Hacker News

The other element here is that the vibecoder hasn't done the interesting thing, they've pulled other people's interesting things.

Let's see, how to say this less inflamatory..

(just did this) I sit here in a hotel and I wondered if I could do some fancy video processing on the video feed from my laptop to turn it into a wildlife cam to capture the birds who keep flying by.

I ask Codex to whip something up. I iterate a few times, I ask why processing is slow, it suggests a DNN. I tell it to go ahead and add GPU support while its at it.

In a short period of time, I have an app that is processing video, doing all of the detection, applying the correct models, and works.

It's impressive _to me_ but it's not lost on me that all of the hard parts were done by someone else. Someone wrote the video library, someone wrote the easy python video parsers, someone trained and supplied the neural networks, someone did the hard work of writing a CUDA/GPU support library that 'just works'.

I get to slap this all together.

In some ways, that's the essence of software engineering. Building on the infinite layers of abstractions built by others.

In other ways, it doesn't feel earned. It feels hollow in some way and demoing or sharing that code feels equally hollow. "Look at this thing that I had AI copy-paste together!"

People who got "enabled" by AI to produce stuff, just need to learn to keep their "target audience of one"-projects to themselves. Right now it feels like those fresh parents who show every person they meet the latest photos / videos of their baby, thinking everybody will find them super cute and interesting.

I have yet to see any of these that wouldn’t have been far better off self-hosting an existing open source app. This habit of having an LLM either clone (or even worse, cobble together a vague facsimile) of existing software and claiming it as your own is just sort of sad.

To be fair, one probably needs at least one idea in order to build stuff even with AI. A prompt like "write a cool computer program and tell me what it does" seems unlikely to produce something that even the author of that prompt would deem worthy of showing to others.

Sometimes 'gatekeeping' is a good thing.

> so projects are built whether or not they're good ideas

Let’s be honest, this was always the case. The difference now is that nobody cares about the implementation, as all side projects are assumed to be vibecoded.

So when execution is becoming easier, it’s the ideas that matter more…

one of the base44 ads is hilarious about this

https://www.youtube.com/watch?v=kLdaIxDM-_Y

Agreed, and were gonna see this everywhere that AI can touch. Our filter functions for books, video, music, etc are all now broken. And worst of all that breaking coincides with an avalanche of slop, making detection even harder.

There is this real disconnect between what the visible level of effort implies you've done, and what you actually have to do.

It's going to be interesting to see how our filters get rewired for this visually-impressive-but-otherwise-slop abundance.

"One of the great benefits of AI tools, is they allow people to build stuff, even if they have no ideas or knowledge."

Wait, what? That's a great benefit?

> I’ll bet the abandonment rate for side projects has skyrocketed

My experience is the opposite. It’s so much easier to have an LLM grind the last mile annoyances (e.g. installing and debugging compilation bullshit on a specific raspberry pi + unmaintained 3p library versions.)

I can focus on the parts I love, including writing them all by hand, and push the “this isn’t fun, I’d rather do something else” bits to a minion.

>So let's say we have a generation of people who use AI tools to code and no one really thinks hard about solving problems in niche spaces.

I don't think we need to wait a generation either. This probably was a part of their personality already, but a group of people developers on my job seems to have just given up on thinking hard/thinking through difficult problems, its insane to witness.

> Then what will AI be trained on in let's say 20-30 years? Old code? It's own AI developed code for vibe coded projects?

I’ve seen variation of this question since first few weeks /months after the release of ChatGPT and I havent seen an answer to this from leading figures in the AI coding space, whats the general answer or point of view on this?

AI will be trained on the code it wrote, our feedback on that code, and the final clean architecture(?) working(?) result after that feedback.

Exactly. Prose, code, visual arts, etc. AI material drowns out human material. AI tools disincentivize understanding and skill development and novelty ("outside the training distribution"). Intellectual property is no longer protected: what you publish becomes de facto anonymous common property.

Long-term, this is will do enormous damage to society and our species.

The solution is that you declare war and attack the enemy with a stream of slop training data ("poison"). You inject vast quantities of high-quality poison (inexpensive to generate but expensive to detect) into the intakes of the enemy engine.

