I’ve found a new online toy, and it’s almost as much of a time-sink as TV Tropes.

Well, OK, this toy is new to me; it actually dates back to 2012–positively ancient in Internet terms. If you already knew about The Infinite Jukebox, I beg your indulgence while I babble for a little while.

Before I get into the details, I should point out that you’ll need to use Safari on a Mac or Chrome on anything else to follow the links. But that shouldn’t be a problem. Everyone has at least five browsers* installed on their computer, right?

* In my case, Firefox, Chrome, Opera, Konqueror, Vivaldi. Windows users, feel free to replace Konqueror with whichever browser Microsoft gave you.

TIJ is founded on a seemingly simple premise: in any piece of music, there will be multiple points where the sound is very similar. Not just the specific musical notes, but the instrumentation, vocal tone, lyrics, and so on.

By mapping those points of equivalence and matching them up, it becomes possible to jump from one to another, extending a piece of music indefinitely.

It can work very well. Naturally, the first piece of music I tried was a BABYMETAL track. It turned out to be an excellent candidate for eternity: there are a large number of matching points, relatively evenly distributed through the piece. I’m slightly embarrassed to admit that I listened for half an hour before I decided it was time to try something else.

Other selections didn’t work as well. Chihiro Onitsuka’s “Gekkou” doesn’t have enough matches to generate enough variety to be infinitely entertaining, and it has a regrettable tendency to get caught in a repeating loop.

Note that there’s no philosophical reason to limit TIJ to popular music, though there are technical reasons. I first tried a favorite Classical piece, Brahms’ “Variations on a Theme by Haydn”*. It took a long time to process–not surprising, since it is more than twenty minutes long–before erroring out.

* The theme isn’t actually by Haydn, but that’s a different story.

So I tried something a bit less ambitious: the overture to Rossini’s “William Tell”. As it turned out, I wasn’t the first person to try that selection. There were already at least three different versions in the database. It worked very well, once I got past the mental hiccup of “that’s not the way the music goes”*. It was much harder to do that with Rossini than with any popular piece I tried. Given the amount of time I’ve spent listing to BABYMETAL over the past few years, it can’t have anything to do with familiarity. Maybe it has something to do with the fact that I’ve played the William Tell Overture, not just listened to it. I invite speculation from any psychologists, amateur or professional.

* Those of you familiar with Rossini’s approach to composition are already asking “How can you tell it was being played by TIJ?”. You may be amused to hear that the playback got stuck in a complex series of jumps shortly before the end of the piece, and it was a couple of minutes before I realized it was going around and around the same territory.

Quite a few people have tried the “Mars” movement from Holst’s “The Planets”. I didn’t think it held up well. The connections were plentiful and well-balanced, but so much of the piece’s impact depends on the way it builds to the climax, that jumping ahead or back is hugely jarring. Portions of the playback did sound oddly like the Main Title from the original “Star Wars,” however; a connection I’d never made before.

Huh. There are a hell of a lot of people who tried the “Cantina Band” track. Now there’s a piece that lends itself well to eternal fragmentation and recombination, since it’s already heavy on repetition.

Ragtime works reasonably well. How about an infinite “Maple Leaf Rag“?

I could go on and on in a sort of textual infinite jukebox, but I’ll spare you. To send you off appropriately, here’s one last link illustrating the perils of dropping down this particular rabbit hole.