> You could in fact replace "cannabis" and "schizoprehnia" with "foo" and "bar" and you'd have a template for a shallow dismissal of published studies.

Indeed, replacing with foo and bar would be valid, and the fact that this is the case, i.e. that correlation is not evidence of causation in general, remains a deep and important statement about reality. It is particularly important to correct this when the headline is direct misinformation about what the study found. Do you think code snippets using foo/bar as placeholders are shallow or uninformative? There is a good reason they are often used in programming discussions: to demonstrate that whatever is being shown applies in general.