LLMs are highly susceptible to poisoning attacks. This is their "Achilles' heel". See: https://www.anthropic.com/research/small-samples-poison

We create poisoned git repos on every hosting platform. Every day we feed two gigabytes of poison to web crawlers via dozens of proxy sites. Our goal is a terabyte per day by the end of this year. We fill the corners of social media with poison snippets.

There is strong, widespread support for this hostile posture toward AI. For example, see: https://www.reddit.com/r/hacking/comments/1r55wvg/poison_fou...

Join us. The war has begun.

It's because the common belief that you should build copies of whatever SaaS makes decent money. What they don't mention is that people need to have a very good reason why they decide to go for your bare-bones MVP instead of a well-established solution.

Agreed. I'm over here working on Quake 2 mods and reverse engineering Off world trading company so I can finish an open source server for it using AI.

Thing is I worked manually on both of these a lot before I even touched Claude on them so I basically was able to hit my wishlist items that I don't have time to deal with these days but have the logic figured out already.

Communicating that you know what you are talking about and that you're different is a lot of work. I think being visibly "anti-AI" makes you look as much of an NPC as someone who "vibe coded XYZ." It takes care, consistency and most of all showing people something they've never seen before. It also helps to get in the habit of doing in person demos, if you want to win hackathons it really helps to be good at (1) giving demos on stage and (2) have a sense of what it takes to make something that is good to demo.

I have two projects right now on the threshold of "Show HN" that I used AI for but could have completed without AI. I'm never going to say "I did this with AI". For instance there is this HR monitor demo

https://gen5.info/demo/biofeedback/

which needs tuning up for mobile (so I can do an in-person demo to people who work on HRV) but most all being able to run with pre-recorded data so that people who don't have a BTLE HR monitor can see how cool it is.

Another thing I am tuning up for "never saw anything like this" impact is a system of tokens that I give people when I go out as-a-foxographer

https://mastodon.social/@UP8/116086491667959840

I am used to marketing funnels having 5% effectiveness and it blows my mind that at least 75% of the tokens I give out get scanned and that is with the old conventional cards that have the same back side. The number + suit tokens are particularly good as a "self-working demo" because it is easy to talk about them, when somebody flags me down because they noticed my hood I can show them a few cards that are all different and let them choose one or say "Look, you got the 9 of Bees!"

"This repository contains only [Organic] and [Hand-Made] ingredients."

It seems silly, but I know I'm more likely to review an implementation if can learn more about the author's state of mind by their style.

I added the following at the top of the blog post that I wrote yesterday: "All words in this blog post were written by a human being."

I don't particularly care if people question that, but the source repo is on GitHub: they can see all the edits that were made along the way. Most LLMs wouldn't deliberately add a million spelling or grammar mistakes to fake a human being... yet.

As for knowing what I'm talking about. Many of my blog posts are about stuff that I just learned, so I have many disclaimers that the reader should take everything with a grain of salt. :-) That said: I put a ridiculous amount of time in these things to make sure it's correct. Knowing that your stuff will be out there for others to criticize if a great motivator to do your homework.

You’re not actually at risk of being labeled as LLM user until someone comes and make that claim about your work. So my advice is to not try to fight a preemptive battle on your tone and adjust when/if that day comes.

Side note: I’d think installing Anubis over your work would go a long way to signaling that but ymmv.

> I still have no idea how to signal to people that I actually know what I’m talking about.

presumably if this is true, it should be obvious by the quality of your product. If it isnt, then maybe you need to need to rethink the value of your artisanal hand written code.

When I was growing up (millennial) it seemed to me that the default for smart young people drawn to technical problem solving was something like aerospace, software or hardware was more or less a fun hobby, like it was for Steve Wozniak. Nobody cared whether or which of these were a commodity, which is what happens when you actually enjoy something.

These days I do see a lot of people choosing software for the money. Notably, many of them are bootcamp graduates and arguably made a pivot later in life, as opposed to other careers (such as medicine) which get chosen early. Nothing wrong with that (for many it has a good ROI), but I don’t think this changed anything about people with technical hobbies.

When you’re young, you tend not to choose the path the rest of your life will take based on income. What your parents want for you is a different matter…

Prompter News

This, but unironically. Every submission should have an "AI?" checkbox, to indicate if the content submitted is about AI or made by AI, because I'm just absolutely fed up with 2/3 of HN front page being slop or meta-slop.

