For example loops, like as Groundhog Day (film), the book is https://www.puntodeheroe.com/elbucle.html
Or open worlds, like as modern RPG or GTA, the book is https://jose-tamayo.itch.io/la-leyenda-okiri .
Or Elite like gamebooks, the book https://sites.google.com/view/spacegom/inicio .
Plus, there are other new mechanics in other gamebooks.
https://www.arborell.com/windhammer_prize.html
The Lindenbaum Compatition is a newer attempt to do something similar and it has also resulted in some books that I enjoyed reading/playing several of the books from the first year (have not taken the time to look into the entries this year):
https://www.lloydofgamebooks.com/p/voting-is-open-for-202420...
It was useful for making those test gamebooks. I also thought (too much) about how it would be possible to use a LLM to generate gamebooks, but probably best to first randomly generate some kind of structure (directed graph) for the story and then make many smaller prompts to ask for the book to be written one branch at a time. However even if I ever get around to experiment with that I will certainly not release any code (or slop) because AI-generated gamebooks seems like the last thing the world needs.
While the cycle removal algorithm is a fine heuristic, it can perform arbitrarily bad.
