Ask HN: Why is the HN crowd so anti-AI?

It's simply divided. With every such division A vs. B, the A team thinks HN is anti-A and the B team thinks it's anti-B. This is an invariant.

You can see from this megathread, currently on the front page, that HN is by no means anti-AI:

Ask HN: What was your "oh shit" moment with GenAI? - https://news.ycombinator.com/item?id=48406174.

Sometimes it just takes the right initial condition (e.g. title) to bring out one side or other.

As for why the community is divided, there's always a temptation to come up with HN-specific explanations, but society as a whole is divided about AI. Surely that is the only explanation one needs. As I've been saying for years, HN can't be immune from macro trends: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

The HN crowd is much less anti-AI than the rest of the country. But that's a low bar

A Quinnipiac poll showed 80% of Americans are "very" or "somewhat" concerned about artificial intelligence, with only 35% feeling excited about it

https://poll.qu.edu/poll-release?releaseid=3955

Gallup found that 71% of Americans oppose the construction of AI data centers in their local communities.

https://news.gallup.com/poll/709772/americans-oppose-data-ce...

A Fox News Poll indicated that 80% of voters believe protecting the public's interests and enacting regulations should be prioritized over unchecked technological innovation.

https://www.foxnews.com/politics/fox-news-poll-voters-see-ai...

And Pew Research found that the majority of Americans are "more concerned than excited" about AI, with that number increasing over the years

https://www.pewresearch.org/short-reads/2026/03/12/key-findi...

I call these AI tools "proprietary non-determenistic database of the free internet". They belong to american companies which can cut off your access if american government doesn't like your country's government. They fed from the free internet that many of us grew up in, store it in humans unreadable form and sell you access to it. If some day claude starts to spit out compiled binaries instead of code nobody will notice, and we'll essentially get proprietary cloud-hosted compiler that most in the world depends on to build software. With built-in telemetry and backdoors and clause in license that allow full overtake of your business if provider wants it ofc. It's a great shift from the internet we all know and love towards the new subscription-based access to world's propriatary knowledge base. It's a perfect "mind control" tool as well - you don't need USAID, "free media" and stuff like that in other countries when all people there including politicians ask chatgpt everything from meaning of life to recipies of pancakes. Once you see these political and philosophical dimensions it's hard to unsee how claudecode running on my PC won't turn into a weapon some day. But in blissful ignorance it's fun to use, and companies love it for the promise of replacing people. Amen

> Over the past six months, there hasn’t been a single day where I’ve checked the HN Best RSS feed without seeing a post about how AI “writes bad code,” “introduces bugs,” “creates technical debt,” or something along those lines.

because it's true

> Users don’t care whether the code was written by AI or by hand, or which framework you used. They care that the product works.

How can you guarantee that it works though? You can verify, but it would be at the same speed as before the AI, or even slower.

> By then, enough real-world feedback would have surfaced to identify the major issues, and tools like Claude Code would make it possible to fix and ship version 2.0 at an incredible pace.

By then you have a blackbox of a codebase which is unmaintainable, or in a worst case scenario you end up losing your data or get hacked or both.

There are two different crowds using "AI":

- One crowd is using to research algorithms, libraries, write boilerplate code, write test harnesses, introspect and integrate with APIs, do hands-off refactoring, and automating what would otherwise be boring tasks. They still think about architecture, best practices, understanding things in detail and the general shape of the solution is in their hands.

- Another crowd is curating prompts, setting up autonomous agents, creating tooling and guardrails around it, anything else but getting actually involved in how the sausage is made. They are working on meta tasks around the problem, in the hope the solution will write itself.

These two crowds are currently living in very different worlds, and getting very different results. We'll see what survives soon.

The more close you are to the fire the more you understand how dangerous it is.

HN always have had a sizable anti-tech crowd (I don't want to say luddite because it's borderline pejorative). If you see the technology from close and you understand the human impacts of it then there is a reason some would rather stay clear from it. I know some FAANG engineers who doesn’t allow their kids to have smartphones or use social media even though they are themselves working at those companies. Why do you think that is? And you don’t even have to be a FAANG employee to see the social and human impacts of modern technology. AI is the same, in fact not even the same because it’s even worse and it will be only worse.

> Let’s face it: by the time I manually ship version 1.0 of a product, the AI-assisted version could have been deployed 10x faster.

