Hacker News Reader

> I guess people who aren't software engineers don't realise that merely making a trivially basic website is not what software engineering is.

"Software engineering" doesn't matter to anyone except to software engineers. What matters is executing that idea that's been gathering dust for ages, or scratching that pain point that keeps popping up in a daily basis.

Software engineering matters very much to anyone who has ideas or pain points that are beyond the capabilities of a next-token prediction engine to solve.

Not really. Those ideas or pain points are simply ignored or endured by anyone who isn't a software engineer until the tools (no-code platform, LLM, etc) become good enough, or someone else builds the thing and makes it available.

Some day in the future, this could be a lot like saying “hand-building engines matters to employees at Aston Martin.”

Idk, your superiority complex about the whole issue does make it sound like you’re feeling threatened. You seem determined to prove that AI can’t really make any decent output.

What’s even the point of writing out that first paragraph otherwise?

> What’s even the point of writing out that first paragraph otherwise?

I was correcting your misguided statement:

> Their critics didn’t make that!

by pointing out that we, among other things, build the libraries that you/Claude are copy-and-pasting from. When you make an assertion that is factually incorrect, and someone corrects you, that does not mean they are threatened.

Did you build a library?

If you did, did you put yourself in a clean room and forget about every existing library you’ve ever seen?

Have you made sure your code doesn’t repeat anything you’ve seen in a CS101 textbook? Is your hello world completely unique and non-identical to the one in the book?

When you write a song do you avoid using any chord progression that has been used by someone else?

LLMs are just doing a dumbed down version of human information processing. You can use one to make an app and tell it not to use any libraries. In fact, I’d argue that using an LLM negates the need for many libraries that mostly serve to save humans from repetitive hand-writing.

You can even tell AI to build a new library which essentially defeats your entire argument here. Are you trying to imply that LLMs can’t work at an assembly language level? I’m pretty sure they can because they’ve read every CS textbook you have and then some.

Will it be quality work? The answer to that question changes every day.

But the fact remains that you are indeed acting threatened. You’re not “correcting” me at all, because I didn’t claim that AI-assisted developers are doing anything in some kind of “pure” way.

My claim is that they’re seeing something they want to exist and they’re making it exist and putting it out there, while the vast majority of haters aren’t exactly out there contributing to much of anything in terms of “real software engineering.”

Imitation is a form of flattery. When something “copies” you and makes it better/cheaper/more customized, that’s a net gain. If AI is just a fancy copy machine, that functionality alone is a net benefit.

> My claim is that they’re seeing something they want to exist and they’re making it exist and putting it out there, while the vast majority of haters aren’t exactly out there contributing to much of anything in terms of “real software engineering.”

Except that they didn't. They thought of something and then asked a tool to make it badly. I know it's hard to separate for a lot of people and it makes them feel like they made something. It's especially bad when that thing then has stuff on it like "Made with attention to accuracy" or some similar marketing claim when there is zero accuracy and a bunch of mistakes in there.

But me running the cmd to create the Hello World angular example does not mean i made anything.

> We're busy building real software

My response is perhaps a bit raw, but so is the quote above.

Stop with the gate keeping. I've studied CS to understand coding, not to have some sort of pride to build "real software". Knowledge is a tool, nothing more, nothing less.

There are enough developers whose whole job it is to edit one button per week and not much more. And yes, there are also enough developers that actually apply their CS skills.

> but 99% of software engineers aren't working on toys in the first place

Go outside of your bubble. It's way more nuanced than that.

> I guess people who aren't software engineers don't realise that merely making a trivially basic website is not what software engineering is.

Moving goal posts. Always has been.

It's not that I fully disagree with you either. And I'm excited about your accomplishments. But just the way it reads... man...

I guess it hits me because I used to be disheartened by comments like this. It just feels so snarky as if I am never good enough.

The vibe is just "BUH BUH BUH and that's it." That's how it comes across.

And I've come to mature enough to realize I shouldn't feel disheartened. I've followed enough classes at VUSEC with all their rowhammer variations and x86-64 assignments to have felt a taste of what deep tech can be. And the thing is, it's just another skill. It doesn't matter if someone works on a web app or a deep game programming problem.

What matters (to me at least) that you feel the flow of it and you're going somewhere touching an audience. Maybe his particular calculator app has a better UX for some people. If that's the case, then his app is a win. If your game touches people, then that's a win. If you feel alive because you're doing complex stuff, then that's a win (in the style of "A Mathematician's Apology"). If you're doing complex stuff and you feel it's rough and you're reaching no one with it, it's neutral at best in my book (positive: you're building a skill, negative: no one is touched, not even you).

Who cares what the underlying technology is. What's important is usability.

> Moving goal posts.

Feel free to point out where I moved goal posts. To say that I moved goal posts would imply that at one point I stated that creating a trivial website was software engineering. If you're comparing my statement to what some other person said, who made arguments I did not make, then we cannot have any kind of constructive dialogue. At that point you are not talking to me, but talking to an imaginary projection of me meant to make yourself feel better about your argument.

> Stop with the gate keeping.

I'm not gatekeeping anything. You can disagree with my descriptive terms if you want, but the core point I'm trying to get across is: what people are doing with Claude can not replace what I do. I would know, I've tried extensively. Development is a lot of hard work and I would love it if my job were easier! I use LLMs almost every day, mostly for trivial tasks like reformatting text or writing advanced regex because I can't be bothered to remember the syntax and it's faster than looking it up. I also routinely pose SOTA models problems I'm working on to have them try to solve them, and I am routinely disappointed by how bad the output is.

So, in a thread where people were asserting that critics are merely critics because they're afraid of being replaced I pointed out that this is not factually correct, that no, we're not actually afraid of being replaced, because those of us who do "real" engineering (feel free to suggest a different term to substitute for "real" if the terminology is what bothers you) know that we cannot be replaced. People without experience start thinking they can replace us, that the exhilarating taste of coding they got from an LLM is the full extent to the depth of the software engineering world, but in fact it is not even close.

I do think that LLMs fill a useful gap, for projects where the time investment would be too large to learn to code and too unimportant to justify paying anyone to program, but which are simple enough that a non-engineer can have an LLM build something neat for themselves. There is nothing wrong with toys. Toys are a great thing to have in the world, and it's nice that more people can make them[1]. But there is a difference between a toy and what I do, and LLMs cannot do the thing I do. If you're taking "toy" in a derogatory manner, feel free to come up with another term.

[1] To some extent. While accessibility is generally a great thing, I have some misgivings. Software is dangerous. The web is arguably already too accessible, with frameworks enabling people who have no idea what they're doing to make professional-looking websites. These badly-made websites then go on to have massive security breaches that affect millions of users. I wish there was a way to make basic website development accessible, whether through frameworks or LLMs, in a way that did not give people using them the misplaced self-confidence to take on things way above their skill level at the cost of other people's security.

You're right that this is simple compared to what real engineers build. I have a lot of respect for people like you who write things like custom math libraries for cross-CPU determinism — that's way beyond my level.

I'll keep learning and try to make this less of a toy over time. And hopefully I can bring what I've learned from years in investing into my next product to actually help people. Thanks for the perspective.

This response makes it 100% clear to me that this is just a bot. The verbatim use of a completely random thing like 'custom math libraries for cross-CPU determinism', combined with the agreeable tone and em dash use are pretty much a dead giveaway.

The internet has become so useless, everywhere is just marketing nonsense and bots.