Popular Analytics: The Data Science of Hit Making

Data science is increasingly accepted as an effective way to accomplish things that used to be the domain of savants and soothsayers – identifying undervalued baseball players, for example, or predicting election results, or modeling the future performance of investment portfolios.

But what about matters of public taste and fashion – things that are more ephemeral and difficult to measure than batting averages and earnings growth?

In Hollywood, where creating hits is serious business, the dreams, fortunes and careers of many, many people are predicated on such fickle matters. We refer to the people at the top of that list – people like Lady Gaga, Taylor Swift and Hozier – as artists. But how much of hit-making is really art, and how much is science?

The fact is that our personal tastes in music, food and footwear can be predicted with increasing accuracy, just like our likelihood of repaying a debt. We’ve all had the experience of getting alarmingly accurate recommendations from Amazon or LinkedIn. In much the same way, apps like scoreAhit claim to predict whether a song will become popular based not on the emotional resonance of the lyrics with an audience, but on such prosaic, easily measurable factors as tempo, duration and harmonic simplicity. Given that, what's to stop analytic scientists from developing algorithms to identify successful novels, or paintings, or plays?

The implications of these questions are fascinating. If our free will isn't as free as we like to think, and if our personal tastes are as knowable as our hair color and height, then it stands to reason that promising, young scientists may soon choose a career in marketing over one in chemistry or microbiology.

But here’s what I’m really wondering: If we can reduce hit-making to a science, and systematically manufacture songs that people will buy to ease their worried minds, then what’s to become of artistic innovation? Would Kurt Cobain, Bob Dylan or Miles Davis have stood a chance in the era of Big Data analytics?