Thanks, great to hear that!

a very important development, congratulations!!

i was curious about this: "Papers in the Anthology should be original work that has not been previously published in any archival format". Does it include pre-prints as well, even if a specific conference allows them?

Not exactly the paper you're looking for, but this study by Jack Grieve basically says that stylistic differences come from variation in register. No "free" style without matter, all structures are weirdly reflected downstream, in a frequency of word "and" etc.
www.degruyterbrill.com/document/doi...

Although, having a short data "passport" for each corpus somewhere where it's visible on the website might be a good idea

Thanks! Since we aggregate a lot of corpora made by other people, it's not governed by a single collection strategy, so each language differs slightly in their principles: versologie.cz/poetree/reso...

See also accompanying paper: brill.com/view/journal...

Interesting paper! I agree that compulsive regularity in LLMs cannot be fully explained by training data; it likely reflects some feedback loops in training and history of the field (NLP crowd had poetry gen as engineering problem for decades). It's all hyperreal - more iambic than iamb.

What’s “human”? 🙈

I really do overuse ChatGPT as an example of generic poet. All my colleagues who invite me somewhere get to read terrible verse about their home cities 💀

There are so many little weirdos in silksong, I love every one of them

Cliff Unger sends greetings