Monday 23 January 2023

Further RPG Transcript Analysis: Old School

Recently, I wrote about a taxonomy for classifying the statements made while playing a role-playing game. I've been looking at an 'old school' style transcript that I was given by Ara Winter.

With the whole transcript annotated, including breaking up longer statements into separate ones so they can be tagged distinctly, it looks like this in my google sheet:

This lets me do a couple of fun but simple bits of analysis, such as looking at the relative proportion of various statement types, by the GM: the players:

..and the group as a whole:

But the coolest view for me is looking at next statement types. What statements tend to follow others? I crunched the transcript and it allowed me to make a sort of Markov chain out of it, showing the likelihood of a statement type following another:

What I love about this is that you can really see the style of conversation that emerges in this session: the rapid back-and-forth of description and inquiry between players and GM. Once in a while there are little conversations about clarifying the fiction, player intent, or how to approach the situation. Joking tends to lead to more joking!

That was just the top ten most common statement types; here's the full diagram for the entire transcript:

In the next post, I'll have a look at a very different play style: Critical Role.


  1. This is brilliant. The top ten actions fits with my own anecdotal observations about the play loop in old school games. Thanks for doing this.

  2. Great analysis! I find it so interesting to see the conversation we all talk about conveyed here in the data. I used a similar coding process to analyse various session and campaign prep documents. Do you think you might apply this analysis to other aspects of the hobby such as prep documents or published modules?

    1. Maybe..I'm not sure how that would work! On the face of it, I don't see how the statement types would map to written materials. What did you have in mind? Looking at things like the relative proportion of stat blocks vs. description vs. GM advice in a published module?