Cast by aviationdoctor.eth
Show the cast this replies toYour comment made my day, thank you! As to why Borges' own works, and The Library of Babel in particular, is not featured in the Catalogue: it's a bit of a rabbit hole! There is no shorter way to describe a book's address in the Library than for that address to contain the entire corpus of the book. That's because any abbreviated address would lead to an equally abbreviated book, which also exists somewhere in the Library. So to be able to locate Borges' complete book from within the Catalogue, I must store its complete contents in my GitHub repo. Now you might see where this is going. Even though Borges' works are no longer under copyright, the prevailing English translation of his "La biblioteca de Babel" still is. I could legally host the original version, but so far I've only added books in English, which I'm not allowed to do in this case or I'd be breaking GitHub's TOS. Now, in my code, I store the ~1K classics that you can find in the Catalogue as Base64-encoded and then zlib-compressed strings, because it reduces the length of each book (and thus the book's URL) by about two thirds. That led me to explore whether I could get away with storing copyrighted materials as long as they were deterministically obfuscated form. After all, all copyrighted materials already exist somewhere in the Library as a result of "infinite monkeys on a keyboard", so all I'd be doing is pointing out those materials among a sea of nonsense. It's an interesting philosophical-legal question. But under U.S. law, a “copy” expressly includes a form from which a work can be recovered “with the aid of a machine or device”, so that wouldn't work. What I plan on doing in the future is expand the Catalogue to contain a few thousand more of Gutenberg's copyright-free books, or even link directly to Gutenberg's index, so that all their books are readable from within the Library, including those in their original version (though that would no longer work in the offline version of the Library, which right now you can download from GitHub and run locally on your desktop completely offline).
2 replies
At any rate, just sharing this between you and me...
I might run afoul of the new FC purity pledge 😅 Few more thoughts: 1/ If you get the error message "The library couldn't be built" upon following the link, as I did, just refresh the page and it will work. 2/ We could get around translation copyrights by machine-translating Spanish texts to English ourselves — AI has gotten good enough for that I think. 3/ I have an idea for linking vastly many more books inside the library without referencing their contents; and that is recording their hexagon and shelf location only after finding them. I'll investigate that further.
I'm really excited about this, im a literature freak and even if am not a great ebook reader, I think this is better than Wikipedia. About the Borges problem you mentioned: La biblioteca de Babel in the original Spanish is something you said you could legally host If possible I would love to contribute on the Spanish side, If it helps, I think I can prepare the texts as plain .txt files or your predefine format. Just thinking out loud if "ñ" or "á,é,í..ü" might be a problem (?).
I found the original in the Library: tinyurl.com/hdsbsv2y The accented characters are automatically replaced by their non-accented (and uppercase) equivalent when searching for a text in the Library. I *could* change the Library settings to accept more than 64 different symbols, to let's say 128–1,024, so that we could include accented letters (keep in mind that even ASCII at 1,024 doesn't include accents, so we'd need to come up with a bespoke set). But then the size of the library would balloon — for example from 64^(2^20) right now (where 2^20 is the number of characters in one book) to e.g., 1,024^(2^20), which in an increase from ~10^1,893,917 to ~10^3,156,528 possible books 🫠 Alternatively we stick with the current 64 symbols and the tradeoff is capital letters only, no accents. I'll need to think deeper about this. Eventually I'd like to have some version of the TinyURL scheme where I can store shortcuts to books — wouldn't be a bijection, only a limited number of shorter URLs can represent a limited number of books given the math of addressing a book. This will make linking much easier, and we could have a whole database in the Library where you can properly add or look up any book.
The 64 symbols excluding accentuated capital letters seems fair enought. Congrats on the project, again, I really love it.
