Earlier quoted context omitted.
That's really fascinating. So if you could find things in pi (or any normal number) quickly enough, then any arbitrary chunk of data could be compressed to just an offset and a length into that number. That's kind of crazy to think about.
It's an idea as old as time. Or 8 years, at the very least: https://github.com/philipl/pifs In expectation, a k-digit sequence will require a k-digit number to encode its offset, so there's no compression there. Then again, there's this gem 762 decimal places in: http://mathworld.wolfram.com/FeynmanPoint.html
That's brilliant. I also like that you don't need to store the offsets anywhere else, because that metadata is also just data that can itself be found in pi.
Actually, that sounds like one of the sects in Library of Babel, who tried to find the one true catalogue (or whatever it was) by looking for the book that told where it was, and the book that told where that book was, etc. There must be an infinite number of books that lead the way to the catalogue.