“I would guess, looking at music choices, you could probably predict with high accuracy a person’s worldview.“
If you're passionate about music and have a good ear, you can't predict the taste.
Tastes are also hard to generalize. Just because I listen to a certain x, doesn't mean I like y or will click on one particular banner over another. A lot of the world is spontaneous, curious, and random.
Unless you're ordinary…
Algorithms curb the discovery process. Amazon tries to recommend you books. Pandora examines your listening behavior to recommend music. Art.sy tries to introduce you to new art based on your preferences.
Algorithmic predictions feel a bit like Google, crowdsourced information that displays results for what the masses are also looking for in the aggregate.
The information, art, and music DJs that really know their stuff ignore algorithms altogether. They have trusted sources and spend the time to find new and emerging sources to pluck gems from. These curators master the art of showing people what they know people will like and what they think people will like.
I believe everyone should research at least one category of art and dig into it as much as they can. That means scouring the Internet for niche blogs, listening to obscure podcasts, seeing what the DJs are recommending, and following influencers on forums and on Twitter.
Discovery is an active process, not a passive one. Turn off mainstream radio and find something new or rediscover something old. The real gems lie in the nooks and crannies. Predict what’s next, not what’s now.
music is the “fastest, most user-friendly way to influence and reset your brain networks without using an external substance.”
Familiar music can be distracting. Unknown, non-playlisted music, as in radio, seems to inspire working. This is probably why Pandora and Songza are so popular.
I work even better with unfamiliar music playing in the background.
I know how many steps I take per day. But these get recorded on a pedometer, not an Internet connected device with algorithms on the back end. As a result, I don’t get push messages telling me how close or behind I am to reaching my goals. I just know some basics.
I love data. It allows us to make wise decisions about where, when, and how to move forward.
But I still believe data is really bad at predicting human emotion. Music, for example, is hard to recommend. There are special algorithms in Pandora that suggest new tracks in accordance to our tastes.
Based on my own experience, rarely does Pandora play something interesting and of good quality. Music, like books and movies, is not something you can predict with precision. Human Genome projects are great for recommendations and starting points but they try to plan too much. And it’s because we let them to, outsourcing responsibility.
The best part about the analog world is discovering something that aggregated data can’t predict. Discovery occurs through randomness as much as it does through suggested data. The data doesn’t know you’re open to completely new things; it’s going to keep feeding you the same stuff within that niche.
Machines that dictate our action dictate our behavior. Plugging out, being open, is just as important as being plugged in. The best recommendation engine may be yourself.
Internet recommendation engines are now the way we discover things, whether it’s music, photography, or art.
At first we thought the best recommendation engines were our friends. Then it turned out those we follow on Twitter, Tumblr, and Pinterest were sharing far more interesting stuff.
On top of the social engine though are experts. These experts huddle in a room and break down a song or piece of art. Their analysis is “then fed into an algorithm” which seeds to users based on their interests. In Pandora terms, this is called Music Genome Project.
Art.sy is mimicking the same model, leveraging the brain power of curators to grade works of art that it can then recommend to users based on their interests.
This expert recommendation engine is now being used across industries to help people discover new stuff. The process is time consuming but the human element is sometimes greater than the smartest algorithm based on category generalizations. It pays to be more specific.