Radio and the End of Personalization
There is a sense of inevitable satisfaction in the idea that we can get from digital platforms exactly what we want whenever we want it, right?
We lean forward with Netflix and find precisely what we want or what Netflix tells us we will like because of what we have watched previously.
We want that one song from Spotify, and so we find it and listen to it.
We want to create a Pandora station that sounds just like one of our favorite artists so it hews as closely as possible to that expectation.
We swim in a bottomless digital lake of recommendation engines, each looking to offer us something like what it knows or thinks we want based on what we have purchased, viewed, or listened to before.
So where does that leave radio?
Radio is a linear platform, where you get what we have and only that. And it’s produced not with your own personal tastes in mind but with the tastes of the collective you – an audience at scale. The “lowest common denominator,” so-called.
But it’s not only that. It’s also sprinkled with new content or fresh content or tangential content or peripheral content – “surprises” we might call them. Songs that add “spice” to the whole and make that whole more “interesting” based on the expertise and the instincts of the human beings who program the station.
You see, whether digital platforms wish to acknowledge it or not, there’s a danger to being rewarded with only what you already like. In the literature they call it a “filter bubble.” The stream of filtered content chokes off serendipity, chokes off surprises, chokes off unexpected delights. It sends you down a rabbit hole of self-satisfying sameness.
The technologists call this “overfitting.” It’s when the model so accurately predicts things that you are very likely to prefer, it ignores that which you don’t know you’ll like until you experience it first. And this is the great danger inherent in personalization: By making things intensely personal you can actually make things intensely boring.Great danger of personalization: Making things personal can actually make them boring. Click To Tweet
Who hasn’t gotten bored of the same Pandora station they once created with such delight?
Who hasn’t seen one too many sequels of a favorite movie franchise?
Consumers don’t want only stuff that’s similar to what they already like, they also want stuff they don’t like at first but will learn to like in time.
A lot of relationships (and marriages) begin that way, after all.
So does every hit movie or TV show that isn’t based on a pre-existing hit movie or TV show.
In a world of filtering and personalization, consumers – listeners – will value that which stands out; that which provides a bit of challenge and a dollop of surprise. It’s why every listener says they want “variety” in their radio stations – they are literally telling you to avoid the boring drumbeat of sameness.
Computer algorithms and music genomes are really good at reflecting your preferences back at you. They’re not so good at predicting what you don’t like but will.
That’s where radio can come in.
#pandora #overfitting #radio #markramsey #netflix #filterbubble #Media #radioindustry #spotify #personalization