Show the receipts. Where are the mobile apps, the photoshop replacements, the video and audio editors, the games and game engines that took a decade to make in the past that have shipped since Claude code came along?

> By then, enough real-world feedback would have surfaced to identify the major issues, and tools like Claude Code would make it possible to fix and ship version 2.0 at an incredible pace.

Again where are the receipts?

My experience with coding agents is that they’re perfectly good at generating a v0.1 that just about passes the sniff test. It does the first 90%, but the second 90% always takes longer than the first 90%. That second 90% is what coding agents are terriblle at, and are what make actually good products.

> I’ll probably make a lot of enemies by saying this, do people realize that code is just a means to an end?

You will need a lot more to make yourself my enemy, but this is the divisor between us... not that you like to use Claude and I don't.

I think it depends a lot on where your interest in (self) development lies.

My main motivator has always been to understand how things work, and myself being able to create as elegant solutions as my technical role models (in the range from colleagues and mentors to the elders of our field), hopefully even pushing it further. Having the LLM just create the product robs me of that, or at least of the most rewarding parts of that. And that's why I don't like to use it.

Different people are driven by different things, I don't think either trumps the other in the objective sense, we're just wired differently.

First of all, I expect to be a loser in the socioeconomical effect the AI brings. This isn't really about the technology itself but about the political systems we have now. From the pure job perspective, I will either lose my job, or will keep it but it will be increasingly more stressful and less interesting. There are literally zero benefits for me as a worker. The only hope is that the economic effect will be so huge, the trickle-down crumbs will be enough to live a decent life, but that is unlikely to happen in my country.

But, even if I had generational wealth behind me to be able to leverage the AI to my advantage, I still see a lot of cons in the way cheap content generation worsens the world around me: facilitating fraud, political shilling, disrupting online conversations (now everyone just sends bot summaries to each other). In a way, I feel a similar change that from the "pre-facebook" Internet to the "pre-chatgpt" Internet that happened in the early 10s.

I use AI tools daily and find them genuinely useful.

However I am increasingly annoyed at how everything has to be framed as a conversation about AI, how every tech-adjacent company has to brand itself as AI-first, and most of all, how overblown predictions are about an LLM being conscious, etc.

In short – it’s a useful technology reshaping tons of industries, but the hype is grating.

A lot of people on HN are anti-overhyping, which comes across as being opposed to the thing being overhyped. It was similar when cryptocurrency overhyping was popular.

>AI “writes bad code,” “introduces bugs,” “creates technical debt,” or something along those lines.

Which is undeniably true. I am not "anti-AI", in fact, it helps me immensely in my current job, but in my experience, using these tools still requires quite a lot of involvement. Otherwise, updating existing or adding new functionality becomes increasingly difficult as the system grows. The funniest thing is that when you start to lose touch with the internals of the system you are building, you cannot even give the AI proper context to pinpoint the problem or guide it to make specific changes, which results it a lot of wasted tokens, wrong assumptions and mountains of sloppy code.

HN is not anti-AI. HN reflects a reasonable ratio of pro-AI and anti-AI sentiments (sometimes held by the same person! because AI covers a lot of ground).

> They care that the product works

This reminds me of Anthropic's post where they say they ship 8x as much code as they used to.

And I stopped to consider how many times I've used an app and thought, "You know what this needs? More code!"

Personally, I don't know a place more hyped up about AI than HN. It turned from my daily dose of tech excitement into a daily dose of tech anxiety.

As for your argument, there's no such thing as elegance. Code "elegance" is mainly maintainability (and, to a smaller degree, some other aspects like security, performance, etc.). The importance of maintainability greatly varies between projects, industries and individual subjective viewpoints, resulting in the diversity of attitudes to AI-assisted coding. That, of course, assumes that AI cannot match humans in maintainability. Which seems to be the case to me right now. But it also seems that the gap is closing, not as much through AI writing "better" code, but mainly through it being increasingly capable of maintaining "bad" code.

>> Let’s face it: by the time I manually ship version 1.0 of a product, the AI-assisted version could have been deployed 10x faster. By then, enough real-world feedback would have surfaced to identify the major issues, and tools like Claude Code would make it possible to fix and ship version 2.0 at an incredible pace.

