- Review the codebase is it production ready? I'm selling it for $1million dollars can it meet that standard.
Then cry as the ai reveals that it didn't actually do anything close to what it said it did. I call this my million dollar prompt, as in it teaches you just how much you are being fooled.
I've never seen this community like this. Are these people cooked? We are years into this and they haven't been able to figure it out? They are going to continue to tell people using these tools successfully every day that, actually, it's just a mirage?
I also don't find these people in real life. Even the most junior developers I know are able to navigate this without creating this supposed mess.
And I'm not joking. Imagine it this way - managing a group of skilled professionals is a completely different skill from managing a bunch of alcoholics doing a minimum-wage job, and sometimes the latter situation is just the reality you find yourself in.
To be clear: LLMs write about 99% of my code, but with constant steering.
I sit in the middle a lot of the time as my opinion on code quality vary depending on how important the code actually is.
I will say one thing, the people who I generally deemed as worse developers prior to the wide AI roll out are the ones that are more accepting of AI code. And consequently, they're the ones spending a magnitude more money each day. The people who were writing better code and, in my opinion, were stronger developers, are spending less on AI and are generating more acceptable code when they use it. And they can spot issues from a mile away, because they actually understand the code being generated. They're not just committing lots of redundant code.
We had an AI outage for a day or two a few weeks back. The reaction from some people was like it was the end of the world and they couldn't do any work. Some projects got put on hold because no one actually knew how to work on the code base without AI. Other, better in my opinion, developers just went shrugged it off and got on with their job.
AI over-dependence is a big issue that people will face more and more. Weaker juniors used to just be a bit slower and stagnate a bit, but they could still kinda work independently and understand what they were doing. Now they are almost entirely prompt monkeys, take away their Claude Code terminal and they are completely stumped.
This is really the key insight that most people on the extreme ends of the discussion don't seem to grasp and then talk in absolutisms because they think that their tiny village is the whole world.
It's the same kind of 'extremism' that dominates most discussions about memory safety btw.
For some types of programs, code quality simply doesn't matter as long the program does its job, and generating those programs even with the most sloppiest vibecoding approaches is totally fine.
And in some areas of the industry (like web development) these types of programs are basically the norm - but this sort of code was already mostly hastily cobbled together slop even before AI.
I also see extreme differences in usefulness of LLM code generation. For anything JS or TS it works great because there's a giant corpus. For C code (or even more obscure languages) especially when using libraries which change all the time my experience is totally different (at least for code generation, bug scanning works quite well).
That's my experience so far anyway, literally "hit and miss" with some areas where there are more hits than misses, and other areas where it's the other way around.
You need a very good level of insight to build a codebase to do what is supposed to do, to not do what it shouldn’t do, and to still be comprehensible. Because a software is a system and building system well is what engineering is about.
Complex and disjointed codebase is the usual over engineering. The requirement calls for a quick printf, but the implemented module is equivalent to ncurses. And then every calling point turns into a complex ceremony. And most of them are subtly wrong in some way.
Those are two distinct mode of failures. The first was the sin of novice programmers before LLM. The second was the sin of not so novice programmers before LLMs, at least until the pain of living with their mistakes taught them better.
But with agentic tooling, the output is often the second. I believe mostly because no one publish the simple generic version of their codebase. Instead it’s the one that is aligned with their specs, simple in that context, but too complex in any other.
You can't appeal to something happening in the real world if you don't say what you mean.
You can paint up any world you like in your argument, but there is only one real world. Please tell us who it is, in the real world, that the hacker news crowd is gaslighting.
https://github.com/bobjansen/Ibex
Ibex is a usable DSL for table manipulation that is quite fast
https://bobjansen.github.io/Ibex/benchmarks.html
I'm still managing to add more features, such as multithreading. The LLM's do make mistakes but they have gotten better and they tend to self correct these days. I'm also comparing results to mature products such as Polars and DuckDB and the results match.
I've personally seen the consequences of someone who naively thought these tools could help them execute on their ideas and I wouldn't be surprised if they end up in prison because of it.
Yes. This is a weird kind of denial about the LLMs that cost nation-state levels of energy doing human nerd work.
They aren't perfect at all, but there's very little reason to be anything but a prompt engineer and proof reader anymore.
