Winter is here and so is the frenzied interest in Game of Thrones. From the first season, it’s seemed that anyone might die in the show. As an episode’s opening credits move into the foreboding black screen, suspense fills the air. It’s the tension of uncertainty. Is death that unpredictable? Or, are there some rules in the game of those battling for the Iron Throne?
Here at Tresata, we look at data. Patterns exist in many places, even unexpectedly where we might perceive total randomness. So, it was natural to dive into Game of Thrones. But first, where’s the data?
Data can become an asset when multiple sources are combined. Taking two data sources, one might even say 1 + 1 = 500. For this blog entry, we worked with death data from the GoT Deaths page at: http://deathtimeline.com/. This gave us the time of deaths, who died, and whether the character is minor, recurring or major. We combined this with information on the writers for episodes from Wikipedia at https://en.wikipedia.org/wiki/List_of_Game_of_Thrones_episodes.
Enough of the background! Give some hints on what might occur on Sunday!
You can get a bit of a hint on the opening credits. At the time of this writing, we’ve yet to learn who wrote Sunday’s episode. Why? Could it be the hints it gives to the death from the data?
Take Episode 2 from this season, written by Bryan Cogman. Cogman has only killed minor and recurring characters in his episodes. What happened in episode 2 this season? No major characters died. In fact, except for Joffrey, major characters have only died in scripts written by David Benioff & D. B. Weiss. So, pay attention to who wrote the episode this Sunday.
Let’s return to Joffrey. In the first six seasons, he’s the only major character to die before Episode 5.
That brings us to this Sunday, which will air Episode 5 of the Seventh Season. Using Tresata’s real-time, full distributed visualization tool TIDES, let’s graph the number of deaths per episode. You see this below:
On the x-axis, we see the episodes and on the y-axis we see the number of deaths. Note how Episode 5 has approximately the same number of deaths as Episode 4.
So, according to that we should see about the same death rate as last week. BUT, it may not be that easy this year! Season 7 of Game of Thrones will have only 7 rather than 10 episodes. Note, in the graph above, how, from 10 episode seasons, episodes 6 and 7 have a dip in the number of deaths. Since 5/7 = 7.1, Episode 5 is 70% through the season. So, we might see a dip in the number of deaths with the show concentrating on plot development before a spike in deaths as we move to the end of the season.
But wait, there is another view…Sunday’s show is 3 episodes from the end of the season. Looking again at the graph above, we see a spike in the number of deaths when we are 3 episodes from the end. This sharp increase consists largely of minor characters.
Wait. If this fits the “pattern” of past seasons, there are three ways to look at this:
- It could be like fifth episodes from seasons past and have approximately the same number of deaths as the week before.
- It could be like episodes 70% of the way through the season and have a dip in the number of deaths. Or,
- Since we will be three episodes from the end of the season, we should get ready for a spike in the number of deaths!
Data often gives a glimmer of what might happen with randomness (and humanity) sometimes changing one’s trajectory from the more customary path. Are there signs we might look for in Game of Thrones? Indeed. Suppose we are about to see a death-fest with Episode 5 being more like Episode 8 in a previous season (since both are three episodes from the end of the season). When might we see deaths? Again, TIDES quickly and easy gives insight:
The graph above gives the timeline of deaths in Episode 8 over the first six seasons. The x-axis give the number of minutes in the episode and the y-axis is the number of deaths. From the graph we see that a lot of minor character deaths occur in the first 10 minutes. Then we see only a smattering of deaths in the middle of the episode. Everything changes about 45 minutes into the episode with a spike in the death rate. Note further that major character have died only in the last five minutes.
So, pay attention with the patterns from the past in mind. After we see who wrote the episode, hang onto your seat as there are several ways this could unfold. The data gives us a sense, even in the first 10 minutes, of where things might go. Sit back and enjoy the ride. The data perspective might enrich your journey. Who will win and lose the game? From Cersei Lannister we’ve been told, “When you play the Game of Thrones, you win or you die.” Depending on how things play out on Sunday, we’ll see who is one step closer to winning the game!