I’m confused why there were ~40 comments in this subthread without anybody else pointing out this pretty glaring error of logical inference (unless I’m misunderstanding something)
I was going to say something similar, that “how dangerous is substance X” only makes sense when you specify how much of the substance X and how often you consume.
Like, when you calculate “the danger of alcohol”, are you describing those who drink one glass of wine each year on their birthday, or those who start every morning by drinking a cup of vodka, or some weighted average? Same question for every other substance.
And if the answer is “the danger of how the average user consumes substance X”, well, what makes you sure that this number will apply to you? (Are you really going to make sure that your use is average, in both amount and frequency? Do you even know what those averages are?)
Then consider the fact that different people can react to the same substance differently. If you specify the “danger” as one number, what is the underlying probability distribution? If substance X causes serious-but-not-crippling problems in 50% of users, and substance Y completely destroys 5% of users, which one is “more dangerous”?
Agreed, there’s two different errors here. One is conflating total harm with per-individual harm. The other, more subtle point you’re alluding to is that a lot of the relative harms of alcohol/tobacco/etc has to do with frequency of use, which is a different question from whether doing X once in an individual or community setting is advisable.
I’m confused why there were ~40 comments in this subthread without anybody else pointing out this pretty glaring error of logical inference (unless I’m misunderstanding something)
I was going to say something similar, that “how dangerous is substance X” only makes sense when you specify how much of the substance X and how often you consume.
Like, when you calculate “the danger of alcohol”, are you describing those who drink one glass of wine each year on their birthday, or those who start every morning by drinking a cup of vodka, or some weighted average? Same question for every other substance.
And if the answer is “the danger of how the average user consumes substance X”, well, what makes you sure that this number will apply to you? (Are you really going to make sure that your use is average, in both amount and frequency? Do you even know what those averages are?)
Then consider the fact that different people can react to the same substance differently. If you specify the “danger” as one number, what is the underlying probability distribution? If substance X causes serious-but-not-crippling problems in 50% of users, and substance Y completely destroys 5% of users, which one is “more dangerous”?
Agreed, there’s two different errors here. One is conflating total harm with per-individual harm. The other, more subtle point you’re alluding to is that a lot of the relative harms of alcohol/tobacco/etc has to do with frequency of use, which is a different question from whether doing X once in an individual or community setting is advisable.