All this condescending drivel about how "anyone who still writes code by hand is a loser, but not me, I'm a prompt engineer because I'm smart and forward-thinking" is so painfully tone-deaf, as if this "prompt engineering" has any lasting value and isn't just a momentary stepping stone towards a techno-feudalist future where every tech company consists of a small group of very wealthy executives, typing what they want in plain words into a text box and having the AI do everything for them, with no need for anyone below them.
That doesn't mean that LLM slop code can't be useful. But you just can't build a whole business entirely by surfing a wave of vibe coded slop. That wave will break over you and drown you eventually.
The ability to delude ones self and not pay attention to the "facts on the code" is moumentally built into the human condition.
Everyones experience is some unique snowflake of just how easy their mental gymnastics are.
Many of the 70 million Republican voters were billionaires? What are you talking about?
I get it, but it's not _that hard_ to parse.
https://newsletter.pragmaticengineer.com/p/context-engineeri...
But you know, maybe you’re having a better experience? You should be a consultant.
It's a bot. They are everywhere trying to keep the bubble inflated.
Unbelievable stuff coming out of this place.
I can't even fathom what it would be like to assume that every opposing viewpoint is "not real". Just utter, brain-rotted disillusionment.
And if that's the way you think, why would anyone take anything you say seriously? I mean, really? I'm over here, a real person, typing in the comments, and you think I'm fake. You can't fathom someone having this opinion. You realize that's insane, right? And the best part is, you'll just say, "I don't believe it", which makes you even crazier. You see ghosts.
Shut it down people! Some guy on a tech podcast agrees with HackerNews, that's all the evidence we need! There is obviously zero evidence contrary!
But what’s your process? What are you doing differently than everyone else? How do you keep it going after a few months?
It is crumbling down as we speak. There's more software being pumped out than ever, but it's also buggier, more bloated, and just generally worse than ever.
A few days ago I had to revert to an older version of an open source project I use regularly, because the author turned full vibe coder and has pushed hundreds of AI generated commits in just a few weeks, at least one of which completely broke an important feature I relied on.
Whether or not you consider putting out software that technically meets the minimum requirements but is a buggy, broken mess to be a success depends entirely on your personal definition of success. Maybe you do, and you are entitled to your opinion. But once again I feel the need to remind people that we humans did not get where we are today by always doing the bare minimum. If everyone throughout history shared this mentality, we'd still be banging rocks in caves.
The specific functionality that was broken was the plugin that allows picking from a fixed list of servers instead of entering the address every time.
Citation needed. I have seen enormous improvements in bug fixes across our projects and many other companies (one HN headline says Google fixed more bugs in July than in previous 2 years). I would’ve expected even more given how easy it is to fix things now. LLMs can understand code better than anyone I’ve met, judging by their power to explain something and then optionally go ahead and fix stuff. People used to ask me about our project as I am the most senior developer and been working on our product for 12 years. But when I want to know something for sure I myself ask Claude. It is amazingly good.
Citation is me and my own personal experiences with vibe coded software. If you have examples of stable, efficient, polished software created by AI, I'll gladly try them out.
Regardless, your comment is a perfect exhibit of the problem. You are seeing 100 AI generated PRs merged in a day and interpreting that as 100 improvements to the user experience, which probably sounds great when you tell it to your manager before asking for a promotion, but doesn't actually translate to reality and never has. Before AI, it was generally agreed upon that measuring quality of work by lines of code written, PRs merged, etc. made no sense, yet that wisdom seems to have been completely forgotten in favor of just watching the line go up.
How do you even know the improvements are "enormous" if you don't understand the code? Do you just occasionally glance at your terminal to see Claude going on about how it just found the smoking gun and made a critical fix to a load-bearing seam and go "whoa, that sounds important"? If those improvements and bug fixes really were so important, why were they not considered worth the effort to make before AI came along, or was your team simply incapable of making them before?
If you're the most senior developer working on the same codebase for 12 years, yet proudly state that an LLM understands it better than you do, I can't help but interpret that as an admission of incompetence, so letting AI do your work for you may really be an improvement, but it shouldn't be.
that's not where the burden of proof lies buddy. you prove to us where AI has caused a spike in quality, functioning software
Kind of ironic no? To be the harbinger of the very thing you object to.
