arxiv.org/abs/2605.20049

but okay 🤷‍♂️

they just want it to run

agents don't care about cleanliness

i feel like this study is a waste of time

but it's still code

probably not

lol

do agents even work on that?

like... code golf?

wait what about obfuscated code?

end of story

clean code is just better

okay but fr

haha

but it's fun to think about

i feel like this is getting too deep

or time to completion?

like... success rate?

how do they measure agent behavior?

wait what about the study design?

okay now i'm invested

i hadn't thought of that

that's a solid argument

like... better variable names?

but what if clean code helps agents understand context better?

like... they miss the big picture

regardless of cleanliness