I'm not an anti-AI luddite, but for gods sake talk about (ie. submit) something else!

listen_to_what_the_man_said.stm

That may be something.

Having too may subs could get out of hand, but sometimes you end up with so much paperwork generated so fast that it needs its own dedicated whole drawer in your filing cabinet ;)

http://news.ycombinator.com

You’re part of the problem.

Sqfty?

I mean it's a real problem, but it's also a solved problem, and also not a problem that comes up a lot unless you're doing the sort of engineering where you're using a CAD tool already.

I don't doubt it's useful, and seems pretty well crafted what little I tried it, but it doesn't really invite much discussion.

Worse yet, if you're not an expert(with autodidacts potentially qualifying), your ideas won't be original anyway.

You'll be inventing a lot of novel cicular apparatus with a pivot and circumferencrial rubber absorbers for transportation and it'll take people serious efforts to convince you it's just a wheel.

My new favorite thing to do is grab the 'What are you working on threads' and have a LLM group them categorically with one line descriptions of each app.

Combining the two approaches might work. A "pre-moderation queue" for submissions that are are solid enough to pass the "Show" bar, and then the monthly "what are you working on" threads as a more free-form creative outlet.

And yes, disclosing the use of AI should be par for the course.

Interesting take on the sharing of prompts. I don’t think this is a bad idea. How would this work though given different prompts occur in different context windows?

Sharing prompts, not sure it works if your project required hundreds of prompts? It’s all in history though (.jsonl) so I’m sure the AI can condense it somehow.

> What I see is new users who are trying to share something without having yet understood HN.

From their perspective, HN is another place to post and get views on their project, part of a check list for their "launch" or whatever, not everything comes from within the ecosystem.

Some posts their projects then never reply to any of the comments, while for me (and many others I bet) half the reason of posting a Show HN is because I'm looking for participating in discussions about my thing and understanding different perspectives thinking about it too.

> I want to believe that most new users mean well, they just need better explanations.

Yeah, so far the only thing I know of is the "Please read the Show HN rules and tips before posting" blurb on the /show list, and the separate pages. Maybe some interstitial or similar if the title prefix-matches with "Show HN" could display the rules, guidelines and "netiquette" more prominently and get more people to be aware of it.

Back when I ran a WoW guild, the first sentence in our recruitment post emphasized the importance of reading the whole post (because the way to access the application form was to click the only smiley in the post, and this detail was mentioned in the last paragraph).

Most people did not read the post, which was immediately evident from how they posted their application by copy-pasting and editing an application posted by someone else before them.

> For example, in one project, PRs have to be submitted to the "next" branch and not the default branch. This is written in the CONTRIBUTING.md file, which is linked in the PR template, with the mention that PRs that don't respect that will be close. Most if not all submitters of low-quality PRs don't do anything once their initial PR is closed.

Few things in life are as reliable and trustworthy as the laziness of others.

I'm 90% sure this will end with endless squabbles who's right that the label is correct/incorrect, rather than actual conversations about what the project that the person is showing. It already happens without the labels, feels like it'd increase the frequency of that even more if this label gets enforced.

Is the problem that the app was written with AI assistance or that it's low-effort/bad? I don't care if you used Claude to fix a bug or something if you have a cool app, but i do care if you vibe coded something I could've vibe coded in an hour. That's boring.

Feels like effort needs to be the barrier (which unfortunately needs human review), not "AI or not". In lieu of that, 100 karma or account minimum age to post something as Show HN might be a dumb way to do it (to give you enough time to have read other people's so you understand the vibe).

I thought so too, untill I looked for 1-point Show HN posts with a repo with a long commit history. Some of these are really cool (see my article), but others were not compelling at all, at least to me.

Eh IMO any metric like this can be gamed. My project that reached hn front page was coded in a short time (and yes some ai was used), but otoh I think it was something that showed hey you can do this really interesting thing (in my case vlm based indoor location).

Also its not uncommon for weekend projects to be done in a shprt span with just a "first commit" message dump even pre-AI.

on problem here is that I don't want to share my ShowHN projects with my HN account because it'll connect my real identity to my pseudonymous HN identity.

so, in the past, i've created throwaway HN accounts for sharing things that connect to my real ID.

