Shuffle has always been one of Spotify’s most-used features, and also one of the most misunderstood. For years, listeners have debated whether Shuffle is “truly random,” often noticing patterns or repeated tracks that didn’t feel random enough.

The Spotify Engineering Blog provides  insights, technical deep dives, and behind-the-scenes stories from the engineering team at Spotify. Covering a wide range of topics, including backend systems, data processing pipelines, and frontend architecture, the blog offers a glimpse into the technology stack and engineering practices used to power one of the world's leading music streaming platforms. Developers can learn about scalable infrastructure solutions, innovative engineering techniques, and best practices for building high-performance software systems. By following the Spotify Engineering Blog, developers can gain inspiration, knowledge, and practical insights to apply to their own projects and endeavors.

Spotify Labs

Spotify redesigned its shuffle feature to balance statistical randomness with user perception. While the previous implementation used pure randomization (Mersenne Twister), users complained about repetitive patterns. The new 'Fewer Repeats' system generates multiple random sequences, scores them based on listening history and recency, then selects the freshest option. This approach maintains mathematical randomness while reducing perceived repetition. Premium users now get this as default, with classic random shuffle still available as 'Standard Shuffle'.

Shuffle: Making Random Feel More Human

How we balanced statistical randomness with listener intuition

Our approach: fewer repeats, same randomness

It wasn’t an article about how to generate random numbers, but rather one that asked the more fundamental question of what randomness itself is. It was interesting.

I suggest to you all to take a look on this algorithm : Fisher–Yates shuffle. It achieves the best possible time and space complexities

Isn’t this algorithm very complex? Sort the list in order of last played, in descending order, take the part of the list that has the oldest items (not recently played) and shuffle it in place.

So basically, a machine learning that design a list of items that user will perceive as random but actually fabricated to seek out maximum engagement?