Blog

Context (Not Content) is King

Posted by Jeff Yasuda on Oct 27, 2017 3:32:47 PM

Miles Davis always said, “Music is the space between the notes. It’s not the notes you play, it’s the notes you don’t play.”

As we navigate through the endless stream of content that invades our inboxes, social feeds, and points of media consumption, we often are looking for that “North Star” to guide us and to help us determine what content is actually important and relevant. I’ll argue that context is that “North Star,” perhaps that important space “in between” and it is wildly more deserving of a royal title than content. Moreover, upon revisiting the development of the Web, the importance of context becomes more apparent.

Recall that Web 1.0 was a world where everybody was able to access content. I remember using browsers like Mosaic to look up inane bits of information, like beer recipes and travel tips, that early internet pioneers sprinkled across the nascent network. Most of this content was just fun stuff that early web developers generated as they refined their UI skills. However, the quality and quantity of content increased exponentially and eventually, you could find whatever information you wanted via the web.

Web 2.0 introduced the concept of UGC, or user-generated content, where anyone with an internet connection was able to publish pictures, blog posts, music, videos, etc. A large amount of this “content” was utter garbage or infringed on copyrights. Soon, an actual problem emerged where there was so much useless information, it became difficult to determine what was important and what was a complete waste of time.

While there are many interpretations of what Web 3.0 is all about, I think one of the most important components is the need for filters to determine what content is valuable for the recipient.

The Stone Foxes at Golden Gate Park (credit: Joe Barham)

The world of filters comes in two basic categories: algorithmic and human. 

In the algorithmic arena, the most popular is the collaborative filtering mechanism, used by companies like Amazon, Netflix, and Pandora. This algorithm creates recommendations based upon what similar people have previously chosen. These collaborative aspects are designed so that multiple inputs can be aggregated and summarized to determine what users like.

Sounds good, right?

In general, algorithmic methodology can often times be effective. However, the algorithm frequently is just not good enough. What the algorithm lacks is context.

Internet radio giant Pandora employs about 100 musicologists (most of whom are musicians themselves) who review and map 10,000 songs per month based on hundreds of attributes. Known by many as the Music Genome Project, these attributes might include gender of singer, beats per minute, genre, musical era, and types of instruments. Each song has between 200 to 450 attributes. These attributes are then used to create song mappings to group similar songs together. These mappings are then inputted into their collaborative filtering algorithm and allow listeners to enjoy similar sounding tracks in selected stations.

For example, if you type "Baba O’Riley" by The Who, the algorithm might conclude that certain songs from bands like Led Zeppelin, Wolfmother, the Rolling Stones, AC/DC or even my local fave The Stone Foxes would be good candidates. Simply enter a song or artist name and a station of similar songs will be created to give you a great, personalized listening experience.

The methodology certainly sounds pretty impressive and it is quite impressive, given that Pandora presently has approximately 100 million users, according to company statements.

However, it is often the case that the songs recommended are often not what listeners were expecting. This is because the algorithm lacks context. Recently, I created a station for indie folk phenom, Bon Iver, on my Pandora account. The algorithm created some pretty solid recommendations, but at one point, a John Denver song came on. Now the only reason that I realized this was John Denver at the time was because my kids came in saying, “Hey! I know that song. That's the song about the jet plane.” We finished the song and because the only similarity John Denver and Bon Iver have is that they both have three vowels in their names, I turned off that station.

Just because a user may like both John Denver and Bon Iver does not mean the two artists are similar, and unfortunately, an algorithm is not able to make that distinction.

Musicologist Dario Slavazza of Feed.fm (credit: PopUp Magazine)

For me, I would get recommendations from a human being rather than an algorithm any day of the week. Whether it's the best burrito in SF, a secret surf spot or an underground bar where you can hear old jazz standards, I'd follow the advice of an actual person before paying heed to a computer generated suggestion. The most important reason is that a human being would provide context around the specific recommendations.

A human being would understand what I like and would take those aspects into consideration. Even if they didn’t know me well, I could still choose a person that likes a similar music genre that could potentially introduce me to more artists and songs in that genre.

Musicologist Dario Slavazza spends his days working with brands and how to best connect them to their customers by using the right music.  He is the lead music curator for my company Feed.fm where he reviews thousands of tracks across dozens of genres, multiple moods, tempos, and numerous other attributes. Most importantly, Slavazza needs to take into account a deeper understanding of music, its connotations, and the social context of that music. 

