2-node-supercomputer.net 2021-10-29

Optimize for Signal

Today I was listening to a podcast about a meat-based diet. In the podcast, the narrator describes how she went from a completely plant-based diet to a completely meat-based diet.

While on the plant-based diet she was suffering from severe bouts of depression, binge-feeding herself at night with all sorts of processed (but plant-based) oils and grains. It got so severe that she could not take care of her son.

The problem was that her diet was missing essential ingredients that are hard to find in plants. Because of that she was miserable.

Any normal person would have given in to their cravings much earlier, or in some other way recognize that they were doing something wrong. In the US, it takes effort to avoid meat.

Yet, here she is talking on a podcast, making money. And I get the feeling this is not an isolated instance. It appears to be a fairly typical story: you overdo something, then sell the solution, which might be in the opposite extreme. I feel like I have heard such stories over and over again. The podcast I mentioned is just the latest example.

What is happening here? The hypothesis I will put forward here is the following. By not stopping their craziness early on, those people are optimizing for signal. A strong signal that something is wrong is easy to recognize. It is also easy to recognize when it is fixed. Never mind that it was foolish to go there in the first place, and most people wouldn’t anyway. But the signal is clear: a plants-only diet, for example, can easily make you miss important nutrients.

Another example comes from another podcast where a neurologist (or maybe he is a psychologist?) was interviewed. The neurologist was studying how, say, you can push yourself further with psychological tricks. For example, when running you can probably run further even as you are getting tired and wanting to give up. That is great and very motivating! On the other hand, I have sometimes gone further (say, play another round of basketball) despite feeling tired. And so I will assert that pushing yourself too far is bullshit: I promptly injured myself. Group pressure is a bitch. Indeed, the interview with the neurologist is full of such examples where you have to wonder if the subjects actually sustained some level of damage or exposed themselves to unnecessary risk.

In all cases I believe this is the problem: Because the person was foolish, the conclusion is pretty irrelevant to most people.

Yet, you hear again and again of those people that went too far. They didn’t stop when the signal was there but perhaps not clear. So my fatal conclusion is that to become famous, you should optimize for signal, not truth!