I want to bring some of the different view-pionts to light here, and then really analyze them further to reveal the true meat and potatoes underneath it all – what really matters, and what doesn’t, and introduce a way of thinking about it that just really makes more sense. There are many misconceptions floating around right now that I feel I owe to people to clear up.
First of all, let’s look at the facts. Fact 1: Music recommendation is the future. Reason: Searching for music wastes time, and sucks. Fact 2: Nobody does it perfectly right now. Reason: I can’t find new music that I like consistently with any engines out right now. Fact 3: A solution exists. Reason: If people can make the problem, they can make the solution.
Let’s take a look at some of the different view-points going around the web about this topic. This first excerpt is a Music Think Tank interview with Berklee College of Music VP, Dave Kusek.
“There are lots of people currently working on this challenge. Someone is going to find a really slick way to find music that we are truly interested in, and that software will become invaluable.
The Echo Nest is getting close. They are combining what people are talking about on the public web, with purchase data and behavioral data. They are also providing a very sophisticated musical analysis of the songs and relating that to the social graph.”
When I first read this I almost laughed and cried at the same time. I mean, it sounds reasonable…let’s take data about songs and data about people and put it together to get a good song recommendation engine.
Unfortunately, trying to come up with an equation to accurately model human behavior, is simply an impossible task. People just change too much. Humans are themselves, infinitely variable, variables. – say that 8 times
Finally, today I read something smart about the topic, from TheLefsetzLetter blog. The author wrote:
“It’s not about the iPhone app. It’s not about the ads on the Website. It’s not about technology at all. That’s what’s been wrong with the past decade, it’s been about science and marketing, not music. Everybody wants to get rich. From the wannabe acts to Tim Westergren and Pandora to the endless purveyors of music distribution platforms. No one’s focusing on the music! And that’s why we’re hurting.”
First of all, he couldn’t be more right about needing to focus on the music first in order to get a real solution, however, music isn’t just about music. Think about what it really IS. – It’s people. We need to focus on the people. The technology is important, but right now many websites are over-thinking it with too much technology. They need to start getting the social aspect – the people – involved.
This next excerpt almost made me sick..and it’s actually a blog post that was in response to the previous excerpt.
“While he’s correct that music recommendation and discovery is difficult, partly due to the increased amount of music content (i.e. signal-to-noise), democratically filtering out which songs and artists are the rising ones based on quantifiable and qualifiable information will require technology. For example, The Hype Machine does this well by mining the blogosphere and Twitter. The question is, will a human element remain necessary to the process? If so, what does that look like? Who are the social gatekeepers? Does their influence on an artist’s popularity help or hinder music discovery? How are the social gatekeepers determined?”
Okay, slow down. The Hype Machine hardly does a good job of filtering out the music. Not everyone who cares about music is blogging about it. The data will always be incomplete, because EVERYONE listens to music, so everyone should have a say. Second of all, technology is needed, yes, but not to take away the human element! Are you crazy!
Technology should exist only to empower the human element, not to do away with it..this isn’t the Matrix. The results will ALWAYS be better when a human input is required. Don’t worry, I will offer my own solution at the end of this post instead of just ripping on other people’s.
This last excerpt is another from the Dan Kusek interview, where he actually said the answer, without even realizing it:
“Just look at SXSW. There are 2 or 3 thousand bands playing. There is no easy way to figure out which ones I really want to go to listen to. I am counting on recommendations from my friends and buzz in the media to help pick what I should listen to. That is true at SXSW and also true on a daily basis. Knowing what to listen to is more important than having it in your collection. That is becoming more true every day.”
The part I want to highlight as bright as I can, is the part about friends and media recommending songs. That is the best, surest way to guarantee that a person will listen to your music…have their friend tell them about it.
So I make that sound like it’s an easy thing to do..but how can you really do that? I’m actually in the middle of testing software right now that will make that possible. I have explained it a million times on this blog, but I will keep explaining it until people really catch on.
This is how it works. There is a social network platform, similar to Twitter or Facebook, but just for music. You have a radio player that stays with you everywhere you go on the site. You “follow” people who you share a similar taste in music with – friends, favorite musicians, media, ect – and anything that those people like enough to save into their own playlist, automatically gets sent to your radio. If you like a song you hear, you playlist it, and it automatically gets sent to anyone following you. There’s a lot more detail in how you’re able to filter the music, but that’s the basic concept.
Now, think about this for a little while. There is 6 degrees of separation between everyone. If your music is good, it will get saved to playlists, and shared automatically, like an automated word of mouth. It will spread throughout hundreds, or thousands of playlists, each one branching out to increasingly more.
The reason this WILL become the new model is because this is both a form of distribution, and promotion (for free & without a label), AND..it could even make piracy irrelevant – the artists get paid based on ad revenue and popularity, and the fans continue to get and share all their music for free…Yea, that’s why it’s the new model.
It’s called Beat-Play, by the way. Check out more info here.
Song recommendation MUST include the social aspect. The reason Pandora fails is because it’s based on boring, static, equations. That doesn’t represent music, and it certainly doesn’t represent the people. Generating playlists based off of your actual friends, or at least people you trust to give you what you like, is way better than trusting a machine. I’m also an independent artist myself, and I created this model not so I could get rich, but so I can continue to have a music career. Without this tool, it’s almost impossible.
People’s thinking is for the most part, in this industry, completely backwards when it comes to this topic. They are over thinking, and over-teching this issue. If we, instead, start trusting our users..or fans..a little bit more, we can begin to deliver a rich user experience that actually exceeds expectations and offers REAL solutions to ALL of our music problems. This is very possible.
Beat-Play, like I said, is being tested now, internally, and will be released for beta later this spring. I hope everyone who has interest in this shares this journey with us in order to allow this experiment to reach its full potential. We are getting very excited..and the answers are right at our fingertips. Look out for them ahead.
Written by: Dante Cullari Founder & President Beat-Play, LLC