Vim's lead maintainer has fully lost his goddamn mind

This is in a PR where Shougo, another long-time contributor, communicates entirely in walls of unparseable AI slop text: https://github.com/vim/vim/pull/19413

What a pathetic state after decades of active, thoughtful work. "I asked the chatbot how to write this code", "Well, I asked my chatbot, and "he" doesn't like it". What a fucking embarrassment.

Add setrepeat() and getrepeat() functions for dot command control by Shougo · Pull Request #19413 · vim/vim

Summary Add setrepeat() and getrepeat() functions to allow scripts to programmatically control the dot (.) repeat command. This enables plugins to: Save and restore the repeat command Make custom ...

GitHub

> Thank you for the detailed feedback! I've addressed all the issues:
> Thank you for the feedback! I agree that following the Vim 8+ naming convention makes sense.
> Thank you for the feedback on naming!
> Thanks for the suggestion! After thinking about this more, I believe repeat_set() / repeat_get() is the right choice:
> Thank you for the feedback. A brief clarification.

This isn't even rewriting his own thoughts in whatever bland style the chatbot can muster, it's just dumping people's comments into the bot and copy-pasting whatever it shits out.

This is The Future that boosters want: An endless bikeshedding session where agreeable chatbots trade excessively polite thank-yous and screenfuls of bullet-pointed lists. A bunch of children, getting paid to play around with tamagotchis.

What a bleak fucking future to dream of.

@AndrewRadev wow this PR is truly painful to read. it feels like these social networks made for AIs to talk to each other. I don’t mind AI writing code for you, but you gotta take ownership over the ideas, you can’t just point a finger and say it this or that, you gotta understand it and own it. You posted it online, under your name, not the AI’s.

@eljojo @AndrewRadev "I don’t mind AI writing code for you"

you really should

@eljojo @dirtside We could give you a bullet pointed list of all the ways generative AI is harmful- but frankly there’s plenty available on the subject and if you’re so ethically bankrupt you’ve ignored it so far, your opinion is not going to change in this thread

@penny @dirtside maybe we can go straight to basics: I'll start by saying I don't believe in copyright. I know it's necessary in today's world, but I'm personally opposed to it. I think it's a true reflection of capitalism and this world is much more worse off because of it than without.

Regarding AI, I believe there's a distinction between the theoretical technology and the current companies that deploy popular models. As well as downstream effects caused by them.

@penny @dirtside to add: I personally try to fight AI slop myself https://github.com/eljojo/no-autopilot
GitHub - eljojo/no-autopilot: You can use AI all you want — just don't submit on autopilot. Gentle PR guardrails that encourage human ownership.

You can use AI all you want — just don't submit on autopilot. Gentle PR guardrails that encourage human ownership. - eljojo/no-autopilot

GitHub
@eljojo @penny @dirtside > Please don’t make AI slop, just make AI gruel, that’s better

@amymagdalena are you open to having an actual conversation about this? or you just want to vent?

as a long time programmer, I legitimately don't understand why out of the sudden everyone started caring about which tools you used to write code

@eljojo All of a sudden? It’s been 3 years of slopfest now after a few years of shitty copilot autocomplete prior. And no, this is not similar to old instances of code automation cause they didn’t do all the logic for you, at most they gave you a bunch of boilerplate to fill in. So why did no one care, cause the tool was a keyboard and a brain and some generator scripts made by people with keyboards and brains, not a suped up Markov chain made without care.

And the reason people aren’t open minded with this with you, because you’re a long term programmer talking about something that has been clearly shitty for years, you should simply know better.

@amymagdalena regardless of "when it began", the bigger point is: why care about the tools I use to write code? shouldn't the code itself be what matters?

saying that people aren't open-minded because I simply should know better is also a bit silly, everyone has a different starting point for everything in life, you can't blame me for not knowing what you do.

I clearly am trying to fight AI slop myself, I published code about it on github, it's not like i've been living under a rock.

@eljojo Why do you claim to write the code when the tool specifically exists to save you the oh so difficult effort of writing it. And yes people do have different starts in life, you already stated your position in life “long term programmer”.
And you didn’t publish code, you published a config file to make the spam bot require a human to approve the spam.

@amymagdalena I claim to write the code because I publish it under my name. I made the git commit. I take responsibility over it. If I hadn't logged into my computer, the code wouldn't exist. What else is necessary for me to claim ownership over it?

