The Watson Supercomputer Doesn’t Find Rappers Conscientious

IBM’s Watson is the supercomputer of the moment in the same way that Fetty Wap and Omi are the musicians of the summer. But what does the machine make of the men? Well, thanks to the new Tone Analyzer, the latest widget in its “Cognitive Computing” cloud, which also has opinions about food and shopping, we can ask. Watson can’t analyze a tune just yet, but it can prove that this summer’s most popular songs aren’t very public spirited.

The tone analyzer breaks writing into three different parts, Emotional Tone, Social Tone, and Writing Tone. The Emotional Tone segment is defined largely by word choice and the frequency of highly demonstrative words. Social Tone is more of a holistic product, but it is determined by the ways in which the writer address the audience. Writing Tone has to do with the ontological approach of the author. It is not a massive help when looking at rap.

What’s notable about rap songs put through analysis is their social tone. Omi’s “Cheerleader” and Fetty Wap’s “Trap Queen” are the Social Tone scores, which show that both tracks demonstrate an extreme openness and a high density of socially significant language. But neither score particularly high on “Conscientiousness,” the third category behind Openness and Agreeableness. This illustrates that the songs are, while cheerful and summer jam-y, not determined to open a dialogue with the audience.

What’s interesting is that “Trap Queen,” the less conscientious of the two songs, has way more writerly language and is actually considered fairly analytic by Watson. This is kind of fair, it’s a song about a relationship sung by someone who appears to be emotionally invested. “Cheerleader” is not that kind of song. To put that in context, Watson considers Wiz Khalifa’s “See You Again,” which feels a bit more thoughtful than the other two songs on first listen, significantly less writerly than “Trap Queen.”

What does this say about the songs of the summer? That it pays to write something people want to sing along to instead of think about, for sure. But it may say more about Watson, which proves to be a fairly effective critique of unexpected material. There’s a density here that a lesser program might confuse for, well, confusion, but the Watson analysis is too mathematical to be led astray by something so secondary (not being sarcastic) as the message conveyed. Watson proves it’s value by analyzing overall impression rather than line editing, which is a damn good tool to have if you’re trying to write a business email — or a hit.