By ignoring the structure in data acquisition phase, you just postpone decisions about structure to data processing phase, now without necessary information about the structure (which could be obtained in data acquisition phase).
For example, such basic functionality like changing sort order between given name and surname would be much more complicated.
Besides, the whole story is about the problem that name including title was in one field and later processing (title removal) misbehaves due to insufficient information.
For example, such basic functionality like changing sort order between given name and surname would be much more complicated.