[goto11]

Denormalization means you (by definition) have duplicate and redundant data in the tables. This certainly requires careful planning around updates to avoid data corruption.

> The number of tables explodes

This is IMHO a backwards way to understand normalization. Normalization tell you how many "logical relations" you have in the data - whether you like it or not. If you decide to squash multiple logical relations into the same physical table then you get redundant data (by definition). This might be a fine optimization, but it doesn't reduce the inherent complexity of the data model. It just saves some joins as the cost of more complex update logic.