Then....then what is the job?

As far as I understand, the argument is that a reviewer should check whether the claims that are made are supported by the data that are reported. However, they do not actually expect reviewers to analyse the quality of the data as such.

To publish advertisements, essentially.