blog.maestro.fm

Music recommendations. They’re all the rage. They are the future of how we discover new music and expand our musical horizons. In fact, last year alone, 46% of music consumers cited free online streaming sites as their sources to learn about new music. But it should be noted, there is a significant difference between the two main types of recommendations that are being featured on social music portals now.  As the two sides part, companies can choose to go one way or another: human vs. technology-based music recommendations.

At Maestro, we have clearly sided with recommendations based on human listening habits, which we believe leads to more exposure to personalized discovery than algorithmic scenarios and calculations.  It’s just that with the current state of technology, human being are better at distinguishing subtle nuances of why you might like one song over another.  In addition, humans immediately consider cultural factors when making recommendations, which is not only very difficult for an algorithm to identify, but also nearly impossible to keep updated. Yes, there are lots of things that computers, and computer programs, can do better and faster then people.  C’mon, we’re a music technology company, of course we know that. However, sometimes there are scenarios which, when left to the whims of a machine-based result, lose something in the process.

Many sites out there are leaning towards the digital brain of recommendations. I don’t want to list off a bunch of names, but there was a recent posting on coolfer.com that called out muffin.com as a prime example of this.  If you want to see a more mainstreamed example though, go give Pandora a try, plug in a song, and see how your results turn out.  Now, find one of your playlists on Maestro, and check out related playlists which are connected to what your peers are listening to.  I bet that the personal touch of an individual is going to lead you to much better results nearly every time.

Maybe a more specific example would help.  I like Michael Jackson’s hit song Don’t Stop ‘Til You Get Enough.  Yea, that’s right, I like the song.  Why? I don’t know why, maybe because it makes me laugh, maybe because it reminds me of a middle school dance, maybe just because I grew up in the 80’s and I’m naturally biased towards it.  Regardless, if we plug that into an automated recommendation engine, I get time-period matches like Holiday, by Madonna (um, no thank you) and similar artists like Lionel Richie and his plethora of greatest hits (absolutely not, buster).  However, if I check into Maestro, I’ll find this song leads me to playlists with songs like When Doves Cry by Prince, and We Will Rock You by Queen.  Now that’s more like it!

While recommendation technology can result in a greater number of songs referred to you, the quality of songs recommended by like-minded peers is much higher.  Think about it this way: if you were going to get to listen to just one song, what would result in music you’re more likely to enjoy…a friend of yours assigned to the task, or randomly tuning into your favorite radio station?  Sure, inevitably, the radio station will eventually stumble onto songs here and there that you enjoy, but if you want a result that hits your tastes right on the head, a person who shares your musical orientation is going to be far more successful.

When it comes to music recommendations, following the products of other listeners is hands down going to yield better results then an algorithmic technology designed to spit out results by the boatload. The difference between the two is paramount to the quality of recommendations you receive…people understand what other people might like far better than any complicated and calculated matching scenario can.

Meanwhile, hop onto Maestro.fm, and find a few people who share some similar music tastes to your own, and check out what they’re putting into playlists.  You might just find yourself stumbling onto tons of great new music you didn’t even know about!