Predictive Taste Algorithms Are A Democratic Revolution

By Graham Peterson

I see predictive taste algorithms get called creepy all the time.  They’ve been held up as proof that we are now living in Orwell’s 1984.  Glen Weyl wrote a small bit about how they mark the Final Emergence of the centralized coordination of information, contra those who believe socialist calculation is not possible.  But let’s calm down for a minute and trace back how this all happened.

Since we’re working in a world of anecdote and conjecture already, let me share my own.  When I was 15 years-old, I was living in a town in Wisconsin with 10,000 people.  There were five radio stations, and kids listened to two of them.

I wanted to be different, and decided I liked electronic music.  Luckily I had the internet.  This was when CDnow.com was still beating out Amazon.com in music sales.*  CDnow.com kept track of who was ordering which CDs, and told you.  That meant that I could have conversations with people half way around the world about which artists (say) resembled Brian Eno’s ambient compositions.

That premise from 1998 is how predictive taste algorithms still work.  The conditional probabilities have gotten more complex, but it’s the same basic idea — take more detailed, granular data on who’s buying what in little variegated pockets/clusters of social networks, and deliver that information to people quickly and cheaply.

It is “word of mouth” advertising streamlined and improved, because instead of talking to your neighbor over the fence, who you might happen to share the same taste in beer or fertilizer or knitting needles with, you can talk to people who happen to share your taste in Found Noise music that would annoy the piss out of 99.9% of the rest of the world.

I hate to keep hammering on critical theory, but it’s so much fun, because it’s so totally wrong, and still so widely believed.  Again, critical theory emerged when cultural products like movies and music had to travel through centralized channels because of technological costs (this was, at the time, a notable improvement over only aristocrats having access to professional music).  By the 1990s when I was growing up, people still clung hard to the idea that everyone was a brainwashed idiot because Top 40 radio was telling them to like the same song.

Now think about it for a minute: where did the Top 40 come from?  It came from the radio’s best guess, conditioned on what people had consumed prior, about what people would like next month.  Except because of the constraints of the technology, radio outlets could only get information about huge aggregates of people.  “Mass culture” as it were, was built right into the technological platform.  It had nothing to do with any nefarious plot to degrade people’s diversity and get them addicted to the same Soma.

The only difference between predictive taste algorithms used to produce 1940s radio, and those suggesting 1995 CDnow.com purchases, was the fidelity of the information about people’s tastes that radio producers could sample from.

As soon as the technology improved (cheapened) and allowed music producers and marketers to get a better sense of what people liked, and market to increasingly differentiated varieties of those people, at low cost, they lapped up the opportunity.  In fact the precision and variation of the development of music tastes has evolved such that many people don’t even buy albums any more, and just consume individual songs that appeal to them.

It continues to disturb many people that A Giant Company facilitates this conversation about tastes, but why should it?  The interests of the company are compatible with yours: they want to get you the best information they can about what other people, like you, like.  In fact, they even provide a platform for people to bash their products right on their own sales pages.  Because it’s the right thing to do, and because doing the ethical thing is often (and here’s a real shocker) conducive to commerce.

Predictive taste algorithms are unequivocally not an example of companies telling you what to think and like; they are an example of companies facilitating a conversation between you and other consumers, who are all democratically working out, right there in the agora, what to think and like.  Even weirdos who think ambient noise is music get to participate now.

*Incidentally, I and most of the free world thought the game plan Amazon had in mind was ridiculous: sell everything under the sun to people, CDs sure, but not a full department store.  Turns out I just bought a shower curtain and dress shirt from them.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s