A screenshot of 'Launch on Yahoo!'
Aug 9, 2022

An Ode to Launchcast, the Best Music Recommendation Engine

I’ve long been an avid music lover, from days as a child dancing around and singing along to Michael Jackson’s Thriller album, to Dave Matthews Band concerts in high school, and to DJing at the campus radio station in college.

Today, I continue to love discovering new music. The music of my formative years continues to hold a special place, but the excitement of discovering something new and extraordinary is infectious.

So, I don’t speak lightly when I call Launchcast, a product that was a streaming music service at it’s heart, the best music recommendation engine to date, even though it’s recommendations ended more than 13 years ago.

One of the few surviving Launchcast screenshots, in German

You’re probably thinking that, surely, with all the music services available today, somebody’s created a better approach in the last 14 years. So, let me outline the two key features that set Launchcast apart:

Five-Star Feedback System - I’ll refrain from calling this a “Five-Star Rating” system to help clarify how it was used. For each song, you could indicate from 5-stars (“I love this”) to 0-stars (“Please, no…never again”) how much you liked the song. While I know Five-Star systems have fallen out of favor for “Likes”, primarily for engagement purposes, for me, there’s a big difference in feedback between “This is good…I could listen to more” and “Yea, just leave this on repeat!”.

Song, Album and Artist Feedback - Not only could you provide feedback on individual songs, but you could provide a five-star rating on the album the song was on, or the artist in general. Not only could you provide it, but the service would provide an estimation of those ratings if you had provided enough feedback on the tracks of an album or artist, giving you an opportunity to confirm or modify the rating.

The granularity of feedback that these two features provide so much more information to work with than today’s services can capture. For example, I can like a song on Spotify, and it’ll eventually recommend other songs by the artist, even if I’m not interested (and even then, it’ll keep recommending them unless I block songs I don’t like). However, with Launchcast, I could specify that I loved a song, but I wasn’t a fan of the artist otherwise. Or even that I loved an artist, but I might not be a fan of their “cover album”.

Because of this additional information, the quality of the Launchcast recommendations that were injected into the radio stream were typically high quality, and helped me to discover artists who were new to me in the 00’s that I might not have discovered or engaged with otherwise, like Billy Talent and My Chemical Romance, in a way that no other music has been able to match since.