Hacker News Reader

Comment by mplanchard | Hacker News Reader

I think the idea of having a bot that does this is interesting, and worthy of discussion, with interesting arguments on both sides.

But what’s really incredible here is how poorly this bot was coded. I cannot imagine writing something this terribly implemented and obviously poorly tested and setting it loose via automated bot upon the world.

View on HN

I think the situation of developers being harassed by bot-generated pull requests is deplorable.

At the same time, they brought it on themselves for using Github, so serves them right.

Not to mention that automatically correcting gendered words is an AI-complete problem, since you need human context to tell which words are gendered correctly.

And furthermore, human context is necessary but not sufficient:

(nsfw) https://www.youtube.com/watch?v=TT8-5KBInls

This is nothing new. Most spam bots in the past have been poorly implemented

What makes you say it is poorly coded and implemented?

If you look at the PR, it’s changing words like “here” to “theyre” and “help” to “theylp.” It’s basically just a simple find and replace with no concern for whitespace or whether any of the generated content makes up actual words.

The diff is moderately entertaining. I thought it might be an April Fool's.

many -> persony

hero -> theyro

che non sono [Italian] -> che non childo

Did you look at the actual commit ?

https://github.com/EbookFoundation/free-programming-books/pu...

Lol. It's literally `find . -print0 | xargs -0 sed -i -e s/her/them/g -e s/him/them/g ... && git commit -a`. This is a bash one liner. Has to be April Fool's.

Reply on news.ycombinator.com