AI, Vim, and the Illusion of Flow
rosipov.com
rosipov.com
I too have often compared AI usage to vim in the past months: AI enthusiasts seem to spend most of their time tweaking their setup; creating custom agents, skills, mcp servers, writing custom tools for their agent to use, customizing their harnesses, loops and whatnot. Similar to how us vim users keep updating/rewriting our config. It makes you feel productive while you really aren't and while in the end it might help you better stay within the flow the author talks about it the article the gains are typically very minimal if one is honest with oneself.
What (or why) are those frequent config changes?
I use neovim and I can say I've migrated[1] to vim packages (from Pathogen) years ago, and I also git-clone a repo once every few years if I want LSP support for a language I haven't used before.
But I wouldn't call this "keep updating/rewriting".
[1]: By removing a line from my config, and then running `mv` once.
I do think the culture of tweaking constantly without ever doing real work exists, but it's far less common than it seems online. People who make posts about their neovim configs all the time are certainly more likely to be in this camp than people not posting at all.
I do use Neovim as an IDE but thats a bit different, as lazynvim does 90% of what I need and the rest is just tiny changes that improves the experience for me. I also use AI to change my lua files simply because its faster and I'd rather focus on programming.
1. llamacpp with custom reasoning budget and reasoning message tailored to a specific agent.
2. opencode with tailored agent that has access to dynamic context pruning and a graph of the code base.
3. Because local models love to reason instead of working, the llamacpp budget stops the reason with a message that reminds them to use subagents and to prune the context
---
This is successful enough that I can make and expand greenfield projects into full blown tools. I've just built this morning a open document templating system that lets me build reports into reports and make them look nice with templates. What would have bloated up to ~1M context stays within the dynamic context prune's limits set 50k-85k; some of the limits are because of the 128GB GPU, but mostly because processing times go down the larger the context.
So yeah, it's pretty easy to want to keep tweaking this stuff. I'm happy though, I don't have to ever contemplate some cloud service changing my work flow because they need to squeeze $$$ out of my job.
You mean, except for the spurious :w here and there...
Reads like a human, and then the more you read, the AI-fluff appears. Not X, Y blah blah blah.
You could have human written it with your words in half the words. Been more authentic.
ps. I just read the whole article. It is fine and it doesn't look like it was written with sloppy vibes. Honestly, you're complaining just for the sake of it. The amount of whining about AI-slop is getting more exhausting than the problem itself.
Even so, I've read a lot of stuff written by ESLs over the years and far too much stuff written by LLMs over the past 3 years, and I will take ESL every single time.
Regarding the actual content, I'm sorry, but this post is complete slop. From individual sentences to the outline of the post itself; e.g. "The Depth Problem", "The Empathy Gap", and so on. This straight reads like someone gave bullet points to Gemini and said "write me a blogpost".
Maybe you haven't had the misfortune of being completely inundated with LLM generated drivel at your place of work, but this writing has all the hallmarks of it.
It's trite at this point, but the aphorism holds true: "If you can't be bothered to take the effort to write it, why should I bother to read it?"
Perhaps they could, but let's not make generalized assumptions. I don't know about you, but for me English is not a second, nor even a third language, and for people like me it is absolutely beautiful yet a supple and extremely nuanced tool for thought expression. I wouldn't risk unleashing the naked nature of my prose on anyone, it may not be easy on the eye - I liberally use all the tools I have in my text editor - spellchecking, thesauri, dictionaries and translation, etymology lookup and yes, LLMs as well. Would you complain that my texts overuse synonyms and that's just "annoying"?
You're honestly acting like some impetuous child demanding that the world revolve around them and their likes and dislikes. You're asking for a "human connection", while at the same time granting them zero empathy for their effort - however sloppy it was. The choice is yours - nobody's forcing you to read it. Shitting all over someone's (no matter how small) genuine work is disingenuous, not to mention that it is pointless zero sum - nobody gets any kind of satisfaction, none of the involved parties learn anything new.
For me, the article didn't "speak much" at all; I'm not defending its value. For someone it may feel different - I quickly went over it and that was all. I (like many of us) have been very inundated with LLM-generated texts, but frankly, people seem to eagerly label just about anything as AI-slop. It is getting more exhausting than simply dealing with text. Sorry, but we can't uninvent the damn thing back into the hellhole we found it in. Let's try to be nicer to each other in this brand new world.