> The tricky part of this proposal is that I’ve encountered many data modellers in my career who can offer the “denormalized” option, but few … nay, none who could tell me what the properly normalized data model was.

We have a data model like User has many Xs and Ys, both Xs and Ys have a foreign key to the User via user_id. The normalized version would be User has many Ys through Xs. We denormalize and store user_id on both for performance reasons. This seems like a) so common, b) so straightforward a refutation of this claim that I must be misunderstanding the author.