The first dimension is the most fundamental: statistical fidelity. It is not enough for synthetic data to look random. It must behave like real data. This means your distributions, cardinalities, and ...
An EV might be slightly quicker to cover a quarter-mile, but it won’t deliver the same thrills.