You call it "luck", but that's what I meant by people really don't understand the risks, I don't think it's luck at all - Doctors make lots of recommendations based on broad trends, but they don't really understand an individual person's risk.
I think it boils down entirely to pure individual chance. Someone may be a chain smoker and die naturally. That does not obviate the medical fact that smoking increases the chances of lung cancer. It is like rolling a die - you can estimate probabilities but that your dice will land two 6's are purely based on the event. Yet, done over a couple of thousand trials we definitely know it happens 1 per 36 tries on average.
In cases such as your MIL, maybe she's less prone to gastric inflammation. That simple fact reduces her chances of other serious ailments. Its a biological chain reaction after all. Cancer is strongly linked to chronic inflammation.
Isn't that like saying someone who won the lottery wasn't lucky, we just don't understand the individual person's chances to win? Like in some way it makes sense but for the purposes of actually deciding whether to recommend whether someone should buy a lottery ticket you can do quite well by looking at the broad trends.
No, winning the lottery is the definition of luck, the lotteries spend a lot of money making sure that's true. There's nothing a person could do (besides play more) that increases their chance of winning.
But I believe that long-lived people are more than just "lucky", there's some physiological difference (possibly coupled with environmental differences) that makes the difference between people that eat similar diets and have much different levels of health/longevity.
Extrapolating this line of thought: Let us say, two identical twins are brought up the exact same way. Can we be certain they will have exact same longevity?
Their genetic make up might be similar but there are trillions of stochastic processes happening every moment at the cellular, and each one is capable of a cascade effect in the long run. I think it would be reasonable to assume that scale of precise personalized prediction / recommendation will be a distant possibility, if ever it is even possible.
We might get capable of creating extremely good probabilistic models however, with less than a quarter of the modeling resources by quantifying the different parameters of life taken statistically (e.g. average calorific intake, medicine dependence, exercise etc). That might actually get very close to examining the uncertainty about health & longevity.
Yeah but the difference is we don't peer review large medical studies on successful startup strategies. We're talking about biomedical research and oncology, not "conventional wisdom"
The controls and reproducibility of studies on human nutrition may not be as rock solid as you think they are. The conventional wisdom on nutrition has changed radically since many doctors, including one in my direct family went through medical school.
There are also considerable distortions due to who is paying for the research.