Really depends on what you are shipping, what your users expect and what your personal preference is. I do not want to go 10x on products that need high performance / high reliability, is deployed at large scale where its not easy to undo. But for other stuff, sure why not. The problem is everyone just puts everything in same basket. Either way, AI is useful but not to the same extent people claim it to be.

I'm not sure what the question is in your comment but this sentence is the only one that has a question mark:

>> I’ll probably make a lot of enemies by saying this, but do people realize that code is just a means to an end?

The answer is that, yes, most people do understand that. Some people like writing code regardless of whether they realise the extent to which that code is useful to others or not. Some people write code just as a means to an end and couldn't are less how that code is produced. There is, from what I've seen, a general expectation that the people who care about code are better at writing it but my experience is also that they just aren't. Some people who are "passionate about prorgramming" are horrible coders, some who are lukewarm or don't care at all are great coders.

I honestly don't know how AI comes into all that. HN is certainly not anti-AI. Just the other day a post made the first page with a title like "HN front page without the AI" but I can't find it now. There's as much AI stories submitted as anyting you could consider anti-AI.

Because for many people writing code is exactly their job, it isn't a means to an end.

It is like replacing people at the supermarket with self checkouts and expext they still feel fulfilled on their job, replenishing products from the warehouse.

Additionally only optimistics cannot see their job is in jeopardy.

If you deploy 10x faster, than me as business owner need less of you for the same amount of work.

No, the need for work doesn't grow exponentially every year, there is a physical limit to distribute among all people offering delivery capabilities.

Finally its environment impact destroys all the progress that was made in the last years, and brings computers prices back to the 1980's.

I use AI for small, stand-alone utilities, with limited risk/impact. These are things I make for only myself to use.

For anything something else will use, this is a real problem you can’t simply wave away.

> AI “writes bad code,” “introduces bugs,” “creates technical debt”

Bad code is hard to maintain, for the AI as well. While all code eventually becomes technical debt, the throw-away nature of AI code, and the sheer volume it creates, tends to turn it to technical debt more quickly. If code is “cheap”/“easy”, and the AI is bad at maintenance, why not do a full re-write for every update?

Bugs are bad for users and harder to find when you don’t have a deep understanding of the code. Logic bugs are the worst of these, because the AI and linter won’t necessarily find them. Some logic bugs can cause a lot of real world issues before they are reported and found. If doing a full re-write for updates, because the AI doesn’t care, each run is an opportunity for new bugs to get injected into the code.

If I’m expected to understand and answer for the code when there are issues, I need to be able to read and understand it, not just deploy quickly.

I've seen it before and over time it always plays out similarly.

- the cloud was invented and we were told CTOs would be able to just point and click and make infrastructure and apps! What did we actually get? Another layer of abstraction to debug through. Does it have is perks? Yes. Does it have it's own problems? Yes. Is it more expensive than setting up bare metal and having a solid team? Well that depends on what you're doing and the economics of it and the team.

- then document storages came along like and got wildly popular like mongo and people were calling it the end of SQL! And no more complexity or relational nonsense. Everything is JSON and life is great. What actually happened? These companies realized over time their data was getting trashed, adding things and fixing bugs became complex in pure docstore systems. So the initial v1 was easy and looked beautiful but only 4 years in you have a production db with orphan data that's twice as big as it should be. New features take forever to see a clear path in adding it to the model because it's no longer as intuitive to get performance for a feature.

Anyway. I see AI taking both of these roads at the same time. In 5 years I believe the code will be a giant pile of unfixable mess for most vibe coded things. I also don't see it getting rid of programmers but just adding another layer of abstraction to the mix that yes is helpful, but only if you already know what you're doing, much like what came of the clouds.

Because a lot of us are engineers. It's our mindset and our job to question hype and broad strokes and easy solutions, to go a few levels deeper and ask "okay but does it really work?". I don't think most people are anti AI more than they are anti any tool.

Interesting, my impression is that HN crowd is aggressively pro-AI, and on a level that is simply bizarre. E.g. the 'spec driven agentic engineering' that seems to be all the rage right now is pretty much a return to the 1970's waterfall development model, just with faster implementers. But initial implementation speed wasn't the bottleneck then as it isn't now - maybe people are too young to notice that we're (more or less) just witnessing another rotation of the big wheel of reincarnation ;)

HN probably has as much as 5 million monthly users. This is not just a small group of insiders, but more of a broadly representative sample of startup and engineering people.

