It's interesting how insightful you are -- boom! easily refuting years of medical research and publications without any statistical or public health training. I think you should now turn your powerful skills to other scientific endeavors now that the realm of statistics is proven fallacy. The Flat Earth Society is always taking on comers. Next -- fossil proof of unicorns. Awesome.
Statistical inference and scientific analysis is based on the study of specific populations. Discussions of an individual are nothing more than anecdotal and offer no wider application. I know it's difficult -- but it's SCIENCE.
I am starting to think you are not discussing this in good faith, but just in case you are I will try to express myself one more time.
BMI is a result of scientists studying the relationship that excess body fat has on health. For purposes of studying populations, they developed a scale using the relationship between height and weight (rather than body fat percentage), because everyone knows their height and weight. The BMI can often reliably predict that people with elevated BMIs are at high risk for obesity-related negative health outcomes. This is science and is useful for studying and projecting health outcomes for populations.
If a random person on the internet tells you they have a BMI of, say 27, a person like Northwestie would say (or simply think) "With a BMI of 27, you are at a high risk of suffering from obesity-related negative health outcomes now or in the future." This is a reasonable assumption to make given the info Northwestie has to work with, but it is still an assumption, and assumptions are often incorrect.
Let's say it turns out the random stranger with a BMI of 27 has a low body fat percentage wears 32W pants and performs Olympic lifts with heavy weights 3-5 days a week. This person would in fact have a LOW (thought not NO) risk of suffering from negative obesity-related health outcomes. So in effect, Northwestie's seemingly reasonable statement about the person's health would in fact be WRONG.
Similarly...
Social scientists have done research into the number of households with firearms, as well as the occurrence of negative gun-related outcomes in households. Using this research, they have determined that people in gun-owning households have a 1 in X chance of suffering from a negative gun-related outcome. This may be of some value when discussing the effects of gun ownership on populations.
If a random person on the internet says "I have a gun in my household", a person like Northwestie (or GuitarSTV) would say "You have a 1 in X chance of suffering from a negative gun-related outcome." This may be a reasonable assumption based on the information they have to work with, but it is still an assumption and assumptions have an inconvenient tendency to be wrong.
Let's say it turns out the random gun owner on the internet has extensive formal training on the safe use of firearms, and is in fact a firearms instructor. This person is in good mental health, religiously follows safe gun handling and storage practices, and doesn't have children in the home. This person would in fact have a LOW (though not NO) chance of suffering from a negative gun-related outcome. Much less than the 1 in X chance seemingly predicted by the study. So, in effect, Northwestie's assumption would turn out to be WRONG.
None of this means that research performed by scientists or even social scientists is USELESS. It doesn't mean that it is not ALWAYS useful.