Synthesizing Steve: Music bots are no match for boutique expertise

One upon a time, I relied on my friend Steve Gedrose for musical advice. He owned a boutique CD store called Rose Music down on 4th Avenue, and he was a walking musical encyclopedia.

One upon a time, I relied on my friend Steve Gedrose for musical advice.

He owned a boutique CD store called Rose Music down on 4th Avenue, and he was a walking musical encyclopedia.

But more than a human tome of raw information, Steve was an attentive music enthusiast and empath.

He maintained a elaborate mental record of his customers’ likes and dislikes, passions and apathies. As a result, he was almost always bang-on with his musical recommendations.

I’ll never forget this one day I walked into Rose Music and he jumped out of his chair.

“Hey, Andrew,” he exclaimed. “I found a disc you’ll love!”

A fat funk beat overlaid with DJ Logic’s erratic vinyl scratching exploded onto the Rose Music sound system like an aural kaleidoscope.

As I gasped in awe, his other customers cringed.

Eleven years later, Medeski, Martin & Woods’ Combustication remains an essential album in my collection.

These days, most of the Steve Gedroses of the world have shuttered their independent stores as the digital music revolution drew their customers online.

And we are all poorer for that.

The Stevebots we’re now forced to suffer – those online music recommendation programs – are far less effective than the humans they replaced.

Take the iTunes Store “Genius” feature.

Or, I’ll just call it the anti-Genius, since it seems to employ such a crudely logical set of recommendation parameters.

Like, if I listen to Plants and Animals, then Grizzly Bear is going to appeal to me, right? After all they’re both neo-hippie jam bands that kind of sound the same.

If only it were that simple. Musical appreciation is much more than a comparative analysis of trends or genres.

Through conversation Steve had learned that, while I wasn’t a huge Miles Davis fan, I love the trumpet player’s jazz-fusion A Tribute to Jack Johnson soundtrack.

Steve also knew I was picking up some rock and hip hop albums at other shops.

So when he read a review of the then-underground MMW album that sported hard-rocking drum beats, killer B3 Hammond riffing and a popular DJ to boot, he knew the album was for me.

The iTunes anti-Genius would never have figured that out.

It would have suggested something obvious like Wynton Marsalis, because he’s a great jazz trumpet player, sort of like Miles. (Whereas Steve also knows I don’t dig such clean playing.)

So while Steve combined a number of my musical interests to introduce me to a new sound I wasn’t familiar with, the iTunes anti-Genius just assumes I want the same-old, same-old.

Technologists will be working until the end of time to synthesize Steve’s musical brain into an effective Stevebot. But they’ll never succeed.

Most artificial intelligence as applied to musical comprehension focuses on the qualities of music like rhythm and melody and patterns of individual consumption.

Resulting systems rely on some form of statistical analysis, which is the core problem. Musical enjoyment is not a structured beast.

Sometimes we like a song, well, just because. It makes us feel good, or sad, or it just plain resonates with our very being.

You can’t apply statistics to that.

So artificial intelligence misses one very important aspect of musical enjoyment, and that is the act of enjoyment itself.

In the 1993 anime series Astro Boy, the Japanese term kokoro is used to describe certain innate human characteristics that robots have been programmed with, such as emotions and empathy.

Kokoro lends Astro and other robots the capacity to recognize and respond to a broad range of both overt and covert sensory input.

In other words, kokoro gives robots the ability to be illogical. Kokoro gives robots a “gut.”

Kokoro is, of course, the state of being human, with which we are all naturally endowed.

It’s something that really can’t be taught or, I would argue, programmed.

That’s why technology-based musical recommendation systems will never be as good as Steve Gedrose.

Steve didn’t used to sit around all night calculating my musical interests, analyzing my purchases for patterns.

Based on what he knew about me, and his consummate understanding of the art and craft of music, he just had a gut feel for what I might like.

So while we’ve gained a magnitude of convenience from the digital music revolution, we’ve paid a great price.

We’ve lost Steve and his ilk.

The online Stevebots we now suffer demonstrate about as much musical sensibility as a Roomba bouncing around from couch leg to wall in search of dust bunnies.

Fortunately, I know Steve Gedrose has moved on to bigger and more important musical pursuits.

But my musical library is poorer for his absence.

Andrew Robulack is a Whitehorse-based freelance writer and technology solutions consultant specializing in Macs, the internet, and mobile devices. Read his blog online at