Saturday 28 January 2023

RPG Transcript Analysis: Critical Role

For the third post in my 'transcript analysis' series, I'm looking at a very different play style than I'm familiar with: Critical Role.

For this analysis, I picked a random Critical Role episode, "Between the Lines". Episodes are long, so I chose to do Part I of this episode, which is about 110 minutes and just over 16,000 words.

Unlike most RPG sessions, Critical Role is played for an audience. The participants are professional voice actors, and this shows in a dramatically different kind of play. The sessions are dominated by huge blocks of in-character dialogue.

In the "old school" transcript, the players spent a lot of time strategizing in a way that's not clearly in or out of character. In Critical Role, all of this clearly happens in character, even to the extent that some actions are declared by telling another character about them.

The GM style has a lot more dialogue than in old school play, but is still recognizably similar:

The difference shows up in the player statement balance, which is dominated by in-character dialogue.

The most common action in the old school play transcript was Inquiring for more information - examining something, looking around. In Critical Role, checking out the environment isn't even top ten.

Here's the overall breakdown, players plus GM:

As I was annotating, however, I realized that the play style did shift a lot. When the group was on their home turf, in-character dialogue dominated. Things changed as soon as the party decided to set out for an adventure location.

This got me curious if I could effectively show the ebb and flow of statement types over time. Here's how that looks:

I broke the transcript down into roughly 1,000-word "chapters", and then rendered the distribution of statement types over time.

What immediately stands out is that while IC dialogue is a huge portion, it's steadily decreasing over time. The session starts off with a bit of "last time on Critical Role" (Prior Events), but reaches its peak dialogue-heavy moment in chapter 3, where it's almost pure IC dialogue with a bit of DM exposition.

That immediately slides of as the group sets out. A green band of PC Action appears ("I do x"), along with Describe (the GM describing things)

By Chapter 17, there's even a healthy amount of OOC Approach (statements about how to approach a situation that isn't clearly in-character dialogue)

One similarity with the old school transcript is how little discussion of mechanics there is. Most commonly, the GM simply declares, "Give me a perception check," there's a die roll, and the players are immediately into either groaning, cheering, relevant-but-out-of-character jokes, or in-character dialogue. There's essentially no rules debate in this at all. That's what you might expect from a short-resolution system with an extremely seasoned group.

Here's the "next statement" diagram for the top ten statement types:

This is only top ten, and so we can only see the tip of the iceberg of PC action and how it resolves—some clarifying of the player's intent, but almost always it's simple acknowledgement from the GM. Apart from little sequences of off-topic chatter, game-related joking, or discussion of prior events, all roads lead to in-character dialogue.


  1. This is a very interesting analysis. I've never seen anyone do something like this before. It would be informative to see how different game systems played by the same group varies. This could be a very deep rabbit hole. It could tell us all kinds of things about games. What would also be amazing is if Immersion data could be overlaid with the videos. Immersion is some neuroscience research done by a lab run by Paul J Zak. They put wearable devices on subjects and have them watch a movie or experience an event and then determine what parts of the experience were most Immersive (a term he uses to describe the most emotionally engaging parts of an experience).

  2. Really interesting! I did something similar though at a much smaller scale which I wrote about here ( I did a categorisation about "who" was speaking or seeing things: the player at the table, the character in the game-world or a combination of the two.

    I did only small bits of discourse around an event from a number of games/actual plays. I found something similar to you in that there is an element of "wave" though I observed that in resolving events most discourse moved to or nearer the table-level and then dropped away after.

    For me I felt that there was a difference in styles of play based on the games. The "story games" had a wider swing between player and character, whereas the FKR and OSR games were more "in the middle".

    1. This is fantastic, and very much aligned with what I'm trying to do. Lovely diagrams.

      I feel like your taxonomy maps to what I'm doing, mostly. You're using an unnamed layer a lot (1.5), which seems work picking at. I've taken your first transcript and annotated it in my own taxonomy, then mapped the two to show the relationship. Hopefully this monster URL works!

  3. Wow, this is an interesting concept that I've often recognized, but never imagine codifying. Definitely has me thinking about it now. I'm just wondering if it will now affect my game behavior. Thanks for getting in my head!

    1. In the 90s, we recorded ourselves for the first hour of a Rolemaster session with a friend's new video camera/tripod. When we took a break, we watched it and couldn't believe how little we were actually playing. Most of our discussion was completely off topic. After resuming, we did the same again and found that our play focus had shifted enormously. Some of it is the inevitable bleed of, "How have you been?" into the first portion of play, but a lot of it was just seeing how we play and shifting deliberately.

      Having said that, I think most people aren't making transcripts of their play style. What I think will be the most use of this will (eventually) be making game designers aware of different play styles and how certain mechanics choices encourage them or make them trickier.