I'm a long term programmer, but that doesn't mean I'm supposed to know or experienced with equal degree everything you know. It's not fair to be shitty just because you don't agree with my current opinion.

There's literal javascript in the repo.

@amymagdalena I use AI because it allows me to do more, I don't have to write the physical lines, just review them. This is huge.

This is what I do at work (review code written by others), I'm a Staff Developer

@eljojo What a surprise, it’s a PM that loves AI, shocked I tell you, I am utterly shocked.

@amymagdalena I could say many things about you that don't surprise me, but I'l hold on the judgement.

I'm not a PM, what makes you say that?

I'm a software developer, Production Engineering actually. Here's a talk I gave a few years ago https://www.youtube.com/watch?v=pFESn51YHIQ

Jonathan Gnagy y José Albornoz - Deploying thousands of MySQL DBs using Rails and Kubernetes

YouTube

@eljojo A PM that was given a different title so as he could pretend he hadn’t retired from the profession, soothing his ego. :)

And yeah there’s JavaScript in the file, that we can assume from your statements you didn’t write. And violating the tenet of not putting JavaScript in config files.

@amymagdalena you're literally saying that the title of "Staff Developer" doesn't exist, and it's no different than being a product manager? L O L

As I told you, at work what I do is review other people's code, direct their architecture. The PMs are in charge of product, I don't touch that. My expertise is purely in building scalable code.

@eljojo I obviously meant product manager /s, I meant PM as in Project Manager, which you are, no matter if there’s another project or product manager above you in the chain.
@amymagdalena the code is in the yml so the repo doesn't have to be checked out. this is a very intentional decision! I KNOW WHAT I AM DOING!!
@eljojo Config files shouldn’t contain nor import code, the file structure is irrelevant. But shout some more, I think I can see why people aren’t open minded about this with you.

@amymagdalena also, why deflect the conversation and turn it about me?

You still can't explain why AI guarantees bad code

@eljojo I can’t explain that? I never intended to claim such a guarantee. Did I claim such a guarantee?

@amymagdalena you jumped the thread where I asked why people shouldn't use AI, and started calling it "AI Gruel". You made your opinion on the subject very clear.

Let me ask you: Do you think qualified engineers can use AI to produce production-ready code?

@eljojo @dirtside I don’t either, but here, I will give you a copypaste of my most recent spiel and then politely, I will block you when I believe it’s been received because I don’t really want to know you at this time

it’s all stolen but I don’t even care about copyright, it’s an existential threat to the internet and to society that is funded by billionaires who hate us

It exists only to cause harm and if you get anything out of it you are only enabling and reaping that harm

I also think by the nature of information theory llms can only digest information and lower the entropy floor- anything they output is worse than what they ingested, anything they remove is destroyed

@penny @dirtside if you're not willing to engage in conversation, then why engage in conversation?
@eljojo @penny @dirtside where did Penny lack in engagement? Being pessimistic about your ability to use conversation for any alternative conclusion isn't a poor decision. There's more than enough people applying rigour before despising AI intrusion. You can't just label every aversion to AI as prematurely taken bigotry. It's an aversion of exhaustion for many, and you don't seem to know how to embrace that exhaustion in others.
@goose penny literally threw a copy paste and then blocked me without caring what I had to say about it. LOL -- not too long after attacking me that my opinion was impossible to change L O L

@eljojo @goose hi! Could you share some commits of yours which are heavily vibe coded?

I look for some non trivial code examples as so far whenever I use AI its generating me well only simple snipets. Anything greater than one file quickly piles up to shit and I have to reject it.

The same goes to pickup the architecture or design patterns. Usually AI generates me quite some trash. Fyi; I dont use claude but mostly open models which has waay less parameters. Thanks!

@grayrattus https://github.com/eljojo/estrella was built entirely using Claude Code
GitHub - eljojo/estrella: A print server for the StarPRNT protocol

A print server for the StarPRNT protocol. Contribute to eljojo/estrella development by creating an account on GitHub.

GitHub
GitHub - eljojo/bike-app-astro: global cycling wiki

global cycling wiki. Contribute to eljojo/bike-app-astro development by creating an account on GitHub.

GitHub

@eljojo thanks! Could you ellaborate what was your workflow while buildingit? Have you made it with one readme file as description for AI or had to generate code function by function?

It looks really impressive.

