Nehaveigur

We want to know why. My career is built around finding the causes for rare diseases. Human genetics, the field I trained and work in, is well suited for this. Germline mutations cause traits and diseases, but never the other way round, which means that human genetics can disentangle correlation and causation.

It’s satisfying to find out why, but there are also practical reasons to acquire this knowledge: The cause you have identified could also be a lever. The hope is that it points to an intervention that will help us to determine which gene or protein needs to be targeted to cure the disease.

Other fields aren’t so lucky. For them, teasing causation and correlation apart is harder. Take epidemiology for example. Smokers are less likely to get Parkinson’s, but it’s not clear why. Maybe smoking protects them for getting the disease, maybe it’s something else. Maybe it’s easier for people with Parkinson’s to quit. Maybe there are confounding factors like lifestyle. Maybe people who smoke a lot die from other diseases before they can get diagnosed with Parkinson’s. Other unexpected findings from epidemiology not necessarily pointing to causation are that eating a lot of ice cream is associated with a lower risk for cardiovascular disease and that drinking coffee is associated with higher life expectancy. For none of those we know what’s cause and what’s effect.

One challenge when looking for causes is that there may be none. This is hard to accept.

Example: A surprising number of metrics such as income growth and energy consumption hit an inflection point around 1971. Why? Nobody knows, but there are plenty of theories out there. I’m skeptical that we’ll ever find an explanation.

Strong, persistent associations can arise from feedback loops. Fisherian runaway selection is an example: Traits that are preferred by the female of a species for almost random reasons can, over many generations, get strongly selected and increase in frequency and strength. That’s even if they don’t contribute to survival or even threaten it.

It may in principle be possible to pinpoint the causes that led to a feedback loop powered runaway process. Unfortunately it’s not likely to be particularly useful as it doesn’t point to any levers that we can pull to manipulate the outcome. That’s because the runaway process, by definition, will persist even if the initial trigger is removed.

My guess is that the Great Inflection of 1971 falls into the same category. It was likely due to a series of societal feedback loops, triggered by not particularly informative or actionable events that nobody will ever convincingly be able to identify.