Data Architecture

Computer systems have brought into sharp focus the importance of keeping information organized. The fancy technical term for lining up your columns and ensuring interconnections are sound and precise is called normalization. Buried in that five-dollar word is a deep philosophical statement about the people and systems of the IT world. All are seeking a desire to be normal.

The value of being normal---and in particular, forcing things that are out of the ordinary into a normal structure---is immeasurable. Westerners typically have a first and a last name, with the latter in common with family heritage. Names consist of a string of letters (plus a few symbols, like apostrophes for the O'Shea family and hyphens for the obsessively progressive). This normalized system of names makes so many things possible, such as printing alphabetized phone books, splitting up long government queues into parallel lines, and other ramifications of a teeming planet. Whenever a set of data (such as the names of all people) is normal, there are marvelous gains in efficiency.

Normalize Like It's 1999

Information which does not follow the norm presents a conundrum to catalogers everywhere. Where do you list people without both a first and last name? What about eccentric pop singers, who decide, with total validity, that they have a new name which is an unpronounceable symbol? He jumps around in the phonebook, ending up finally near the front with an awkward, but normalized result: "Artist Formerly Known As Prince, The."

Think like Plato

Data architecture, unlike other aspects of IT, is all about platonic ideals. Defining an information infrastructure is about visualizing pure, perfect forms that model the real-world data, and then compromising for practicality. Let the aspects of the information broil in your mind, until you are certain you have worked out every possibility. It's more philosophy than engineering. Data architecture is about beauty, and beauty cannot be rushed.