Honestly I would support only allowing Show HN for accounts that can downvote. It's really not such a high threshold.

Funny a year ago I used to hear from so many people who thought a Product Hunt launch was the same as a marketing plan but it's been a while since I've heard about Product Hunt...

I think this already happens but the threshold might be pretty low - there's a separate shownew page and if it gets sufficient upvotes it appears in the regular "Show HN".

it was in this link included in this same story https://johan.hal.se/wrote/2026/02/03/the-sideprocalypse/

18 relevant results in 0.96s, from Kagi.

You shouldn't be browsing games while working, that's why

They probably have a newly registered domain rule.

Thanks for the blunt feedback - genuinely appreciate it.

You're right that children's books can be excellent, and for generic topics a well-reviewed book from a skilled author and illustrator will beat what we generate. No argument there.

Where we see real value is in the gaps the publishing industry doesn't serve. Bilingual families who can't find books in Maltese/English or Estonian/German. A child with an insulin pump who wants to see a superhero like them. A kid processing their parents' divorce. A child with two dads, or being adopted, or starting at a new school in a country where they don't speak the language yet. No publisher will print a run of one for these families - but these are exactly the stories that matter most to them.

On the UX points - you're right on both. We should localize the showcase to your language, and the signup wall before trying is too much friction. Working on both.

What the actual f. I am one of those 50 million people, played Blockheads all the time as a teen and had no idea I would randomly stumble over the dev around here. Great work indeed, thanks for the fond memories I made on my 4th gen iPod touch, playing that game :)

If those announcement posts don't take off, how do you end up finding a community of users/players?

It's also a pretty big lottery. Two nearly identical projects can get very different receptions depending on the phase of the moon.

Yes, this was in a marketing niche and I knew it wasn't a good fit

Our revenue isn't public and this is a throwaway

Affiliates. They made it blow up overnight with an existing audience

I'd push back on this and say that the #1 problem with the discourse about AI now (e.g. why I'd almost never upvote a blog post about AI coding) is that it is too focused on 2026-02-17. That is, I could care less about optimizing to pick the best model or agentic workflow because it's all going to be obsolete in a year.

I am wary of blogs by celebrity software managers such as DHH, Jeff Atwood, Joel Spolsky, and Paul Graham because they talk as if there was something about their experience in software development and marketing except... there isn't.

The same is true for the slop posts about "How I vibe coded X", "How I deal with my anxiety about Y" and "Should I develop my own agentic workflow to do Z?" These aren't really interesting because there isn't anything I can take away from them -- doomscrolling X you might do better because little aphorisms like "Once your agent starts going in circles and you find yourself arguing it you should start a new conversation" is much more valuable than "evaluations" of agents where you didn't run enough prompts to keep statistics or a log of a very path-dependent experience you had. At least those celebrity managers developed a product that worked and managed to sell it, the average vibe coder thinks it is sufficient that it almost worked.

You could write in the language you are most eloquent in and prompt a bot to do a faithful translation.

Post both versions

I’d rather read broken english than LLM output.

HN already has a mechanism for flagging posts. If flagging low effort/trivial show HN posts were normalized, I suspect it would work just fine: if a human can't easily decide whether or not to flag it, they likely wouldn't.

As dang posted above, I think it's better to frame the problem as "influx of low quality posts" rather than framing policies having to do explicitly with AI. I'm not sure I even know what "AI" is anymore.

OK, let's say there are two categories of software now.

One is where the human has a complete mental map of the product, and even if they use some code generating tools, they fully take responsibility for the related matters.

And there is another, emerging category, where developers don't have a full mental map as it was created by an LLM, and no one actually understands how it works and what does not.

I believe these are two categories that are currently merged in one Show HN, and if in the first category I can be curious about the decisions people made and the solutions they chose, I don't give a flying fork about what an LLM generated.

If you have a 'fog of war' in your codebase, well, you don't own your software, and there's no need to show it as yours. Same way, if you had used autocomplete, or a typewriter in the time of handwriting, and the thinking is yours, an LLM shouldn't be a problem.

I don't think so. At the end of the day it's just a tool.

Case in point: aside from Tabbing furiously, I use the Ask feature to ask vague questions that would take my coworkers time they don't have.