“No one is just listening to a song,” he explains. “What they’re listening to is a story and a narrative that includes things such as their friend group, locale, and generation. Music has cultural touch points that require a deep understanding of events and relations that are seemingly benign on the surface but make all the difference.” In many ways, it’s that cultural context that can attract customers to a brand or drive them away.

Providing context around content is not a new idea. Radio DJs in the past have talked about upcoming songs, why they're so great, and which of those great aspects to listen for in a piece of music. This commentary has always hooked me because it gave me the information I needed to actually understand songs.

A few years ago, many of the music streaming services tried to individualize themselves by boasting the amount of content they had in their libraries. One service said it had 5 million tracks, one said it had 7 million, one said it had 30 million tracks. But as everyone knows, it would be impossible to listen to that amount of music in an entire lifetime. So again, I return to my initial point that content is everywhere, it's the context, however, that points each user in their own right direction in order to find what music they’re looking for.

Laurence Scott of The Golden State Warriors (credit: Scott Strazzante, The Chronicle)

For the past year, Feed.fm has been working with the Golden State Warriors. Not only do they curate music that they feel is important to the fan base, but they have meticulously taken the time to actually provide context for every playlist and the songs in rotation.

I had a chance to catch up with Laurence Scott, Senior Director of Content and Entertainment Programming. To Warriors fans, he is the beloved commentator and on-air personality. He’s been working in media and journalism for years and has successfully brought meaning to hundreds of video and audio clips over the years. He has to make tough decisions on what content is relevant to the fans and what should be scrapped (as painful as it may be). Here’s what he had to say about the importance of context in a world of endless content.

"Having worked in the media before joining the Warriors and now communicating the team’s messages direct to our fans, it has always been important to deliver content where context is the primary driver. How all of our team news and information, music and audio content relates to each other in a meaningful way to our audience is through the stories we tell to weave it all together and through this, the context is key."

Storytelling is an important contextual glue that can piece together seemingly unrelated bits of content.  

Scott recently interviewed Warriors’ broadcaster, Jim Barnett, and asked him about his stint with the original New Orleans Jazz team.

“I Started with Pete Maravich,” said Barnett. “That’s what was really fun was being around with Pete Maravich, to tell you the truth, just the first few weeks here. Getting to know him and playing on a team with that kind of star.”

After Barnett elaborates on his time with the New Orleans Jazz, Scott says, “Some things never change, like great music. And your choice would be a Fats Domino song.”

“Always like Fats Domino,” replies Barnett. “And he made a song called ‘Walking to New Orleans…’ (He) Still lives here in New Orleans. One of my first two or three records was a Fats Domino, Blueberry Hill.”

Scott then rolls the tape and it’s like you’ve time traveled back to Barnett’s years with the New Orleans Jazz. Suddenly images of a New Orleans from days past would spring to mind as the song meanders through a man’s journey through life and old neighborhoods in the historic city. Barnett’s connection to the song and the city of New Orleans gave context to a song from our past and made it relevant to fans. Fats Domino may you rest in peace and thank you for your amazing music.

You can check out how Laurence Scott is using context around content at Warriors Sound HERE.

Jason Hirschhorn's REDEF (credit: REDEF)

In many ways, today’s challenge is finding the right curator that can help me find what I want. 

Jason Hirschhorn and his team meticulously scours the web in the early hours of the morning to create an email newsletter called Redef that usually is the first thing I review after waking up. He covers everything from politics to tech to health and sex across tens, perhaps hundreds or media sources. In the early days, this was a passion project for fun and people had to fight to get on his email distribution.

In 2006, he turned the passion project into a real company called Redef Media. He opens every email with a personal note which provides his thoughts for the day and an overall contextual background around the articles that he’s chosen as noteworthy. He is a master curator and it astounds me that he is able to get up so early every morning to make his picks on a consistent basis. Please check it out HERE.

 

Samuel Taylor Coleridge wrote, "Water, water, every where, nor any drop to drink." We often float without a compass in a sea of meaningless content.  Social media, paid media, content marketing, fake news (I couldn’t resist) and legitimate content sources compete for our attention at every second of the day. Content is simply everywhere and the excess surplus has rendered most of it worthless. There is hope, however. Context is that North Star and it is king. As Miles said, it may not be about the songs in a playlist or pictures in a photo stream. It’s really more about the spaces in between explaining WHY content is important. Context is all the rage.

Topics: Feed.fm Blog