So there is a wide range of judgement, and more importantly, a diverse set of worldviews. These are beliefs that form the foundations of cognition and perception. In the general population there are a massive number of people who do not understand technology and/or do not really appreciate it at a deep level. This includes a significant percentage of startup and engineering people unfortunately.

Because AI use correlates with sloppiness, and due to the fundamental attribution fallacy us engineers don't like sloppiness.

Let’s face it: by the time I manually ship version 1.0 of a product, the AI-assisted version could have been deployed 10x faster. By then, enough real-world feedback would have surfaced to identify the major issues, and tools like Claude Code would make it possible to fix and ship version 2.0 at an incredible pace.

This isn't really related to AI because it relates to manually coded things just as much, but on this point specifically this is only true for your very early I-gave-it-to-a-bunch-of-interested-people-to-try customers. It's much less true for your first paying customers, especially if the 'major issues' make their pain worse (e.g. data loss, time wasted, etc). You lose those ones for good, or until there's a critical mass of social proof to tell them the early problems are solved.

'I can dash out an early prototype with AI and then fix it later' is a dangerous mindset. If you're working in a small market with a limited number of customers you might piss off enough people that you won't be able to recover. There still has to be some level of quality. But it is a balance.

I think it's a negative feedback loop: If you are convinced that AI is worthless you are not going to be willing to spend $100-$200 to try the best tools, and if you're not using the best tools or you have to work under severe token limitations you are going to have a bad experience which is going to further convince you that AI is worthless.

LLMs as provided by Antrophic and OpenAI represents an enormous centralization of computing, and are black boxes that end-users are completely at the mercy of.

I think it also has to do with the experience after using AI for a while.

I have decided to leave what has been my workplace for over 15 years - mainly to do other types of work. Why? I stopped thinking in my current job as a system architect and dev. AI output is better or on par with my thinking, and I lost all feeling of achievement in my daily work.

I still want to use my brain, so I will find a job where I can continue to do that.

The elegance of the code is not superfluous at all. It correlates with the developer's understanding of both the code and the domain.

Many kinds of software cannot be yeeted 10x faster with AI. Someone has to sit down and understand what the right thing to do is, first.

It also matters how many users you expect to be able to reach. If you're Facebook you can afford to use the first 10,000 users as unpaid QA. If you're an indie shop that's barely getting downloads you really want to make a positive impression on your initial users or you're toast anyway.

At my workplace, we outsource a great deal. Of all the companies we outsource to, the employees of two are very upfront that they use LLM "assistance". Their output has been getting worse and worse since that started, about a year ago. Firmware produced with LLM assistance results in hardware that does not work reliably. Tools created or maintained with LLM-assistance do not function reliably. In short, LLM-created product doesn't work in my direct experience.

AI has done widely-accepted-as-bad things (e.g. spamming social media, datacenter physical consequences) and seems to have done less widely-accepted-as-good things (code improvement, but many people aren’t convinced. Is software today better than 1-2 years ago? I can’t find clear examples.)

Good or bad, I believe the genie is out of the bottle, and we should do the best we can using and mitigating AI. But I don’t blame or disagree with anyone who’s critical.

I’m more of an “AI centrist”, as I think the topic is extremely nuanced. As with most tech hype, there tends to be a black and white “AI good” or “AI bad”. I think reality is somewhere in the middle, personally.

> Let’s face it: by the time I manually ship version 1.0 of a product, the AI-assisted version could have been deployed 10x faster. By then, enough real-world feedback would have surfaced to identify the major issues, and tools like Claude Code would make it possible to fix and ship version 2.0 at an incredible pace.

It’s takes like this that remove the nuance completely and ignore so many facets of the debate. That being said, let’s assume this is true because I think vibecoding a CRUD app does make this realistic on the face of it. When I say vibecoding I mean prompting and dropping, not reading the code.

You do your adversarial reviews with multiple agents, you have your UX agent look over it, your security agent etc. Under the hood there are architectural issues. The code is probably passable, but rough.

You release, customers start using it for their business that they depend on for income. Issues start cropping up, you burn more and more tokens to fix the issues as they come up. Expedience starts sacrificing quality even still, architecture (if there was any) starts being violated and it degrades more and more.

I consider myself a professional, I would never want to end up in a situation like this for mission crititcal products. So, what do I do? I read the output, I make sure I understand it. Why? I care about my customers and secondarily I’m the one with the pager when something breaks down.

