Goodbye to 8tracks after 1000h of use?

I have been a big fan of 8tracks for 5 years. I seriously considered investing in them during their recent $2.1 M Series A crowdfunding. On a cumulative basis, I probably spent over 1000 hours on 8tracks. Yet I am not sure I will keep coming back.

8tracks lets you explore playlists through the use of tags (genre, emotions, artist name, …). These playlists are created by its users who assign the tags.

My primary use of 8tracks is to listen to music while at work. I realized that listening to a combination of chill, indie, minimalistic, happy and electronic music helps me focus and be more efficient. Unsurprisingly these 5 words represent my most used tags.

You might think these tags could also apply to you. However are we really thinking along the same line?

What sounds happy to me is possibly boring to you. “I’m going down” by EFIX & Henri Pfr truly trigger chill emotions to my ears.

Whereas you might consider that only the likes of Metallica — “Master of Puppets” do that. Or worse, that “I’m going down” reminds you of your ex and makes you cry.

“The best and most beautiful things in the world cannot be seen or even touched. They must be felt with the heart” ― Helen Keller

There is a reason why words are often not enough to describe our feelings and emotions. A lot is happening up there in our not-so-little brains — on average we have 70,000 conscious and unconscious thoughts per day. Now let’s assume a song takes 4 minutes. This means we are exposed to 200 thoughts while listening to it. What is the likelihood that my 3–5 words summary of the song matches yours — let alone for an entire playlist?

A tag approach on entire playlists was ideal when no other alternative existed. I got used to skipping many songs in suggested 8tracks’ playlists in the hope of finding a song which exactly matched my emotions.

However it is clear now that artificial intelligence and machine learning algorithms can vastly outperform humans in curating content. Hand over my 1000 hours of music history to a machine learning algorithm and it will find the new song that I really want to hear next when I need to focus.

Spotify, Soundcloud and the likes already make song recommendations using AI. I imagine that with data such as:

• correlation between searched tags and songs liked/listened to end-to-end
• cumulative listening time per song
• genre and artist name of songs liked
• moments the “like” / “next song” / “volume up” buttons were hit while the songs were being played (to break down songs into its parts)
• songs sequence per tag
• social engagement per song (post on FB / Twitter / etc)
• …

empowered computers with proper AI algorithms will become our new favorite DJs.

8tracks can remain one of my favorite music players if it adapts. Suggestions for a hybrid human-computer 8tracks — use AI algorithms to:

• Tag individual songs (not only playlists) to improve accuracy of recommendations
• Create curated playlists personalized to each listener based on the searched tags and user history (+ provide option to users to select between pure human-made playlists and hybrid playlists)
• Provide adequate song suggestions to creators of playlists in harmony with the set of songs already listed, as well as adequate tag suggestions for newly created playlists
• Improve songs sequence in line with tags (feature extension: mood changing playlist, e.g. from sad to happy)

Only 8tracks team can figure out which above feature needs to prioritized first — if at all. They have the data and the access to the user base. But I feel they should run their customer tests now and start building AI-powered features at the earliest.

media, …

If you want more content right now, check out which single feature is missing to get you hooked on UberEats or Postmates here!

Enjoyed that read? Click the ❤ below to recommend it to other interested readers!