Normalization is a method for organizing data to fit into a computer. Unfortunately organizing data is not the same thing as organizing information.  Without clear understanding of the differences between data and information, normalization relies far more on what data looks like than what information means.  Information possibilities and potential are sacrificed. 

Normalization requires restricted perspective and criteria.  It fits well with and reinforces the IT tradition of looking at the little picture instead of the big ... looking at some factors to the exclusion of others.   In the real world, outside the artificial world of IT, we work very hard to overcome normalized thinking.  A little picture, limited factor, normalized perspective of "flight" for example, would make it difficult to tell the difference between a vulture and a jumbo-jet because they both have wings. 

Prejudice springs from normalized thinking.

Ironically, many if not most system development efforts reach a point that, in order to satisfy requirements, the rules of normalization are set aside.  Big picture reality intrudes on little picture restrictions.    This should be a red flag to developers that normalization is seriously flawed, but the concept is so entrenched in the IT psyche it is never appropriately questioned.

Normalization improves the process of making systems but fails the function.

Because we have no criteria for what constitutes effective information development, normalization has never been validated as an effective way to organize information.  It is not.  It virtually guarantees poor information development.  Normalized data is almost always out of context. Though standard practice, informationally, normalization has always been a bad idea.