I Am The Gatekeeper (And That’s Okay)

John Schaefer (photo credit: @thesundncekid)

Many years ago, when “New Sounds” was still new (and dinosaurs roamed the Earth), I played a few tracks from a Japanese musician named Osamu Kitajima on the show.  At that time, WNYC-FM was still largely a classical music station; “New Sounds” was on at 11pm, when all the grownups had left, and I was accustomed to getting mail from older listeners who didn’t care to hear anything after Brahms.  But the Kitajima record provoked one elderly gentleman to write in to say that it was the best thing he’d heard since Enrico Caruso, some sixty years earlier.

I’ve been thinking about that listener and that letter recently, as news comes of AI bots developed to do the work of local radio announcers, and algorithms define what many of us listen to.  Music can do many things: it can calm you down, rile you up, move you to tears… but sometimes, music can be startling.  Those can be some of our most memorable musical experiences – the time when we heard something totally unexpected, something that took us unawares.  Look, I understand the utility of a good algorithm.  I have discovered any number of musicians and songs that I now love by clicking on something that was sitting on the bottom of the Soundcloud or Bandcamp page, or in the list of videos that runs down the side of YouTube.  But I’m willing to bet that no bot or algorithm would take an Enrico Caruso fan to Osamu Kitajima-land.

Music can do many things: it can calm you down, rile you up, move you to tears… but sometimes, music can be startling.

John Schaefer

I’m willing to bet that no bot or algorithm would take an Enrico Caruso fan to Osamu Kitajima-land.

John Schaefer

AI and algorithms work on logic.  Music, and discovery, work beyond logic, in the realm of intuition and inspiration and chance. 

John Schaefer

“I know what I like” is a perfectly reasonable way of approaching music.  For those listeners, a good algorithm will be decent company, and will occasionally throw in related things that offer a little of that sense of discovery. But for more curious listeners, and especially for anyone straying far from the world of pop, that simply won’t do. 

John Schaefer

When it comes to the democratization of music in the digital world, the theory differs greatly from the practice. 

John Schaefer

The further you get from the musical mainstream, the more important it is to have a guide, if only to get you started. 

John Schaefer

Turning on the radio means inviting someone into your home, your office, or your car, and sharing that space with them.  Late at night, if you’re alone in your room or your car, this experience is heightened.

John Schaefer

Curiosity is one of humanity’s greatest traits, and perhaps one of its most undervalued.  I don’t think there’s an AI program that can recreate that.

John Schaefer

AI and algorithms work on logic.  Music, and discovery, work beyond logic, in the realm of intuition and inspiration and chance.  I was not trying to convert old opera listeners into Japanese electro-ethno-pop fans.  That example was a total fluke – the element of chance at work.  Now, with more music available to us that at any time in history, the potential for surprise, for the thrill of discovery, is greater than ever.  But when almost everything is available, where do you even start?

This is where the editors, the gatekeepers, come in.  They’ve always been there:  record companies, record stores, radio programmers, music critics.  In theory, all of these people stood between you and your freedom to choose what you’d hear next.  Each one essentially said, “you will hear this”; but the subtext was, “and if you want to hear something else, go find it yourself.”  And for many folks, that was fine; “I know what I like” is a perfectly reasonable way of approaching music.  For those listeners, a good algorithm will be decent company, and will occasionally throw in related things that offer a little of that sense of discovery.

But for more curious listeners, and especially for anyone straying far from the world of pop, that simply won’t do.  Back in the day, we were the ones decrying those gatekeepers, lamenting the major labels’ desperate search for the next Thriller, and complaining about (but still watching) the Grammys’ annual parade of commercially safe pop.  So instead we took note of the record label that consistently did interesting stuff, the critic who made room for musicians no one else seemed to notice, or the DJ who blew your mind that night when she followed It’s A Beautiful Day with Lothar And The Hand People (and yes, those are real band names).  These people were our guides, because even in the analog age, there was already more music than any one person could handle.

People like that are needed now more than ever, because when it comes to the democratization of music in the digital world, the theory differs greatly from the practice.  Yes, countless additional artists and recordings from all genres, time periods, and parts of the world have become available to us, but somehow we find the digital system behaving in ways that look suspiciously like the old analog one.

It has always been a challenge to discover experimental music, contemporary classical music, and any other type of music that is unlikely to make much headway in the commercial sector.  The further you get from the musical mainstream, the more important it is to have a guide, if only to get you started.  That was the idea behind “New Sounds” all those years ago (forty and counting): to open a door to different musical possibilities – for rock fans bored by tight playlists, jazz listeners orphaned by the loss of their music on the radio, even the classical aficionados who’d fallen asleep and forgotten to turn off the radio.  I saw it as a door, not to a room, but to a hallway full of more doors, with different sounds and traditions and ideas in each room, many of which revealed themselves to be connected – because that was how it worked for me.

Could an AI program do that?  Well, probably, eventually – if someone had the desire and the funding and the time to do it.  Of course that someone would have to know that part, or more precisely, those parts, of the music world and so that person would be the gatekeeper and this is making my brain hurt just thinking about it.

But there’s something else at work here, something specific (I believe) to radio.  You may understand on an intellectual level that other people are hearing the same thing as you; but emotionally, it feels like it’s just you and this other person who is playing music for you.  Turning on the radio means inviting someone into your home, your office, or your car, and sharing that space with them.  Late at night, if you’re alone in your room or your car, this experience is heightened.  “New Sounds” began as a late night experiment; these days, I know (again, on an intellectual level) that people are listening to it online at many other times of the day, but I have always felt like it’s still a late night show at heart.  And anyone who has ever hosted a late night radio show will tell you tales, both horrifying and wonderful, about the connections that individual listeners make with this disembodied but still clearly human presence in their life.

Now we get word that some of the big media corporations that own radio stations across the country, filling them all with programming from one central location, are realizing the importance of sounding local and are experimenting with AI as a way of “localizing” their fare.  I have no doubt that this can work, within limits.  A simulacrum of a radio announcer can impart local weather info, concert listings, and the like, and at some point I’m sure that simulacrum will sound almost indistinguishable from a human announcer reading the same material.  Tellingly, no one is yet talking about an AI bot crafting a music show and acting as a guide/companion for the listeners.  That doesn’t mean such things can’t happen, only that they’re probably years away (he said, in a burst of implausible optimism).

I guess it comes down to FOMO.  If FOMO means fear of missing out on what everyone else is doing, or listening to, or watching, then algorithms and AI might be just the ticket.  But FOMO could also mean fear of missing out on everything else – the new, the distant, the overlooked… you know, the stuff that might make you think, “that’s the best thing I’ve heard in sixty years.”  In that case, FOMO is just another term for curiosity.  And curiosity is one of humanity’s greatest traits, and perhaps one of its most undervalued.  I don’t think there’s an AI program that can recreate that.