In the taxonomy of my personal Spotify playlists, there are two phyla: temporal and tonal.

Temporal playlists grow organically and without premeditation. Their purpose is to capture a sonic snapshot of a particular period, typically delimited by the beginning and end of the temperate seasons, thereby allowing me to relive the emotions of the time through its soundscape. When I find myself repeatedly listening to a song, I eventually add it to the playlist for the current period. In this way, these playlists grow quickly, all at once, and then not at all. Once a new season arrives, a new playlist is minted, and the old one becomes paralyzed, an audio time capsule frozen in time forevermore.

Tonal playlists are more similar to the conventional playlist in nature, in that their constituent songs all share some common trait. For example, the songs in my running playlist all possess a fast, consistent tempo, whereas those in my study playlist possess a soothing instrumentality devoid of lyrical distraction. These playlists are harder to develop, and songs must be sought out with intentionality and purpose in order to curate the collection.

Within the phylum of tonal playlists are two sub-classes. In order to name these, I'll borrow two terms from investing: passive and active management.

Most of my tonal playlists begin as actively managed ones, as I spend the time and effort to sift through the dozens of candidate songs to which a few are granted admission. However, with time inevitably comes neglect, and the playlist eventually becomes passively managed, a fate similar to that of the temporal playlist.

Except, while this fate is definitional for the temporal playlist, it is undesirable for the tonal one. An unchanging running playlist will fail to motivate me with the new and energizing beats I seek. A constant study playlist will eventually distract me with familiar and predictable melodies that disrupt my flow.

I found myself wanting a way to automate the active management of my tonal playlists: a process that would evolve their content dynamically, while holding their essence constant.

Enter THESEUS.

THESEUS is an application I built that achieves the automatic active management I outlined above. The name pays homage to the metaphysical question originally posed by Ancient Greek philosophers. In the context of playlists, specifically tonal ones, I’d venture to say that the playlist remains the same despite its constituent songs being substituted over time. The soul of the playlist is its ability to invoke particular feelings, which is greater than the mere sum of its parts.

Once you register a playlist with THESEUS, it will slowly add songs to the playlist based on Spotify's Recommendations API. Eventually, it will remove old ones. In this way, the playlist will completely change over the course of time, while maintaining its essence. I’d encourage you to read the How To guide on the application page itself, since any documentation here is bound to grow stale.

Building THESEUS gave me the opportunity to use a variety of technologies, some familiar, others novel.

The frontend web client was written using TypeScript, React, and Redux. Naturally, I used the Spotify API to retrieve and update user data.

The backend was written using features from Google Firebase. In particular, I used Firestore to persist the set of playlists a user registers to be managed by THESEUS. I also used the pubsub feature of Functions to periodically run the job to update all user playlists.