Extremely cool idea! Are you planning to share the source or prompts at all?
Do you know what are the scaling limits of this kind of graph?
EDIT: In the past people used to pore over each line of series like asoiaf to hand compile their own wiki of this kind of data and GRRM was known to consult them to ensure consistency in his universe. So you never know, a tool like this may finally help him get over the finishing line for resolving his nebulous knot of plot lines! (tongue in cheek...)
Thanks! On the source: I honestly never planned this as an open-source project — the prompts plus the programmatic validation between stages are most of the product. But I'm planning a technical post soon that goes deeper into the architectural details I've been converging on over the past six months.
On scaling: context almost completely stops growing after chapter ~70 in the case of the Oz series. By my estimates it's chapter ~90 at the latest depending on the story — for practically any story, context growth flattens out around there. The theoretical limit of the system should be somewhere around 80 million words.
I like the idea of using something like this as a writing aid, as if to "compile" a story to shake out what contradictions or open plot lines or implications are yet to resolve.
That's right! The system quickly updates the plot chart as new chapters are released, and the author doesn't have to create and maintain the wiki themselves.
Oz is the perfect series to show this off and I'm delighted to see others still going deep into the lore, which is so rich and expansive. Once you consider how he also weaved in other parts of his bibliography into Oz, it's a wonderfully huge universe. In my opinion, Hollywood has never come close to capturing the true spirit of the books.
Absolutely — The Wandering Inn was on my mind from the very start when I was designing this. At 14M+ words it's pretty much the ultimate stress test of the whole approach. But I won't run it without the author's consent — I'll be reaching out to them soon.
Primarily it's a way to build a model of a story's world, and two things fall out of that.
1. A wiki for the story. For the author and for readers, so you don't forget what happened 300 chapters ago, which promises are still active, and so you can explore a story you love structurally.
2. Translation. The system scales to millions of words, so when it translates chapter 100 it has the full, precise context of what happened in chapter 1, including speech registers, relationships between entities, and a synopsis of the world.
I read very long web serials myself, and I kept losing the thread of stories I loved. So primarily it's a way to build a model of a story's world, and two things fall out of that.
Extremely cool idea! Are you planning to share the source or prompts at all?
Do you know what are the scaling limits of this kind of graph?
EDIT: In the past people used to pore over each line of series like asoiaf to hand compile their own wiki of this kind of data and GRRM was known to consult them to ensure consistency in his universe. So you never know, a tool like this may finally help him get over the finishing line for resolving his nebulous knot of plot lines! (tongue in cheek...)
Thanks! On the source: I honestly never planned this as an open-source project — the prompts plus the programmatic validation between stages are most of the product. But I'm planning a technical post soon that goes deeper into the architectural details I've been converging on over the past six months.
On scaling: context almost completely stops growing after chapter ~70 in the case of the Oz series. By my estimates it's chapter ~90 at the latest depending on the story — for practically any story, context growth flattens out around there. The theoretical limit of the system should be somewhere around 80 million words.
I like the idea of using something like this as a writing aid, as if to "compile" a story to shake out what contradictions or open plot lines or implications are yet to resolve.
That's right! The system quickly updates the plot chart as new chapters are released, and the author doesn't have to create and maintain the wiki themselves.
Oz is the perfect series to show this off and I'm delighted to see others still going deep into the lore, which is so rich and expansive. Once you consider how he also weaved in other parts of his bibliography into Oz, it's a wonderfully huge universe. In my opinion, Hollywood has never come close to capturing the true spirit of the books.
can you try running it on [The Wandering Inn](https://wanderinginn.com/)?
Absolutely — The Wandering Inn was on my mind from the very start when I was designing this. At 14M+ words it's pretty much the ultimate stress test of the whole approach. But I won't run it without the author's consent — I'll be reaching out to them soon.
For what purpose did you make this?
Primarily it's a way to build a model of a story's world, and two things fall out of that.
1. A wiki for the story. For the author and for readers, so you don't forget what happened 300 chapters ago, which promises are still active, and so you can explore a story you love structurally.
2. Translation. The system scales to millions of words, so when it translates chapter 100 it has the full, precise context of what happened in chapter 1, including speech registers, relationships between entities, and a synopsis of the world.
I read very long web serials myself, and I kept losing the thread of stories I loved. So primarily it's a way to build a model of a story's world, and two things fall out of that.