A few days ago I discovered this initiative that I found very interesting and fun: during the month on November you have to develop a program to generate a novel of around 50k words. This is the 7th edition of the event.

I started reading the dev-logs (GitHub issues) of the past editions and discovered some great novels both in term of technology involved and also content.

Some gems:

* MARYSUE (https://github.com/catseye/MARYSUE), which aims to create an interesting novel with a plot and everything else (and the related write-up "A story compiler" (https://git.catseye.tc/MARYSUE/blob/master/doc/Overview%20of...)

* LIFE OF THE AZAR (https://github.com/NaNoGenMo/2017/issues/39), an archive/enciclopedia of some sort of an imaginary city of a couple of thousands of citizens describing the people, their relationships, events that happens in the city and much more

* THE DESERT OF THE WEST (https://github.com/dariusk/NaNoGenMo-2015/issues/156), a generated guide of imaginary worlds together with a map generation alghoritm with rivers and erosion (http://mewo2.com/notes/terrain/) and a generator for city names based on natural language theories (http://mewo2.com/notes/naming-language/)

* MEOW (https://github.com/dariusk/NaNoGenMo-2014/issues/50) meow meow meeeooow mew

Some projects are based on Neural Networks, some on Markov Chains, some on simulations, some on Tracery grammars (https://tracery.io/), some on Prolog, some on "plain" text processing and some on a mix of all of these. (I found an overview of the approaches for the 2016 edition here https://github.com/NaNoGenMo/2016/issues/154)

I find all of this very fascinating and might start my own adventure with text generation in the next days.