@snuffsahoy - Excellent points. My “common sense” take on it is that inhaling anything other than clean air into your lungs is, in all probability, at least less than ideal. Sure, you might have a case here and there of someone smoking cigarettes for many decades to no ill effect - no cancer, no COPD, no bronchitis, no cardiovascular consequences, but those are instances where “the exception proves the rule”. Probability theory says that it is entirely possible to flip a coin 50 times and come up with heads each time; that doesn’t mean it’s likely. Perhaps more to the point, it’s possible to play Russian Roulette 50 times without shooting yourself, but it’s certainly not the wisest form of recreation.
In any case, you’re absolutely right that epidemiological studies, or rather the conclusions drawn from them (especially in the media), notoriously conflate correlation with causation, whether in studies of smoking, dietary habits, or anything else. But there are instances where the correlation is so strong, that it at least implies, though granted does not equal, causation. Apparently Buffalo Bird Woman understood this.
A complicating factor is that both provable and non-provable factors are at play at the same time. For example, nicotine is a clinically provable vasoconstrictor. This is absolutely undeniable. But as for the long-term effects of persistent vasoconstriction through long-term use (this applies to caffeine as well), there is no clear-cut data regarding cardiovascular outcomes in real life cases.
With observational epidemiology, I think the best we can rely on is what the law would call a “preponderance of evidence”, without futilely looking for slam-dunk conclusions at the micro-science level, as long as correlative factors are accounted for and variables controlled. For example, the majority of smokers do not get lung cancer; however, the majority of people with lung cancer, across populations and controlling for lifestyle variations, are smokers. Therefore, we can reasonably conclude, just through the use of common sense, that smoking in all likelihood has something to do with it. Absolute proof of causation, in the most restrictive forensic sense, is not necessary to draw reasonable conclusions.