Post

i think im more careful about typos in my prompting than i am in my slack messages

more and more i wonder why i bother to correct typos in prompts that i know the LLM will just ignore and interpret correctly anyway

Grass Toucher Georg

@avengingfem.me

5 June

interesting. why?

austin (eeek!/ackkk! 👻)

@thebadcode.com

i have a bad theory of mind for LLMs and treat their attention as limited in ways similar to that of a human

5/6/26, 10:31 pm0 reposts8 likes

austin (eeek!/ackkk! 👻)

@thebadcode.com

5 June

also im somewhat partial to the argument that clean, typo-free prose overlaps in latent space with the set of all good engineering practices

Grass Toucher Georg

@avengingfem.me

5 June

i like that argument i'm going to adopt it

Amber 🌸

@shimmermathlabs.com

5 June

LLMs absolutely do respond differently based on where your prompt lands in latent space. lists of explicit rules to follow spelled out in exacting detail push into "sr engineer berating a jr engineer that just fucked up". co-workers trusting each other and working through a problem works better.

gerge

@gerge-lemons.bsky.social

5 June

I might be imagining it but I think sprinkling in some LaTeX when referring to specific variables gives a slightly more precise answer

eva (^_^)/

@eva.computer

5 June

i always start my AGENTS md with something like “we are colleagues and collaborators. we’re pals.”

Singularity's Bounty e/CC

@catblanketflower.yuwakisa.com

7 June

I drop "collaborate" in the context window a few times and the models go nuts. It's like catnip to them

Nathan

@nathanklisch.dev

6 June

The why / framing matters more then the telling. They respond well of you give a reasonable reason, like we are pre-release, with no users, you don't need compatibility, versioning right now