Actually a lot of software engineering involves breaking complex systems down into manageable chunks, so this is a lot like what I do for a living. If you view this as the task of taking a real-world business process and formalizing it, then it's exactly what I did for a living for five years.
On to the problem: Dave wants to formalize the current taxonomy of literature to find something more workable. He has some good ideas and he clearly knows more about the topic than I do, but he got off on the wrong foot. He is using as the basis of his formalization an analogy to biological taxonomy. This turns out to be a poor choice for literature.
Works of literature don't have an overall family relationship predicated on a small number of factors. Yes, you can find some family relationships, but each family is organized by different kinds of factors. The overall system ends up either looking chaotic or looking like a tree pasted over a simpler organization.
Instead, think of individual of properties. There is a Chinese food restaurant in Tucson --I think it's at Park and University-- that has an interesting menu. I only ate there once because the food wasn't very good, but I was impressed with their menu. It looked something like this:
Choose a Meat: chicken, beef, pork, shrimp (add $1)What was remarkable is that with just ten menu entries, they were able to describe twenty-four dishes. The cook had found the essential organizing element of Chinese cooking: multiple dimensions. He noticed that he had sweet&sour pork and sweet&sour shrimp and cashew shrimp and cashew chicken and Kung Pau chicken and Kung Pao beef,... and he realized that it could all be combined in different ways. This is multi-dimensional because you can view it as a two-dimensional table: meat on one axis and sauce on the other.
Choose a Sauce: Lo Mein, Fried Rice, Sweet&Sour, Kung Pao, Garlic, Cashew
Linguists use a similar method to organize sounds, but there are lots more than two dimensions. A consonant in any language can be described by giving a set of phonetic features including tongue position, how the sound is produced, whether the voice is used, whether it ends in a stop, and a few other properties.
I think this is the best way to organize literature as well. Expanding on Dave's suggestions, I'll propose the following literature feature set:
Format -- prose, poetry, scriptWith this system, you categorize any work by specifying all of its properties. A Star Trek episode would be
Setting -- modern, western, gothic, science fiction, ancient Greek, etc.
Genre -- action, adventure, romance, mystery, horror, etc.
Length -- *by number of words/minutes*
Subject -- fiction, biography, history, etc.
Content -- neutral, heroic, Christian, erotic, etc.
[script, science fiction, action, 48 minutes, fiction, heroic]A Zane Grey novel would be
[prose, western, action, 90,000 words, fiction, heroic]Of course you could always shoe-horn this system into a hierarchy because the hierarchy is a universal data structure, you can shoe-horn any organization into it. Most data structures are universal in this sense. But there is usually a simplest and most natural organization, and in this case that seems to not be a hierarchy.
Slightly off-topic: here are some previous thoughts of mine on the issue of genres --specifically about science fiction and why Star Trek isn't.