You're clearly wrong as it relates to the post you're responding to. This one data point is going to get him to engage in a behavior that's healthier than the one he otherwise would've been engaged in, which is clearly good. Thus, the data point is useful!
Not everything his to be a double blind study showing significant effects to be useful.
Suppose that 99.99% of the 80+ year-olds who try this end up breaking their leg, getting a heart attack, etc., and this one data point is from the 0.01% which benefits.
Likely not - but that's why we have double blind studies.
Not everything his to be a double blind study showing significant effects to be useful.