@grayrattus I'd suggest looking at this https://github.com/obra/superpowers it's an entire workflow
GitHub - obra/superpowers: An agentic skills framework & software development methodology that works.

An agentic skills framework & software development methodology that works. - obra/superpowers

GitHub
@AndrewRadev I used to think how much nicer the internet in general and FOSS communities specifically would be if people were a bit warmer and nicer to each other. Acknowledging others' contributions and ideas, thanking each other for good points etc. instead of the popular attempts to outnerd each other on technicalities.

Should've been more careful about what I wish for I guess...
@kunev "There, I fixed it"
misschien iets voor jullie @Noortjevee

@AndrewRadev @kunev I vividly remember having to discuss an essay article from the NYT or Guardian about "Visions of the future" for my English final exams in 2011.

In it the author described scenes where users' soon to be ubiquitous AR glasses, might show them likable cartoon characters from their youth instead of homeless people on the street.

This feels like that

@wall_e @AndrewRadev @kunev

Mark Osborne's MORE short film on a similar theme:

https://www.youtube.com/watch?v=cCeeTfsm8bk

Mark Osborne's MORE

YouTube
@oblomov @wall_e @AndrewRadev @kunev love this film. Thank you for topically posting it here. Think about it a lot.

@kunev @AndrewRadev "the monkey's paw curls"

I've had minimal non-consentual contact with llm-app responses, but those leading sentences of a stochastic parrot reply really really get to me (due to people copypasting slop at me with zero human interaction/filtering/etc)

@AndrewRadev everyuone I talked to when the dumbass AI boom started told me not to worry but I knew straight away vibe coding trash will make things crap. Seeing people I used to look up to when I was a young developer are now AI zombies. My current, thankfully non-dev, work is impacted a lot with certain vendors releasing updates that break a lot of basic functionality. What a f-ing mess :(
@thereverend4253 @AndrewRadev the real tears will start flowing after the crash, when the AI users will have to pay the actual cost of their computations, and start to save money by actually modifying generated code by hand.
There's going to be a lot of hand-wringing and ugly-crying.
And if i don't misjudge humanity, they will probably blame the people who fought AI spread for the crash and their predicament.
@mavu @thereverend4253 @AndrewRadev Get’em hooked then jack the price. Now where have I heard that business plan before?
@qhstone @mavu @AndrewRadev History repeats itself. Hard not to go memberberries about all the fecking tech that was supposed to replace IT people and just made a mess and added more tech debt. Or like Meta, I can still laugh about hard.
@AndrewRadev Sounds like every corporate/business email I’ve ever received. I fucking hate that style of communicating. Maybe it’s just me, but I never say thank you in an email unless I really mean it!

@AndrewRadev Is Shougu a native English speaker?

Some tolerance should perhaps be given to those who are not.

@simon_brooke @AndrewRadev what does it have to do with anything?

@simon_brooke @AndrewRadev absolutely not a valid excuse. There are translation services that do not make you sound like an artificial twat.

Google Translate is fine for just about most things. I would much rather read “obvious Google Translate” slop because you can work backwards from ooorly translated text to find the original meaning.

AI slop strips out the original intent and replaces it with… slop.

@simon_brooke @AndrewRadev i am not a native English speaker either but i am perfectly capable of thinking
@simon_brooke @AndrewRadev
I once saw a non-native English speaker on a forum who's not confident with their English ability that they uses LLM to rewrite their words.
They were suggested to use machine translation (since it's mostly already pretty good now) and add a note that they used machine translation when writing this, instead of letting LLM reword them, which is prone to errors, by another non-native English speaker.

And in the end their English is actually already good enough to communicate.

I myself isn't native English speaker either, and I'd very much more prefer machine translations instead of LLM for a language I don't have confidence with. The thought of LLM probablistically slipping words and sentences I don't intend to into the final paragraphs I sent (which sometimes happens), it terrifies me.
AI Translations Are Adding ‘Hallucinations’ to Wikipedia Articles

AI translated articles swapped sources or added unsourced sentences with no explanation, while others added paragraphs sourced from completely unrelated material.

404 Media
@simon_brooke @AndrewRadev Sad state of affairs if non-English speakers think they will get more tolerance running their thoughts through a stochastic parrot than by just writing their thoughts out directly. Surely they get plenty of tolerance for grammar and spelling already from anyone who is not a total a-hole.
@AndrewRadev what the hell is going on man :(