Carl,
The message was a hypothetical situation. Like any hypothetical situation it includes assumptions that are givens. It is not a relation of any actual situation. It is an example used to demonstrate the process of denormalization and to describe what that process is and where it can and shoiuld be used.
The change to the data structure introduces a violation of the 3rd normal form. In the story the justification of for the change is a hypothetical performance gain. The only reason the performance is even mentioned is that it is necessary to have a REASON to denormalize. If there is no reason then there is no excuse for a denormalized structures.
In truth, I don't really care if you agree or not. I give some degree of creditability to my own 20 years of experience with xBase and its performance. And yes I have tested multitable joins using SQL and using SET Relation against a single table and there is a measurable timing difference. Using a single table is invariably faster. It is usually not faster by much so it is irrelevant until the record counts become very high.