πFS

The GitHub repo is a novelty filesystem that presents a classic gag as software: instead of storing your file, it “stores” offsets into the digits of π and reconstructs the bytes from there. The catch is the whole point. This only works if π’s digits behave like a normal number, which is suspected but not proven, and even then the index of a string in a random-looking sequence is usually enormous. Several commenters grounded the joke in plain information theory. If a file is long enough to be interesting, the first place you should expect to find it in a random sequence is so far out that describing that position costs about as much information as the file. The repo dodges that by splitting files into tiny chunks. In practice it encodes each byte separately, which turns the idea into a guaranteed expansion scheme rather than a compressor. People also connected it to older versions of the same trick like Borges’s Library of Babel, the Sloot coding myth, and “illegal number” arguments. The common landing point was that πFS is useful as a teaching prop. It makes the hidden cost of metadata painfully concrete, and it exposes how often people smuggle in unproven assumptions about π when they talk as if infinite digits automatically contain every possible file in a usable way.

Treat this as a vivid demo of why magical compression claims collapse once you count the metadata. If you build or buy anything that promises huge savings from clever indexing into a shared corpus, ask how many bits the lookup keys actually cost and what assumptions about the source are still unproven.