Lmao. Lmfao, even.
Here, I’m gonna save you some time and summarize the article for you:
- everyone is talking about how good LLMs are [citation missing], but they’re missing the point!
- because really, we’re already so much further ahead! There be magic tools out there!
- wait, you want to know which ones? Uuuuh sorry, the people building then are keeping them a secret…
- …because they are just THAT FRIGHTENINGLY GOOD! [citation missing]
- seriously, I used one of them to build a 30k/month product in 2 hours with zero coding!
- I mean… A subset at least. Oh I also did not do any review, so no idea just how badly fucked it is. Ah, and I guess the price tag comes from the existing reputation and legal guarantees provided by the original tool, which obviously I also couldn’t replicate.
- but worry not! I’m sure I’d be able to build the full tool in DAYS!
- I won’t though. Even though it could TOTALLY make me 30k per month per user. Totally.
- instead, I’ll PROVE to you JUST HOW GOOD these SECRET TOOLS are! Checkout this GitHub repo!
- …ah, no, sorry, I misspoke. I’ll obviously NOT be using the secret, but TOTALLY REAL [citation missing] tools that this entire article is about. For some reason. They’re totally real though bro. Trust me bro. I’m sure one more data-center will fix it bro. Just need to prompt right bro. Like bro, people don’t realize how gooooooood LLMs are bro. I swear bro. My prompts are so good I need to keep them secret because it’s SCARY bro. Bro.
Ahem. Maybe I editorialized a tiny bit. Not much though, trust me bro.
It has been really fun watching any LLM debugging a bug in a moderately large codebase. It has been so bad in my experience (I really wanted the hype to be real) even in really simple and readable programming languages.
It doesn’t stand a chance though for more complex languages. There was an obvious bug in the way one Rust project was loading a config, decided to “test” Claude sonnet 4.5, I watched it identifying the issue but over engineering paragraphs of new Rust utilities without getting there while the solution was in the second paragraph of the github repo of config-rs, just adjusting the prefix and dropping the separator entry.
He sure loves to evangelise “cloning”, a.k.a. “copying other people’s work”. One wonders how anything new or even slightly tailored for each situation will creep into this ecosystem.
To make an analogy, there’s usually nothing terribly wrong with the mass-produced clothes that big box stores sell. But it rarely fits that well, looks too stylish, or does much more than cover your nakedness. Often, that’s acceptable. But sometimes you need special gear or want a really good quality, tailored shirt.
People still make a living as tailors. In fact, a lot of their work these days is making small alterations to the mass-produced stuff for individual clients.
The authors of these articles always make it crystal clear how inexperienced they are.
The role has transformed. Engineers are no longer writing software — they’re designing higher-order systems.
Whoa you don’t say?! We’ve only been doing that for thirty years now. Glad you finally noticed.
You know what the major problem has been for the past couple decades of my programming career? Hint: it wasn’t writing the code.
Leave it to a finance bro to think everything he does is “DiSrUpTiNg.” What year is it?
Where did you get the impression that the author is an inexperienced developer and finance bro? The introduces himself as someone who started programming from the age of eleven.
I’m Michael Arnaldi, Founder and CEO of Effectful Technologies — the company behind Effect, the TypeScript library for building production-grade systems. I’ve been programming most of my life. I started at 11 with the goal of cracking video games. Since then, I’ve written code at every level: from kernel development to the highest abstractions in TypeScript.
I started walking 33 years ago, I am by no mean an expert in walking.
This part
I’ve always been passionate about finance. I used to be a regulated person— an executive director at a firm that created derivative products. Since moving on from that world, I still get the occasional urge to check macroeconomic data and dig into market dynamics.
A few weeks ago, I decided to analyze Polymarket. I wanted to spot insider trading, whale activity, derive volatility — the kind of stuff only a finance nerd would care about.
Maybe just me, but that screamed “finance bro” to me. “Since moving on from that world.”
The inexperience in professional programming is indicated by the entire article. There’s a huge difference between a hobbyist who starts at 11 and a professional developer. Real professional developers don’t start with “I started when I was 11.” I started when I was 15, 20 years before the author did. Who cares?
His LinkedIn shows little experience in professional development. Just a bunch of “CTO” positions.
He dropped out of university in 2018 after eight years.
I’ve also never heard anyone experienced say “I checked the code. It’s good.” Because it’s embarrassing. And he put it in writing. “LGTM 👍” Insightful.
Yeah when I see a slew of Founder and csuite positions I clock as a grifter rather than a competent technologist.
He also seems to refer to himself by the title BDFL (“Benevolent Dictator For Life”) on his own company’s blog.
Which is so cringey it makes my skin crawl.
Every child with a blog is convinced they’re Guido van Rossum like it’s 2010.
Gross
Knowing a programming language is not software engineering.
It’s like building a bird feeder and then claiming you can build a skyscraper.
Maybe I can build a bird feeder that is as tall as a skyscraper. 🤔/s
Trust me bro the finance terminal it wrote had no bugs and was fully featured. No I won’t release the source code, instead I’ll promise to release the source code of my next incredible AI project!
If it was so impressive, just show us. We know AI is great at building tech demos that look like they work, but does it actually work? Who knows!
nah only a finance nerd would want to see it, none of them will be reading this /s
Personally, while I see what the author means, I don’t agree 100%. What I think is bound to happen is that manual development will lose market share, but not fade away completely. Like stone aqueducts still being needed in some cases despite pipes being a thing, some companies may require proof of work, which could be better achieved with humans. Also, not from a commercial angle per se, but for code reviewers, knowing how to write their own codes would help the oddities in machine-generated code, and fix/improve where needed, meaning at least for study they’d need to do it.
