GLOBAL RANDOM — a single-file art piece built on MusicBrainz data
I built a small web piece called GLOBAL RANDOM: it continuously scans random slices of the MusicBrainz database, builds a rotating pool of candidates weighted so every country gets a fair shot (not just the usual US/UK-heavy defaults), and plays them via the embedded Spotify player using the MusicBrainz→Spotify relationships your editors have curated by hand.
No backend, no database of its own — one HTML file, pure client-side JS against the API. No ads, no tracking, no monetization of any kind — purely non-commercial (CC BY-NC-SA). The concept side is about breaking the recommendation-algorithm loop: real randomness instead of taste-modeling, with a deliberate “lottery, not majority” fairness mechanism across the ~247 countries/territories in the data.
Genuinely pleasant to just let run — no Spotify ads showed up for me even without Premium, though I haven’t tested that thoroughly and it might just be uBlock Origin doing its thing on my end.
There’s also a donate button linking straight to GiveDirectly — money goes directly to people in extreme poverty, no detour, not affiliated with this project financially in any way.
Try it here: https://global-random.radio
Longer writeup on the concept (chance in art, the skewed geography of music metadata, etc.): GLOBAL RANDOM — Democracy of Sound
Would love feedback, especially from anyone who knows the API better than I do — happy to share more about the query/pooling logic if useful.