The sixth season of the HBO program Game of Thrones begins April 24.

And if you're a fan of the books on which the program is based, you're not quite sure what to expect. Author George R. R. Martin has not yet announced a release date for the sixth book in the series. Which means the television show is about to move ahead of the books, in terms of where the plot is heading.  

The result? Fans of the book are preparing for spoiler alerts of epic proportions. In a finely timed bit of marketing, IBM researcher Vinith Misra used a Watson program called Personality Insights to analyze how Game of Thrones characters have evolved in the first five books--and anticipate what readers might expect in the sixth book. Here's what the data revealed: 

  • As Daenerys Targaryen embraces her inner Khaleesi and queenship, she is slowly losing her openness and liberalism, and is becoming simultaneously more worried, angry, assertive, and dutiful.
  • Sansa Stark's original extraversion and cheerfulness has been replaced by self consciousness, an imaginative internal life, and dutifulness.
  • Her sister, Arya Stark, has become hardened throughout the books. She is now less vulnerable and less prone to worry. 
  • Tyrion Lannister's initially more trusting and disciplined persona has given way to a vulnerable and emotional alcoholic.
  • Jon Snow has changed from being the angry, vulnerable, adventure-seeking youth to a disciplined, intelligent, and cautious leader.

If you've watched the series or read the books, your first reaction to all of the above might be: Duh. I didn't need Watson's Personality Insights to tell me any of that. What's the big deal?

The big deal is that "Watson is not reading the actions of these characters the way a human would," says Misra. Watson came to these accurate readings of the characters strictly through an analysis of word choices used by the characters in the first five books.

For example, when Watson is determining a character's altruism score, it's picking out first-person plurals ("we," "us," "our," "ours") and words like courage, bravery, daring, certain, definite, confident, easy, and faith. Swear words impact negatively on a character's altruism score. The most altruistic character, according to Watson, is Bran. The least altruistic is Tyrion. 

Likewise, when determining a character's assertiveness score, Watson looks for words such as fight, forgive, gave, meet, negotiate, explain and persuade. The most assertive characters, by Watson's analysis, are Arya and Brienne. The least assertive are Barristan and Davos. When Watson is determining a character's emotionality score--a trait combining self-awareness, empathy, and passion--it looks for words like hugging, holding, ache, alcohol, affection, admire, adore, and grin. The characters rating highest in emotionality are Cersei and Daenerys. The characters rating lowest--you could call them the most stoical--are Davos and Ned.

Now imagine you had this capability--to assess personalities through frequency of word choices--in a large organization. If you had to sort through thousands of candidates for a particular job, you could use a tool like Personality Insights to give you hard data on which candidates had the most (or least) altruism and emotionality in their resumes and cover letters.

Likewise, if you were forming an innovation team, and you wanted the participants to have high assertiveness scores, you could use Personality Insights to analyze internal documents (emails, Slack messages) for the data. As in all cases with analytics, be it baseball or business, you wouldn't want to rely on the data alone. But the insights from the data, says Misra, would provide a "sanity-check component." And surely prevent you from trusting a Lannister with any matters of emotional sensitivity.

Published on: Apr 19, 2016