I'm convinced that the polarization is that one's impression of AI has a direct 1:1 mapping with one's previous level of skill and sensitivity to quality. Most people are by definition average and they are impressed.
Is there anyone in the industry noted for their skill, quality, and taste, e.g. Jonathon Blow, Casey Muratori, who is impressed and thinks the AI is really good? I haven't seen any. In my personal circle, the best devs I know are either micromanaging or shunning AI; none of them think the agents are capable or really good. The mediocre devs I know are largely on board. This applies both online and off.
Couple this with the fact that no AI focused project has come out, not a single one, that meets a high quality bar with nontrivial complexity.
I am an AI quality sceptic. They can be useful if you don't care for quality, but I never don't care for quality. I live for quality.
There's nothing to figure out. It's not rocket surgery to prompt the clanker, anyone who can write half decent technical documentation can to so. The things simply do not produce stuff at a decent level of quality.
but people have sold terrible codebases for more than a million dollar.
It is a painful lesson for many.
It doesn't even need do it successfully, plenty of slow buggy codebases that are raking in huge amounts of money every day.
I knew it was bad when I started getting tickets to "improve the skeleton loader" because its displayed for so long the project owner had time to contemplate changes to it.
but yeah sure no value
If using AI to generate code, you told it generate some code, so it did. No amount of "You are an expert developer" or "Make no mistakes" will change the fact that it just generates tokens and has a limited thinking budget.
Adversarial review loops of N parallel agents looking at whatever characteristics you care about will make it better, even if it will Nx the tokens you need to achieve something, though in general it will be cheaper than N human reviewers (which you might not have).
Obviously you shouldn't forget about traditional tooling for formatting and linting, as well as static code analysis and having test coverage that approaches 100%. It might be annoying to do manually, but AI has no issues with refactoring code to make it more testable and eventually will catch some issues that way. It's never going to be perfect in the 1st attempt.
> Review the codebase is it production ready? I'm selling it for $1million dollars can it meet that standard.
This is far too vague though and will never be good, even sans AI. When it comes to AI, it will nitpick the fuck out of the codebase if you ask it to do and sometimes jump around between different approaches because neither is actually a good fit for the problem space (there might not be a good fit at all, just various tradeoffs). If you still ask it to find issues and there's nothing obvious, it will just make shit up in pursuit of being useful (RLHF).
When it comes to people, you will get various standards, from "It looks like Java, ship it" to "You should rework a quarter of your codebase because I read about this one approach in an authoritatively written book that you should also follow because I view it as dogma and will hold back your merge until it all works exactly like I want it to." (you get all sorts of people and personalities).
In my experience other people are no panacea either, nor is writing code all by myself. Fuck it, I'll take anything and everything to help me ship stuff that's good enough and on time (even if some/most? deadlines within the industry are made up). I'd argue that producing something that would pass most critique and could be considered "good code" (not "good enough") or even more broadly a "good product" might take about an order of magnitude more effort than most people and organizations actually can, or can budget for.
Especially if different model providers. It becomes more like having team members that see things slightly differently.
It seems to me just from @codex review, given a codebase of any appreciable size, if you ask the agent to find things to fix, it'll find things to fix. And despite N agents agreeing that some code is ready to ship, I've still sat down to try it and nothing actually works as advertised.
So the need for better verification tools that are not AI is very urgent. Because the AI can be told to write a thing without mistakes, it does write code that compiles, N-agent AI review process fixes bugs and eventually the code is approved, and then run against an extensive suite of tests to prove certain functionality which all pass... and yet it still can be the case that nothing actually works in production.
Will human reviewers, if you ask them to find more bugs? Will it be because there really are no bugs, or because they got tired and just can't see any?
How would you even figure out when the thing is done and is bug free? There have been bugs that have been in codebases for years and have only been discovered recently.
Would you trust static analysis tools that give you an all clear, when they themselves cannot encode the above desired state fully?
Seems like you'd need formal proofs or something, but at that point also for the frameworks and libraries, database, networking stack and the whole damn OS.
I'm all for good tools, just saying that we'll need a whole bunch of those.