[miku86]

As a doctor, your primary job is to make sick people healthy, not to make healthy people more healthy.

This is why you want to have a high compliance, meaning that the sick person does what the doctor recommends. This is mostly caused due technical or behavioral complexity.

You can increase compliance by reducing complexity. At the same time, the accuracy decreases.

Example: High Accuracy / High Complexity: "Do not eat frugugle, fergerio, flululu and fnyvoo." Why low compliance? Because the behavioral complexity is very high, especially for sick people: 1. Find the food ingredients list. 2. Loop over the ingredients list and compare the current item with the "do not eat"-list. 3. Make a decision for every ingredient, if this is in the list ("frugugle" is on the list, but one ingredient is "fruguglelase", what's my decision?)

Lower Accuracy / Lower Complexity: "Do not eat products with more than 3 ingredients." Why higher compliance? Because the behavioral complexity is lower: 1. Find the food ingredients list. 2. Count the amount of ingredients. 3. Make a decision after reading max. 3 ingredients.

IMHO it's easier to start with the second approach, because you make progress faster, and keep the momentum going.

[lordlic]

This whole argument strikes me as very silly. If I want to eat healthy foods badly enough to change my diet and check the ingredients every time I buy something, I'm not going to want to oversimplify it to the point of near meaninglessness by adopting a rule like "3 ingredients or fewer." I'm perfectly capable of learning what ingredients to avoid.

And if I'm sick, and there are clinical details that are important to my treatment, then I'm absolutely not going to be satisfied with a low-accuracy simplification.