Reddit users battle ChatGPT's compulsive linebreak-heavy storytelling style that ignores instructions
Dogbold · reddit · 2026-09-11
A Reddit user complains that ChatGPT constantly breaks stories and roleplay into one-line paragraphs for dramatic emphasis — punchy fragments like "He strikes with heavy force. The force of a mountain." — and even after being told to stop, it reverts after a few turns.
The post asks why the model does this and how to disable it. This fragmented, dramatic prose style is a known RLHF-trained quirk; community fixes include explicit system-prompt bans on short dramatic paragraphs, few-shot counterexamples, and periodically reasserting formatting constraints in long sessions.
More from Fun
- One-prompt 3D muscle anatomy site built on GPT-6 via WorkBuddy — lxfater · 2026-09-11
- Satirical startup claims Navier-Stokes singularity will cool GPUs, raising $1B seed — AIFlow_ML · 2026-09-11
- Nick Walton jokes about training a fly to solve a Rubik's cube — Al_Grigor · 2026-09-11
- After Planning Two Books with ChatGPT, He No Longer Feels the Need to Write Them — tinyfool · 2026-09-11
- After 9 Years, Bots Finally Show Up on a Blogger's Bounty Page with AI Text — gleech · 2026-09-11
- Astra builds a surprisingly polished Catan game in three.js, reusing past UI and 3D assets — FinanceYF5 · 2026-09-11