’re led to believe that more is better when it comes to experiments–so the animal fat thing seems like it complicates the message. Maybe a little disclaimer like “I’m not your medical doctor, I’m not advocating that you change your diet, I’m just saying what worked for me” might help ease things. Even though the animal fat part wasn’t even the core message, I still think that explains some of the strange reactions (including the awkward crowd response when you bring up the pork fat slide).
Very interesting overview of your experiments from the last several years. I noted that you did not discuss the effect of butter on your cognitive abilities which you have been posting about for a long time. There was only so much time in that 20m talk; I will still impressed by the amount of information you were able to convey.
I wasn’t in the room, but from watching the video the laughter sounded “off.” My take was that it was derisive like, “This guy is crazy.” I’m not at all surprised to read that a few people asked if you were serious about this.
In particular, there was a large and sustained reaction from the audience when you displayed the slide about animal (pork) fat and sleep. You followed it up with another slide showing that more animal fat consumption is correlated with lower incidence of heart disease, but clearly they weren’t buying it. It goes against “settled” science as explained by the Lipid Hypothesis.
Between the work you are doing, Taubes, Eades, Kock (HealthCorrelator.blogspot), and several others, all of this settled science is becoming quite unsettled. Keep it up, please.
Alex, plotting your data is 95% of a good data analysis — that’s my concise guideline about this. And it’s easy to plot your data using R. I think statistics textbooks are misleading about this. They emphasize the hard parts of data analysis, which are much less useful than the easy parts.
I have a maths degree, and I’ve done a fair amount of statistics in business. I absolutely endorse Seth’s point to start by plotting your data. You can do this with graph paper and a pencil. If the eye can see an effect, you have an interesting result, something to build on. If the eye can’t see one, chances are that any effect is too small to be interesting. If I was going to take the next step and do some mathematical analysis, I would use the sign test – simple, obvious, and requiring few assumptions about the data.
I think this leads to another point about self-experimentation. If you are experimenting on other people, and with the aim of benefitting other people, you have ethical duties towards both populations, especially to be very certain that your findings are valid and not artifacts of the experiment design or data processing. Experimenting on yourself, you do not need such a high level of certainty, since the issue is ‘How does this work for me?’
I think that a lot of people would fail if you told them to take a .csv file and plot it in R.
There might be a much of people who could benefit from a step-by-step instruction to record their data on a piece of paper, enter their data into the computer and then plot it in R.