Now, for some fun hobby project to track my hobby paints for Warhammer… who cares? I agree. I have used Claude for such projects and not really cared. But your statement does not hold up in the enterprise world with mission critical software.

> At some point, execution speed starts to matter more than the elegance of the code.

This is reductive. You’re assuming people’s concern is “elegance”. It isn’t solely elegance. It’s domain understanding. It’s quality over all. It’s being a professional.

Writing the code was never the slow down for large scale enterprise products.

The post asks a question and then presents a strong opinion with the confidence of it being a fact. There is a pretence of curiosity veiling a complaint. I think it is this perceived lack of curiosity, casual exaggeration (“10X faster”) and implication of the “one true way” (“Let’s face it”) among AI supporters that grinds my gears at least.

It is more of a reaction to misrepresentation and falsehood, which AI and its rhetoric seems to have generated a lot of.

I think there are multiple aspects at play.

People have reasonable concerns about the ethical, political, economic dimensions before we even get to the technical capabilities.

Even within the narrower question of the technical utility, I think there are a lot of factors which differentiate people's different experiences which are largely unacknowledged and lead to people talking past each other and failing to understand how others have such different experiences and opinions.

The sentiment that "users only care if it works" for example implies that all considerations beyond "does the feature work today" are developers self serving their aesthetics, but overlooks many other concerns which _do_ impact users at a later point.

I wrote about just this division of experience and the polarisation which manifests on hacker news just last week as it happens https://www.julianhaeger.com/posts/Disparate-Results-with-AI...

I am against it simply because it corrupts reason.

Humans are trained for easy wins, not thinking about long term consequences and assuming risks will not play out. It's the lethal trifecta of emboldening our worst characteristics.

Not only that the majority of problems that we have with software engineering are self-inflicted and everyone has forgotten that they don't have to be. All this does is allow us to operate the pile of shit we built over the last 25 years with somewhat less efficiency than we did 25 years ago.

The customer does indeed care that the product works. But the engineers should care what they build and with what. And they don't any more. The status quo is using a statistical model to assemble a pile of trash.

Go back 30 years and I single-handedly built a full custom ERP system for a company in 3 months without an LLM in sight, with no internet connection and thus without pulling a single dependency off the internet.

Add the me-too culture over the top "we have to have an AI proposition" and it's a gold rush but the gold is muck.

The other day, I had a similar thought about the relative importance of code.

I'm working on a game, and I've been fussing over the code quality. And yeah, having code that isn't awful is important for various reasons. But with a game, it got me thinking, the code is literally the only part of the game the player doesn't experience.

The time I'm spending on the code, I could be spending on the art, the game design, the music, the story...

But my natural tendency is to hyper-focus on the only part of the game nobody will ever see. I thought that was interesting.

(That being said the codebase is ass and I do need to clean it up!)

This does not really match my observations. While it does feel to me the sentiment is shifting towards a more negative one, overall HN feels reasonably balanced between those that are pro-AI and those anti-AI (with the middle ground somewhat absent).

Unlike what many other comments here seem to suggest, HN seems much more pro-AI than what I see in real life amongst developers - at least where I live.

And I do think many users would care more than we might think, but unlike art etc. it is often more difficult to tell.

I think it's very useful but the hype promises so much more than it delivers. And a lot of the proponents are all in on the hype it gets annoying.

I use claude to write a design, review the design, turn that into an implementation plan, spend 2-3 turns reviewing that, but still when that is turned into code it misses things or creates helpers that's not actually used or... It creates massive files and unless I explicitly tell it to it never refactors them. It often just silences errors and warnings instead of actually fixing the problem.

It saves a lot of time, and I'm building things I couldn't have on my own. But it makes a lot of mistakes, it's far far from one shots which the hype keep going on and on about. It's tricky to put firm limits on what it does. A lot of the mistakes I catch because I've spent 15 years without an agent and sometimes it's just "hm, this smells weird" and I begin digging. I worry about the next generation.

For me the mental framing of "It's all hallucinations, some of those hallucinations are useful" is helpful to keep frustration in check as I ask it to review the same implementation plan for the 4th time and it turns up different issues because the input was slightly different, or review the output code and see allow(dead_code) despite my claude.md forbidding it.

Everyone knows eventually AI will impact their career prospects. How can you be in love with that

A few observations I’ve made.

