The Consumer vs. the Beautiful Randomness of the Beatles

Advertising doesn’t work if people are not to a large extent predictable. In fact, if people aren’t largely predictable, then society pretty much breaks down. The insurance industry is premised on the idea of predictability. The supply chain. Banking and the stock market. The whole economy, in fact. Driving. Movie production. Autofill. Pitch selection in baseball (the pitcher has to anticipate what the batter considers predictable and then go against that). The list is pretty endless.

This analysis, written up in The MIT Technology Review in 2019, claims that machine learning studied the writing style of Shakespeare as well as that of another playwright named John Fletcher, and was able to determine that some of Henry VIII were written by Fletcher, and which parts. 

This was possible only because we all – even Shakespeare, the most genius among us – have stylistic “tells.” Tendencies. Preponderances. I myself probably overuse the one-sentence paragraph for dramatic effect. And it’s fair to call my reliance on the em-dash – while often valid and honest – a writer’s tic. Sure, human beings are less predictable than dogs, but we’re still pretty predictable.

Except when we aren’t.

What about the intentional aberration? What if we aim for unpredictability, to have a style-less style? Or an abundance of styles, all at once? I haven’t watched the World Series of Poker in a while, but I’m sure one of the assets of the best players is unpredictability. Of all the remarkable things one can say about the Beatles, there has never been a group, not even close, that has produced such a range of popular songs. True, some were composed by Lennon & McCartney, some by Lennon, some by McCartney, some by Harrison, a few by some other mix (all of them together, a tiny bit by Starr). Still, the range is astounding; if you compare their output just to other contemporary British bands with roughly four to five contributors – the Rolling Stones, the Who, Led Zeppelin, etc. – the variety of music produced by the Beatles is almost laughable.

That’s not to say that the Beatles were trying to be different; just that they achieved a level of unpredictability (especially after their first few albums) almost unprecedented in art, modern or ancient. If someone appeared on Earth and knew very little about the origins of particular songs but enough to have a human conversation, and we played for them The White Album, or Abbey Road, or Sergeant Pepper’s, they would likely believe each was a collection of songs from different bands, not from one and the same.

What will AI learn from work (text, music, imagery, video) produced by the most unpredictable of us? Will that content be treated as an aberration largely to be ignored/marginalized*, because it comes from such a small percentage of the population, and has so few correlates? Or will it be weighted all the more precisely because it’s unique? Tim Ferriss famously said “you are the average of the five people you associate with most”: Just how off-center can that average ever get if (a) the vast majority of us are predictable and (b) we’re averaging, a fundamentally flattening, predictability-seeking endeavor? At what point does AI understand and appreciate the most unpredictable among us?

(*AI, like any human-created system, or maybe like any finite system period, will necessarily favor certain data over other data. No matter how fair we endeavor to make it.) 

What happens if we all – gradually, inexorably – become a bundle of quirks? We’re all, essentially, one-offs? If there are fewer and fewer archetypes? (Maybe not so much of a stretch, since we all now carry around with us our own personalized information/entertainment networks.) Will marketing and advertising remain as responsive as can be because of AI’s ever-broadening power?

If the firehose approach of marketing and advertising - for example, television advertising in its infancy, when the goal was to have a very few programs on very few channels that appealed to the broadest possible audience - is being gradually replaced by the needle-threading approach - with each person’s desires and peculiarities understood and catered to - then isn’t AI really the ultimate champion of the quirky, the unpredictable, the persons who don’t fit in?