Paul Nixon's Blog |||

What to eat

1. Introduction

We all want to live forever, and part of our strategy inevitably acknowledges the role of nutrition in support of this goal. As in all matters, we look to science for answers.

Nice plate of food

However, science can’t give us answers. Granted, that sounds defeatist, but there are first principle reasons that this is the case. Here’s a test: what matters around nutrition have been settled with data and no longer provoke debate? I see none. It’s possible to stake out any position in the nutritional landscape and find abundant science’ to support you. Great news, except for that fact that folks that believe the complete opposite can make an equally (un)compelling case.

2. How science proves things

Science has a sure-fire method of determining causality - a well-designed experiment. It works like this

  1. Hold all potential causes of an outcome constant.
  2. Change only the possible cause of interest.
  3. Wait long enough and see whether the outcome changes.

So if, for example, we want to find out if coffee shortens or extends our life we would

  1. Get a large number of people together.
  2. Split them into two groups.
  3. Hold every other potential cause constant.
  4. One group drinks coffee, and the other does not.
  5. Wait 50 years and see what happens.


2. Two minor problems

The first problem is that there is an incalculable number of potential causes to try and hold constant. These factors include, but are not limited to, genetics, childhood, environment, life situation, amount and type of exercise, diet, alcohol intake, smoking, sleep, stress, self-esteem, interpersonal relationships and atmospheric pollutants.

There is no way to hold all of these factors constant. Of course, we can try and take some of them into account; for example, by factoring smoking into the calculations. But it’s impossible to lock everything down and eliminate all of the sources of bias and error.

Even if this were not a catastrophic problem - which it is - the influence of anything under test will takes decades to become material enough to be detected. Impossible squared.

3. Observational Studies

So, experiments are out, and we do the next best thing instead - observational, or epidemiological, studies. Observational studies are relatively cheap and easy to do and are quite useless.

Here’s how they work

  1. Take a bunch of people.
  2. Ask them what they’ve been eating for a few decades.
  3. Have a look at the outcomes.

It does not take a genius to see the flaws in this idea. It relies on the participants having extraordinary memories, reporting honestly, and skilfully aggregating variable behaviour across decades.

If we see any effect from these weak studies, there is still a big problem. This type of research can only show an association between a cause and an outcome and can only hint at cause and effect.

4. Healthy User Effect

There is a further curious and pernicious skewing effect in observational studies that acts to reinforce the status quo.

The conscientious among us live longer - they eat better, exercise more, drink less, play by the rules, have better relationships and generally live in ways conducive to longevity.

When you ask people about their past behaviours guess what happens? Any factor that has been considered good practise will come out looking better than it is because it will be inextricably packaged up with a range of other positive behaviours.

An example - let’s see if eating breakfast is good for us. Take a large number of people split them into two groups, those that have and those that have not regularly eaten breakfast. Without even seeing the data we know that those that eat breakfast will have better health outcomes.

But is it down to their virtuous morning habits? We’ll never know because that behaviour that we know is good for us comes with a life of other good choices that in aggregate produce a positive effect.

The opposite is also the case. We all know, deep down, that bacon must be good for us! However, science’ has been steering us away from it as a healthy breakfast option for decades. When we do observational studies on bacon, guess what, we find some small negative correlation with health outcomes.

But who’s been eating bacon for years? Not the conscientious that’s for sure, they have been eating their muesli instead. Anyone cavalier enough to regularly eat bacon will be more likely to have made a range of negative lifestyle choices - and in combination, these will be responsible for the effect we see in the study.

5. A thought and a strategy

While the science of nutrition is entirely hopeless, I don’t think the situation is.

A liberating thought: most of the things we eat don’t make much difference.

Does coffee cause cancer? Does beetroot make you live longer? Is bacon bad for me? Perhaps Betteridge should have had a law about nutritional questions. The answer, rounded to the nearest binary conclusion, is no.

Strategy: Avoid anything that wouldn’t have been in our ancestors’ environment.

Our long-distant past has shaped our physiology - this is the essence of the Paleo movement for which I have sympathy. However, I’ve not seen anything to persuade me that any particular era should serve as the model to guide us on what to eat.

However, we can invert the idea, and use it instead as a guide on what not to eat.

Would our ancestors have eaten industrially processed seed oils, sugar in the quantities we eat now, or highly-processed grains? Easy answer: no. Does this mean that these foods are necessarily bad for us? No easy answer: perhaps yes, perhaps no and, and with no clear answer, the sensible approach is to adopt the Precautionary Principle.

The Precautionary Principle tells us that when there is the potential for harm, and our scientific knowledge is incomplete, we should employ caution.

In conclusion,

Don’t sweat the small stuff and avoid anything your grandmother wouldn’t have eaten.