Strawcup does podcast stats

Exactly two years ago, a group of us launched the TMA Fan Survey as part of the fan project Eye Love TMA to celebrate the podcast that means so much to so many. In the two months between survey launch and the March 2021 finale, we got over 3,500 fans (and a few cast/crew members!) to take the survey. I had a blast getting to examine such a large set of data and analyse how the fandom related to the podcast through items such as:

Favourite character

Favourite season

Scariest episode

Charisma levels of the fandom as compared to the Archivist

What the Admiral looks like

How the Admiral wears trousers

Which entity fans are most afraid of

Which dancing Theresa May haunts the Archivist in the gap between his nightmares

And so, so much more!

Now, since the fandom has only continued to grow since the finale, and that means there are even more people who are able to contribute to this project, I decided to readvertise the survey.

Through the launch of TMagP, I’m going to be running a drive to collect Even More Data for the survey and release an updated newsletter come October.

Look at the finale-era results here.

Tumblr post of highlights from newsletter launch by the incredible niksfake

Adorable comic by shanni used to advertise the survey

Just to note that, going forward, I’m only going to be looking at the multiple choice questions on the survey. I already read through 3,500 responses to each of the 6 short answer questions and I am not going to read through any more, thank you very much.

ZOLF - Since Zolf did the bulk of the party's healing across the campaign, below is a bar chart containing all 27 of his heal checks across his days as both cleric of Poseidon and cleric of hope. Overall, he does exceedingly well at it, which is part of why he healed the party for the most points.

BERTIE - Despite his insistence on his own greatness, Bertie was largely statistically uninteresting. However, as seen in the table below which contains stats on the saving throws in the campaign for character, he rolled the most saves per capita, averaging nearly 1 save per episode. This is likely due to a combination of James antagonizing Alex & Bertie antagonizing everyone.

GRIZZOP - I have zero idea how Ben pulled this off, but the spread of nat 1s and nat 20s for Grizzop is ridiculously beautiful. The tables below contain information on the crits for skill checks, saves, and attacks, as well as a combined total for all three metrics. Skill check & save crits were perfectly balanced as all crits should be, and there was only a 3 crit disparity in attacks. Plus, the total number of natural 1s and natural 20s was just about 10%. This is incredibly close to what you would expect to see using d20s in an ideal system, which our world patently isn’t, so wow!

HAMID - You might have trouble remembering, but Hamid used a little spell called "fireball" once or twice amidst other firey attacks. Here's a nice bar chart comparing just Hamid's fire damage, which was 71% of his total damage dealt, to everyone else's total damage. Cel is the only one whose total damage just barely eeks out Hamid’s fire damage, and that only happened in the last 3 episodes of the campaign!

AZU - A bar chart of Azu’s Knowledge (Religion) rolls. Of the 16 rolls, 3 were natural 1s, which had a 3.6% or 1 in 28 chance of happening. Understandable it did, though, Azu had better things to be doing than paying attention in seminary...

CEL - From Cel’s 28 Knowledge (Engineering) rolls in the bar chart below, we can see that they really know their stuff! Nothing was below a 12, and a decent portion of rolls hovered around 30.

WILDE - And for my next trick, I shall perform the most difficult of statistician's techniques I toiled for years to perfect: glaring in the vague direction of London and spamming memes.

(thanks to serhawke (@cityelf​) for the Dragon!Alex Dice Hoarder doodle)

SKRAAK - A bar char of Skraak’s 8 stealth rolls. These checks themselves rolled a natural 20 on stealth. They look deceptively poor until you remember to look at the scaling on the left (how are none of these values below 25?!?!).

And now, weary, my labours completed, I shall impart upon you all one last tidbit:

WHY SKILL CHECKS PER MINUTE (SCPM) IS THE SUPERIOR UNIT FOR CAMPAIGN ANALYSIS

So I’m sure you all can hear the voice of my PhD supervisor in the background shouting “Straw, skill checks per minute (scpm) is an utterly meaningless measurement for evaluating episode content across a large campaign since the small amount of interesting information it could give you is in a format that is difficult for humans to conceptualize, what could you possibly use this for? What are you even trying to say with this unit, Straw?!” 