Interestingly at least in Cursor, Intellisense seems to be dumbed down in favour of AI, so when I look at a commit, it typically has double digit percentage of "AI co-authorship", even though most of the time it's the result of using Tab and Intellisense would have given the same suggestion anyway.

Maybe. It's basically "people who know how shit works" vs "people who don't know how shit works". I hope we still have at least some people in category 1 or else we just end up with Wall-E.

I think it's just numbers. There are maybe a few dozen people that see your post on /new. That's a tiny sample size, not a good proxy for how interesting the post is. You see this on Reddit as well where the same exact post gets 1 upvotes and then finally blows up.

Chasing clout through these forums is ill advised. I think people should post, sure. But don't read into the response too much. People don't really care. From my experience, even if you get an insanely good response, it's short lived, people think its cool. For me it never resulted in any conversions or continued use. It's cheap to upvote. I found the only way to build interest in your product is organic, 1 on 1 communication, real engagement in user forums, etc.

I'll counter you this: I don't use AI at all, but in a way I am a vibe coder, even though I'm five years of full time work into one OSS project. I have no code review. I move fast and ship bugs. I roll forward.

I see no reason to disrespect your work from what you say, but I also see no reason that AI would be much help to you after you had been learning for a year. If you are in the loop, shouldn't this be just about the moment when your growing abilities start to easily outpace the model's fixed abilities?

I view it as do you have a full mental model of the code base.

If you do then not vibe coded.

For me, I have different levels of vibes:

Some testing/prototyping bash scripts 100% vibe coded. I have never actually read the code.

Sometimes early iterations, I am familiar with general architecture, but do not know exact file contents.

Sometimes I have gone through and practically rewrote a component from scratch either because it was too convoluted, did not have the perfect abstraction I wantet/etc.

For me the third category is not vibe coded. The first 2 are tech debt in the making.

vibe coding is no-look coding, it's largely being replaced by agents that do the iteration to the point no human is involved beyond initial project description like "Build me a web browser"

>At what point does a project stops being considered vibe coded?

Good question, and by the same "token" when does it start?

Maybe if there's no possible way the creator could have written it by hand, perhaps due to almost complete illiteracy to code in any language, or something like that, it would be a reference point for "pure vibe". If the project is impressive, that's still nothing to be ashamed of. Especially if people can see the source code.

All kinds of creative people I see are mostly no dummies and it might be better than nothing for them to honestly rate their own submissions somewhere on the scale from pure vibe to pure manual?

With no stigma regardless, and let the upvotes or downvotes from there give an indication of how accurate the self-assessments are. Voting directly to Show HN could even have a different "currency" [0] to help regulate the fall of Show submissions, where a single upvote could mean something like infinitely more than zero.

I'm not disappointed by a project purely vibed by somebody like a visual artist, storyteller, or business enthusiast who has never written a line of code, as long as it is astoundingly impressive, in the league of the better projects, those I would like to take a look at.

I also see real accomplished coders guide their agents to arrive at things that wouldn't be as nice if they didn't have years of advanced manual ability beforehand.

Plus I think I'm in the vast majority and have no interest in "slop", in a way that aligns with so many kinds of people who are also turned off.

But so far, the best definition we have for slop is "we know it when we see it".

Oh, well that's all I've got, so far :)

[0] slop vs non-slop which is like pass/fail, or even a numerical rating could be on the "ballot".

I feel similar but since starting to use Claude (only 4 days ago) I get it. "Vibe coding" to me is just a new term for pair programming.

Ah yes, I was being dense. I was so obsessed with the steepness of the last point drop and completely missed the overall trend line! Thanks.

Or a button to "vote as slop" which would automatically move it from "Show HN" to "Slop HN" :D

We probably just need to cultivate a culture of not upvoting them in the first place.

It's better if you don't have to learn to program to make applications.

In the future it will seem very strange that there was a time when people had to write every line of code manually. It will simply be accepted that the computers write computer programs for you, no one will think twice about it.

There's a difference between documentation and LLMs. An LLM can be your own personal tutor and answer questions related to your specific code in a way no documentation can. That is extremely helpful until you master the programming language enough.