1. Many criticisms of AI’s usefulness focus on the present/past. They may well be right about the current state, but the trajectory things are improving at (unless we’ve hit a brick wall) means they’ll probably be wrong in a few years.

2. Some of the most vocal anti AI statements I’ve heard have come from people the most impacted by its disruption. Does the concept it cheapens your prior work, or perhaps threatens to replace you, not lead to a huge amount of bias?

Like any technology, AI has good and bad aspects. There should be criticisms of it, we need those. But focusing on where it’s headed, and being aware of bias, is going to produce the most important discussions.

> Users don’t care whether the code was written by AI or by hand

This is very much not settled, and very much depending on your market. Selling games to gen-z? Yeah, they are going to care a lot.

It's a gut feeling.

We _know_ LLMs can't be _that_ good as they are promoted.

I've spent the last 6 months creating a production grade app from scratch with Claude where I wrote no single line of code. I've reviewed code and it was looking good, almost completely following my templates, workflows, skills.

Now I've started to make minor manual updates and I'm horrified. Claude has no idea why there were those templates and instructions in place. It followed them blindly without grasping their spirit. The end result is like a very junior dev copy-pasting answers from Stack Overflow into the codebase. No consistency, chaotic application of different conventions, duplicated code, ghost code (does nothing), and perhaps more as I'm digging in.

The pros: The code works, all tests pass (43% code / 57% tests, 1:1.3 ratio), the UI looks good with visible glitches

The cons: I'll have to rewrite most of the code on the long run, make it fit, easy to maintain.

The verdict: I wouldn't started this project alone. Claude get me through to v0.1.0 / MVP where I've focused solely on the product: technologies, architecture, functionality, and usability. Now it's easier to refactor all for v0.2.0 manually without Claude.

So this might be our gut feeling: we know it's something good, but not as good as the stakeholders might promote. We know it helps in some ways but it's a nightmare in other ways.

We are not anti-AI but rather pragmatic: Not that AI enthusiasts we are expected to be.

> Why is the HN crowd so anti-AI?

For most software engineers LLMs represent future financial uncertainty, pushing them into a tailspin. For a substantial subgroup the challenge runs deeper IMO. They are experiencing an identity crisis, as a central component of their self-concept is being violently stripped away on top of the financial uncertainty.

Everyone knows AI will impact their jobs eventually. How can one be in love with that idea

I think it depends on which side of the regression-to-the-mean machine that you land on (above or below the mean) for any given skill that is being disrupted by AI. From above, AI is frustrating; from below, it's magical.

https://halecraft.org/software-engineering-is-the-new-manufa...

Saying "code is just a means to an end" is like saying a Car is just a Vehicle to get from A to B. Which is true to a certain amount of people, but certainly not to people who need a special car like a truck and certainly not to the people who build cars. Developers are car builders.

Personally I'm not anti-AI, I'm anti-Stupid.

I also don't consider LLM's to be AI. I put it in the same category as PageRank.

It feels like the "Hype Wave" is going through oscillations before it damps down.

You know, it starts with hype about this new thing that is, perhaps, smarter than humans—or at least may soon be. And then the backlash comes and AI's reputation follows in descension.

And from such backlash and hate, some people start to say, "Hey, it's not so bad. It works for me." And maybe AI's reputation begins to swings toward the positive…

I suppose this is to be expected and perhaps ultimately healthy. I know for myself the swings in attitudes regarding AI have caused me to give pause and consider both sides with more regard than I might have. It has blunted some of my criticism and praise, sharpened others.

> there hasn’t been a single day where I’ve checked the HN Best RSS feed without seeing a post about how AI “writes bad code,” “introduces bugs,” “creates technical debt,” or something along those lines.

Confirmation bias. There has been pro-AI post every single day too but because you already decided that HN is anti-AI you didn't notice that.

I tend to agree with your overall point, but I think you reveal far more about yourself than you accurately reflect anything about HN. Just read back what you wrote:

> AI-assisted version could have been deployed 10x faster. By then, enough real-world feedback would have surfaced to identify the major issues

You want to ship major bugs to your users, let them find them, report them and fix them afterwards. You passively assume this is a good way to build software without even really questioning it.

Aside from some people just not liking this in principle, there are a lot of contexts where bugs cause actual harm and cost actual money. In some cases, "people dying" and "go to jail for it" type harm.