Thread by @Tan_Shelly

You've might heard I published the last and FINAL update to the #GameOfThrones death count graphic here: wapo.st/thrones

I was a naive, tiny baby when I first pitched this graphic 5 years ago shortly after season 4. Here's a lil thread about how this all started:

*extreme 80-year-old narrator voice* It was the summer of 2014. I had just graduated from NU and joined WaPo, and lil ol Shelly Tan was eager to pitch her own major graphic for the first time.

"Another character bit the dust on GoT. What if I counted all of them?"

You fool

I pitched the idea. Eds were interested but not convinced about the feasibility of counting everything. (NO KIDDING I SHOULD'VE LISTENED.) They said if I could get the data, they were in.

For the next 6 months, I slowly counted the first 4 seasons in my spare time...

It quickly became obvious I needed some ground rules. First: I decided to ONLY COUNT ON-SCREEN DEATHS. Caps bc I've gotten MANY angry comments/tweets about how the count is clearly off cuz "the wight army was millions strong blah blah"

SHUT UP LEAVE ME ALONE HERE'S WHY I DID IT

1. If any of you decide to count for yourself, you'll realize some things very very quickly: The things that are said about battles/deaths rarely match up with what's on screen. For example, the Golden Company supposedly had 10,000 men. Here's an overhead shot of them from S8, E5

2. Sure, could be more of them hiding off screen. But we sure don't see them, and I have no way of verifying. Any army size claims are just wild guesses. Same for the wight army's "millions" and for earlier seasons, like the 2000 men Robb sent to their deaths at Whispering Woods

2b. Or what happens if we get an army estimate for one battle, but then no estimates for another? Do I just start wildly guessing on my own, based on assumptions about one side's power? Inconsistency galore! So I settled for on-screen deaths, because at least I could verify those

2c. (This also applies to places, like the Burning of King’s Landing. Again, I can guesstimate how many people are in the city. But I don't know for sure, and some of them may have survived ¯\_(ツ)_/¯)

3. Finally, I chose on-screen bc I also knew I wanted more granular data. I released the data publicly on GitHub (github.com/washingtonpost…), and if you look through, you can see I included stuff like:

- Killer
- Method
- Location
- Allegiances

Can't do that w/ off-screen deaths

Anyways, moving on: Besides setting rules for on-screen deaths only, I also needed a way to check over my own work. Cue the 4 Watch Rule

1. First watch: Scrub until I hit a death scene
2. Second watch: Recount any large battle scenes
3. Third watch: Second recount of battle scenes
4. Fourth watch: Watch through w/o scrubbing, cause sometimes deaths happen in the bg or in transition cuts and I'd miss them

I'm gonna confess right now that I absolutely did NOT recount the Battle of Winterfell. If any of you wanna fact check me on that battle and count for yourself, please feel free to do so and I will absolutely ignore you after you're done

Another ground rule was to still include off-screen deaths of prominent characters, bc it'd be silly to have them obviously missing. Off-screen needed to either be confirmed or assumed due to imminent danger. So no, I'm not taking Syrio Forel off this damn graphic stop @-ing me

So those were the initial ground rules I set for myself. Obviously, my methodology section expanded DRASTICALLY as time went on, but that's a story for another day

Fast forward a few months. I've crawled out of my hole with my finished spreadsheet clutched in my hands.

"Please ma'am, can I make this graphic now."

And THAT was when the illustrations started

You ABSOLUTE FOOL

To be totally fair, I knew from the beginning that I wanted to make this an illustrated guide to the deaths. I totally could've changed my mind by the time I finished getting the data

But nope, fresh-out-of-college Shelly Tan was AMBITIOUS

To save SOME of my sanity, I did decide early on that, the more important the character, the more detailed their illo would be. So major characters got full body illos, secondary major got busts, so on and so forth

When I first started, I actually sketched the major characters on paper before scanning it in and then cleaning up in Illustrator. Since then, enough major characters have died that I started using old characters with similar builds as a base and just customizing from there

I try to give each character its own flair, but if you compare them side by side, you can see the base body I use

#GOT does a really good job of connecting characters w/ subtle costume details, so I wanted to do the same w/ color + accessories. Ppl from similar geographic areas got similar skin/clothing tones + fashion details

For example, everyone at or beyond The Wall has a pale, slightly sallow tone, as if they haven’t gotten much sun in a while. (On a side note, this also speaks to just lacking in diversity this show is, but also another thread for another day)

Another example: People from the North, like Robb Stark or Ned Stark, share fashion accessories -- fur cloaks, leather bands that cross over their chests in an “X,” etc

The Pattern tool was my best friend for a lot of this. The clothing is super intricate, and patterns helped with the tiny details, like the woven Dothraki armor or the really lavish Qarth clothing

Since I'd gone with a pretty minimalist, vector style, facial features were also important to get right. Small details, like the shapes of their eyebrows or hair, were important identifiers

Level one characters were the easiest -- I just recycled the same basic head and bust, alternating hair/clothing colors and beards as appropriate

Level four characters, on the other hand, needed a bunch of layers for all the costume details

Here's what my Illustrator file for level four character illo now looks like, at the end of all this

Level threes...(I tried to keep characters from similar geographic locations in the same columns)

Level twos, aka a bunch of floating heads

And finally, all the level ones, aka the poor bg extras that kept dying over and over and over again

This graphic had a lot of numbers, so here are the final ones:

- 5 years of working on this project
- 6887 deaths
- 290 character illustrations
- Far too many hours of sleep lost

And now, at the end of it all, my watch has finally ended. Thanks for following along

Also, after finishing the ENTIRE damn thread, I've only just realized I have a typo in this very first one. UGH

You *MIGHT'VE heard