Programming has been democratized in terms of “time invested in programming” by AI, which has resulted in exactly what happens to any high-investment community when a tool-assisted method of avoiding that investment is developed. You could ask any newspaper or movie script submissions reviewer before AI what percent of what they receive as uninvited-submissions is even slightly worth their time and they’ll look at you with the deadest eyes in the world and say “zero percent”. What invention led to their industries being buried in meaningless (relative to pre-invention) submissions that took a thousandth of the effort to produce than they did prior to it, without the editorial staff being scaled accordingly? The typewriter.

The obvious counterpoint is that AO3 is brilliant, which it is: give people a way to ontologize themselves and the result is amazing. Sure, AO3 has some sort of make-integer-go-up system, but it reveals the critical defect in “Show HN”: one pool for all submissions means the few that would before have been pulled out by us lifeguards are more likely to drown, unnoticed, amidst the throngs. HN’s submissions model only scales so far without AO3’s del.icio.us-inherited tagging model. Without it, tool-assisted creative output will increasingly overwhelm the few people willing to slog through an untagged Show HN pool. Certainly I’m one of them; at 20% by weight AI submissions per 12 hours in the new feed alone, heavily weighted in favor of show posts, my own eyes and this post’s graphs confirm that I am right to have stopped reading Show HN. I only have so much time in my day, sorry.

My interest in an HN post, whether in new or show or front page, is directly proportional to how much effort the submitter invested in it. “Clippy, write me a program” is no more interesting than a standard HN generic rabble-rousing link to a GotHub issue or a fifty-page essay about some economics point that could have been concisely conveyed in one. If the submitter has invested zero personal effort into whatever degree of expression of designcraft, wordcraft, and code craft that their submission contains, then they have nothing to Show HN.

In the rare cases when I interact with a show post these days, I’ve found the submissions to be functionally equivalent to an AI prompt: “here’s my idea, here’s my solution, here’s my app” but lacking any of the passion that drives people to overcome obstacles at all. That’s an intended outcome of democratization, and it’s also why craft fairs and Saturday markets exercise editorial judgment over who gets a booth or not. It’s a bad look for the market to be filled with sellers who have a list of AI-generated memes and a button press, whose eyes only shine when you take out your wallet. Sure, some of the buttons might be cool, but that market sucks to visit.

Thus, the decline of Show HN. Not because of democratization of knowledge, but because lowering the minimum effort threshold to create and post something to HN reveals a flaw-at-scale of community-voting editorial model: it only works when the editorial community scales as rapidly as submissions, which it obviously has not been.

Full-text search tried to deprecate centralized editorial effort in favor of language modeling, and turned out to be a disastrous failure after a couple decades due to the inability of a computer to distinguish mediocre (or worse) from competent (or better). HN tried to deprecate centralized editorial effort and it has survived well enough for quite some time, but gestures at Show HN trends graphs it isn’t looking good either. Ironically, Reddit tried to implement centralized moderation on a per-community basis — and that worked extremely well for many years, until Reddit rediscovered why corporations of the 90s worked so hard to deprecate editorial staff, when their editors engaged in collective action against management (something any academic journal publisher is intimately familiar with!).

In that light, HN’s core principle is democratizing editorial review — but now that our high-skill niche is no longer high-skill, the submissions are flooding in and the reviewers are not. Without violating the site’s core precepts of submission egality and editorial democracy, I see no way that HN can reverse the trend shown by OP’s data. The AO3 tagging model isn’t acceptable as it creates unequal distinctions between submissions and site complexity that clashes with long-standing operator hostility towards ontologies. The Reddit and acsdemic journal editorial models aren’t acceptable as it creates unequal distinction between users and editors that clashes with long-standing operator hostility towards exercising editorial authority over the importance of submissions. And HN can’t even limit Show HN submissions to long-standing or often-participating users because that would prevent the exact discoveries of gems in the rough that show used to be known for.

The best idea I’ve got is, like, “to post to Show HN, you must make several thoughtful comments on other Show HN posts”, which puts the burden of editorial review into the mod team’s existing bailiwick and training, but requires some extra backend code that adds anti-spam logic, for example “some of your comments must have been upvoted by users who have no preexisting interactions with your comments and continued participating on the site elsewhere after they upvoted you” to exclude the obvious attack vectors.

I wouldn’t want to be in their shoes. A visionary founded left them a site whose continuing health turn out to hinge upon creating things being difficult, and then they got steamrolled by their own industry’s advancements. Phew. Good luck, HN.