I agree with a lot of this, you correctly note the serious limitations of research that depends upon 'correlations' but you do not go far enough. To clarify, I have to comment on your comparison of what you term;
"Solid Correlation-causation: Smoking and lung cancer." and "Shaky Correlation-causation: Breakfast and obesity."
To put it lightly, BOTH are 'shaky'! The problems associated with the 'shaky' are EXACTLY the same with the 'solid' - and those problems are many.
The only difference between your 'solid' and 'shaky' is that your 'solid' was effectively the first use of that methodology - it was novel, and several $billions have been spent over several years to disseminate propaganda to reinforce it and promote the 'everyone knows' slogan that many have internalised as (wrongly) definitive. The 'shaky' is a mere a copy of the 'solid'.
"Evidence for the increased-risk causation is solid, starting with statistical studies in the 1950s."
In the 1950s there were very few NON smokers - how many of that cohort were NON smokers? 1950s epidemiological 'evidence', and repetitions of the same, is the ONLY evidence suggesting harm. No other research using alternative methodologies have corroborated those early correlations and as smoking prevalence has declined, there are now far more NON smokers who die from lung cancer (over 80% are never and ex-smokers)). ie the correlation is now an inverse one - smoking has NEVER been proven to have caused any harm!
Correlations can never prove causation, they can only ever 'suggest' a 'possible' link. It does however provide the propagandist with any number of possibilities to manipulate the publics consciousness, as you point out with "If I was a doctor..." comment.
Re: “Correlation ≠ Causation”
I agree with a lot of this, you correctly note the serious limitations of research that depends upon 'correlations' but you do not go far enough. To clarify, I have to comment on your comparison of what you term;
"Solid Correlation-causation: Smoking and lung cancer." and "Shaky Correlation-causation: Breakfast and obesity."
To put it lightly, BOTH are 'shaky'! The problems associated with the 'shaky' are EXACTLY the same with the 'solid' - and those problems are many.
The only difference between your 'solid' and 'shaky' is that your 'solid' was effectively the first use of that methodology - it was novel, and several $billions have been spent over several years to disseminate propaganda to reinforce it and promote the 'everyone knows' slogan that many have internalised as (wrongly) definitive. The 'shaky' is a mere a copy of the 'solid'.
"Evidence for the increased-risk causation is solid, starting with statistical studies in the 1950s."
In the 1950s there were very few NON smokers - how many of that cohort were NON smokers? 1950s epidemiological 'evidence', and repetitions of the same, is the ONLY evidence suggesting harm. No other research using alternative methodologies have corroborated those early correlations and as smoking prevalence has declined, there are now far more NON smokers who die from lung cancer (over 80% are never and ex-smokers)). ie the correlation is now an inverse one - smoking has NEVER been proven to have caused any harm!
Correlations can never prove causation, they can only ever 'suggest' a 'possible' link. It does however provide the propagandist with any number of possibilities to manipulate the publics consciousness, as you point out with "If I was a doctor..." comment.