by Raq, Aeluric, and Random McNasty
Data science mecca 538 took on Dungeons and Dragons character generation last week, but their write-up didn’t align (no pun intended) with our years of empirical observation. We checked their numbers and determined that their thesis is totally wrong.
As the 538 blog reports, in August, D&D’s publisher released D&D Beyond that streamlines the process of setting up a new character. Players created hundreds of thousands of characters in the site’s first month, and Curse, the developer behind D&D Beyond, sent users’ most popular picks for races and classes from the game’s Fifth Edition to 538 to crunch. This was the resulting blog:
Is Your D&D Character Rare?
We got a peek at what kind of characters everyone is building, and a lot of players are sticking close to reality.
According to their heat map, human is the most popular race and fighter is the most popular class, and almost half of all characters created on D&D Beyond are human fighters.
We side-eyed that finding, along with a few other assertions in the post. For example, the author suggests that people play the combination halfling rogue because halflings are inherently stealthy, but also because they’ve seen this type in Lord of the Rings. So yeah, they kind of got that backwards – LOTR is the seminal work that helped spawn D&D. Halflings were designed as stealthy in D&D because that’s how Tolkein imagined them. Getting your correlation/causality correct builds confidence for your data science.
Anyway, the heat map in their blog was OK, but Aeluric thought percentages would be more helpful, and he fixed their maths as well:
This helps show that while “human” is popular, it doesn’t necessarily show that most people play “human” rather than a non-human character. And we think that’s the real choice – not human or elf or dwarf or gnome or whatever the heck an aara-cloaca is. The choice of race is human or not.
Here’s where we step away from math and towards lit crit
We offer that in D&D you are really choosing between human and non-human (choose your sub-flavor) most of the time, because that is the way that fantasy writing has always been structured –it’s hard not to think about the human perspective when you are, in fact, a human. So the fact that more than 75% of characters generated have been non-human says something interesting about how players approach the game. If we control for elf and half-elf, which are arguably sub-types of human (super-human and charismatic human respectively), there’s a fairly evenly distribution of non-human sub-flavor.
Looking at the class axis, fighter is the most popular, which makes total sense. The game is designed around fighting. If you take a party of four druids you should name your group the TPKs, but four fighters will do just fine.
Interestingly, despite being the largest overall grouping, in truth fewer than 1 in 20 characters are human fighters. As the author of the 538 blog says, there are a lot of human fighters in media. Honestly, that’s the fantasy genre; if D&D players were responding to the Aragorn/Conan/Boromir/Fafhrd/King Arthur/Roland model of human fighter, way more than 50% of characters would be Jon Snow or Galavant.
We can’t prove it without a lot more research, of course, but these numbers suggest that most people are, in fact, charting their own path in creating characters, rather than pacing the well-trodden boards. They are playing unusual races and classes rather than, as 538 says, staying close to reality.