There are about 4 million songs on Spotify — 20 percent of the app's entire catalogue of music — with zero listens. No one has ever clicked "Play" on them, even once. Forgotify will pick one of them and play it for you, turning you into a very special unique snowflake of musical knowledge.
Although you're going to find a lot of pretty rough tracks this way, there's a certain appeal in clicking play on something that, in Spotify's world, was before neglected. And as Techcrunch notes, it will almost certainly never be played again once Forgotify finds it and sends it your way. A brief tour through a sampling of Forgotify's finds picked up movements of Beethoven, Stravinsky and Purcell; a track off of a best-of-album from Indian pop and jazz singer Usha Uthup; multiple vacation-themed albums that were not by Jimmy Buffett; a bongo drum instruction book; and a sound effects track for a sea lion. Clicking through brings you to the entire album, meaning that Forgotify users can either stay awhile with their new find, or move on to an entirely new thing.
This will not appeal to everyone. But Forgotify shares a resonance with the strange finds available for years on WFMU's Beware of the Blog, (like this year-in-review from Kellogg's, 1971), from Crap From the Past, or the random collection of strange songs on April Winchell's site. Forgotify's collection is a little more limited than some of these finds — Spotify's collection contains music that, after all, was digitized by someone on purpose, presumably because they believed someone would want to hear it.
Hacker News seems to be an early Forgotify adopter, so much so that the site apparently briefly overloaded with demand.. Users there note that among the strange finds are a few less exciting tracks, like the 100 songs per day uploaded by Matt Farley, who floods the zone on the theory that he gets a little bit of money each time someone clicks play on one of them. ("On the Media's" TLDR recently profiled him here).
Forgotify is kind of the anti-Pandora. Instead of trying to guess what you'll like, Forgotify gives you something you probably won't. And that's exactly what makes it interesting.