If you’ve ever played Dungeons and Dragons or any classic role playing game, you’ve probably heard about the concept of ‘character alignment.’ This is when a fictional individual is placed in a relative position along a two axis grid depending on their ethics (good vs. evil) and morals (lawful vs. chaotic). Adding a midpoint to each axis (neutral) allows for nine discrete character alignment configurations: lawful good, lawful neutral, lawful evil, etc. This is great for comparing and contrasting what drives character motivations within a given story. But what if you want to categorize the stories themselves, not just the characters within them?
Categorizing ‘story alignment’ requires a different pair of axes. Let’s use tone (noble vs. grim) and setting (bright vs. dark). Again with a neutral midpoint, we have nine discrete variants: noble bright, noble neutral, noble dark, etc. But shoehorning stories into predefined boxes pares away much of the narrative’s rich diversity and we end up losing data. Why not use a gradient scale instead? A story’s tone gradient would run from ‘noble’ in which the outlook and characters are optimistic, hopeful, and transitory to ‘grim’ in which the outlook and characters are dour, pessimistic, and stagnant. The setting gradient would go from ‘bright’ in which the world is a golden age of adventure to ‘dark’ in which the world is a bleak, war-torn hell.
Now we could simply place our favorite stories anywhere we please within the (X,Y) vis a vis (tone, setting) coordinate system of our 2-dimensional story alignment plane. We could do that . . . if we wanted to be terrible data scientists! Oh, the horror. Such subjective criteria would veer wildly depending on the order the books were read, who was reading them, and even from emotion-tinged memories of past reading experiences. We leave ourselves wide open to reader bias, subjective emotional skewing of our data, and good old-fashioned human error. Let’s see if we can smooth out those lumpy variables and provide some consistency, rigor, and duplication. Ideally, we want a process that works in a similar fashion for any given person analyzing any given story.
A better method would be one that uses objective criteria: we create a standard of measurable variables that can be universally applied without bias for all stories before we even begin to examine them. We can then analyze these variables using non-Bayesian statistics to place the stories in our given 2-D plane relative to each other.
To establish tone, we could examine the following variables:
1) Frequency of named character deaths
2) Frequency of nonviolent conflict resolutions
3) The Hero Alignment Metric: #heroes/(#heroes + #anti-heroes)
4) Presence of a HEA fairy tale ending
5) Frequency of pages devoted to witty banter.
Setting could be examined using these variables:
1) Frequency of village, town, or city destruction events
2) Frequency of chapters in which the protagonist(s) engage in battle
3) Mean economic level index of major narrative waypoints
4) Percent of chapters that feature rainy, foggy, or overcast weather
5) The frequency of historical wars mentioned in the backstory.
This is a good start, but hardly exhaustive. So, let’s make this a group effort. I am throwing down the gauntlet to you, fair reader. What unique measurements or improvements to the existing ones might you devise to quantify a story’s tone or setting along the noble-grim and bright-dark gradient axes, respectively? I’m shooting for 10 variables for each, but the more the merrier.
Remember, your variable has to be something we can objectively measure. Email your suggestions to firstname.lastname@example.org by Nov 1, 2017 and then the fun begins: we grab some novels and put this sucker to the test.