And, well, to an extend, he (and literally all of us who are thinking this) would be correct. However, and hear me out on this:

- You can pronounce scpm as “skip-uhm” and that’s just the kobold’s knees - Anything above 1 or really close to 0 is an obvious outlier and should be easy to spot in the mass of data. Sure, you could find them just as easily in other formats, but this range is fun and perky. -It’s just a really neat unit and you can pry it from my cold, dead claws.

In all seriousness, to all who worked on RQG - thank you. If not for you, we would not have had this fun sandbox of numbers to play with, nor this wonderful adventure to all embark on together.

Thank you.

Stats nerds like me will be happy to know that I compared normalizing by episode against normalizing by a rough approximation of time due to the variation in episodes across the podcast, with episodes getting shorter on average as the crew hit their groove. There was no major statistical difference between the two, and normalizing by episode corresponds exceedingly closely to by runtime

Since I looked at this, I discovered that time-based stats that stand on their own. Here's some data on episode times.

Bertie was around for 69 hours; Zolf for 69% of both episodes & runtime. Nice.

Next, here's how much the party dealt & took damage by episode. 2 highlights:

1) One of the few times where damage taken is greater than damage dealt is 39 & 40. Party visibly got rekt by an old man 2) The biggest bar is in 216. Cel dealt 265 of that 453 damage, or 3/5 of the combined bar size

Editors and producers do a lot of magic behind the scenes to make the final product that reaches our ears. Here's a table of who edited/produced both RQG campaign episodes and specials.

Cathy's edits have a release runtime of 0.69 days. Nice.

Alex’s edits have a release runtime of just over 9 days; Lowri’s edits have a release runtime of just under 5 days. Please someone get them to nap.

This post (and subsequent parts) will comprise of stats I found interesting but was unable to fit into the book. The data covers numbered campaign episodes, Bertie's Sidequest, and Ancient Rome. Data collected by myself and Oddly Specific using the RQG wiki & relistening. Thanks to the wiki contributors.

If you want to take a look at all the raw numbers that got us here, this linked spreadsheet has them. Feel free to copy whatever data you like for your own purposes. If you want to know something not included or if you spot an error, contact me about it!

Thanks to serhawke (@cityelf), Dubious Gem (@regulationblues), Sumi, nik (@niksfake), Nix (@scatteredheroes), Oddly Specific, and Lowell for being such fun people to work with on this project.

Thanks to NotusLethe, Booksquirm, and cuewulfir for additional support. 

Finally, massive thanks to Lowell for turning all of my doodles, rough charts, and data into the beautiful infographics contained herein. They’re the one who’s made these stats sing.

And now, after 6 months of work, I present to you: Rusty Quill Gaming as told by statistics.

We start with an animated scatter plot detailing the values of every skill check stated in the campaign that wasn't a natural 1 or natural 20. I like how clearly it shows the growth in power of the party (can’t figure out how to embed it, so we’re going with an imgur link for now).

https://imgur.com/cgPkrXV

Up next, a line chart containing skill check overview for how many skill rolls each character had accumulated at each episode of the campaign. I was surprised by how pretty and overarchingly linear the accumulations were.

Here's yet another skill check overview, this time divvying up all 2,818 skill checks rolled during the campaign (including natural 1s and 20s this time) by character. From this pie chart, we can see that Sasha emerges with the most rolls at 617, despite being around for only 57% of the campaign.

Next is the same pie chart as before, but player-based this time. We can see that Lydia carried the team with a whopping 870 skill check rolls, which is almost 1/3 of the d20s tossed for skills in the campaign. That's a lot of dice!

Another player-based skill check chart, this one averaging skill checks per every 10 episodes for ease of reading. I like how this one smacks you upside the head with a visual of how the pandemic move to remote recordings altered Alex’s GMing style - once remote recordings were the standard, the average number of skill checks per episode decreased slightly, and the party started rolling skill checks more as a unit more than individually, unlike before the pandemic.