A single counterexample can prove that something is wrong with a model, but even a large number of data points that fit the tenets of a model can never confirm a model.
A basic asymmetry in empirical science that we just have to live with.
Consider the facts about human longevity:
- Generally women live longer than men. When heart disease is finally conquered, this may no longer be true.
- Heavy smoking reduces lifespan on average more than 10 years.
- Heavy drinking reduces lifespan, but we do not have a good number of years, in part because we do not have good data on how much individuals are actually drinking. I suspect that the average person drinks a lot more than he owns up to.
Consider two people of predominantly Irish descent (i.e. very similar genetic background):
- The man – his father lived to only 45 years and died of heart failure and his mother lived to only 61 years and died of a stroke. A hyper-stressed, type A personality, the man smoked 3 packs a day of cigarettes every day for 45 years, and then gave it up cold turkey, and drank about 6 alcoholic beverages a day for more than 50 years, and then gave up drinking cold turkey. Enormous willpower. Reasonable fit and trim -never overweight – his whole life. Exercised moderately and occasionally and slept well. His diet was pretty much the Standard American Diet, but with less processed and junk food, more meat, and fewer sweets.
- The woman – her father lived to his early 80s and died of natural causes and her mother lived to her mid 90s and died of natural causes. She never drank and never smoked, her sleep was of average quality, and though she did not exercise much, she was just modestly overweight since having her third child. Her diet was pretty much like his, the Standard American Diet, with less processed and junk food, but heavier on the comfort food carbs and lighter on the fat and the protein.
In view of the above facts on lifespan, we would predict that the woman would live much longer than the man, not only because of her parents’ much longer lifespans, but the absence of heavy drinking and heavy smoking.
The facts: the woman lived 78 years and died of the complications of cancer in very frail health, and the man lived 87 years and died of the complications of cancer, and even just days before death, his doctor said his vital signs were strong, meaning he could have held on longer, if he accepted treatments.
Does this make any sense? No – not a jot of sense. We are clearly missing something in our model of lifespan. Human lifespan is a function of hidden variables, including a least one that is likely non-quantitative, one that science has simply ignored because it cannot quantify it. Science currently has a very poor model of lifespan, as these killer counterexamples show.