Hacker News Reader

Comment by alexyoung | Hacker News Reader

I realise I’m pissing into the wind here, but I find the LLM text on the homepage and docs quite funny/ awkward.

> “Ctrl+M turns it into a tiny player that wears any classic Winamp 2 skin, spectrum analyser, equalizer, and playlist included. 2000s vibes, pixel for pixel.”

Everything is said with too much intensity, and phrasing that sounds impressive but doesn’t really mean that much. Like “wears any classic Winamp 2 skin” feels so awkward, what’s “wears”? You mean it can use it?

I feel like if you can’t be bothered to write the code, at least document it yourself and write the marketing copy so I know you understand the product. Otherwise how can I trust running it on my computer? Did the LLM generate some amazing rm -rf somewhere, or another blunder that wrecks data I might care about?

View on HN

Author here.

Yes, the code is generated. Yes, the docs are generated. I agree LLM copy is annoying and will rewrite them myself.

But I built this for myself out of frustration with the official client, and I use it every day, so every bug I find is fixed immediately. I also care about bugs and PRs from other people, as you can see the amount that have been closed on the repo. The only reason I would stop maintaining this is if Spotify kills it.

I also wouldn't call this vibe coded either. Human directed and LLM generated is fairer. It's been an intense 5 days with tons of design decisions and testing on 3 platforms. I think we need a new vocabulary to distinguish it from "author doesn't care". I do.

I would use the phrase "Responsible Use of Generative AI", from https://www.debian.org/vote/2026/vote_002#texte

I wouldn't.

> The Debian Project nevertheless expects that all contributions submitted to Debian, regardless of how and with which tools they were produced, satisfy the same standards of quality, correctness, maintainability, and legal compliance.

Slop text on the homepage does not scream "quality" to me.

The code is worse, somehow. Endless paragraphs of comments that say nothing: https://github.com/crmne/fastpotify/blob/d9aafbf8e0b0c4d2da3...

    /// Rule: every block leaves exactly as long as it arrived, whatever
    /// is being held back, or the output would run short.

Uhm yeah, that is "literally what" territory.

Wtf? It’s putting its whole chain of thought in comment form, that’s insane.

No wonder people can’t get Claude to stop commenting if this is what it does unrestricted.

I'm going to sound harsh here, but please don't take this as criticism of the effort, it's criticism of how it's being portrayed. The quality might for all intents be great and useful.

> But I built this for myself out of frustration with the official client

But you didn't build it. This is akin to me telling an intern to build me a widget and then claiming it was me who built it. Or a construction manager saying they built a skyscraper.

Yes, the intern and the construction crew needs guidance and management, but I can't then go to the management committee and claim ownership of the effort.

> I also wouldn't call this vibe coded either.

You didn't write any of the code or even the documentation for that matter. You guided a code generator that relied on you to guide high level decisions but then did all the implementation on it's own. That's vibe coding.

The construction manager can’t say they built a skyscraper? In that only for someone who personally laid every brick and ran every wire?

But they know the plans, and knows people where all the wires are, etc.

With vibe coding you know nothing.

No they can't. The construction crew - which includes the manager - can say that they built a skyscraper.

If the construction manager did lay the bricks or did the majority of the work in all aspects, they could claim to have built a building. Otherwise they contributed work, no denying that, but they can't claim credit for building it. Same way that architects can't claim they built the skyscraper, even though they probably spent a lot of effort designing it to specs. To the best of my knowledge, no serious architect claims they "built" the building. They have designed it yes. Same way, the construction manager would claim they have managed the construction.

I admit that the classification is fuzzy, but there's an explosion of vibe coders going about claiming to have built things, whose contribution is basically spending a total of one hour with a prompt and and LLM doing basically all the heavy lifting.

That's not necessarily the case of the author here, but the line has blurred a lot and people very giddily claim credit for much less actual accomplishment than what would've flown in the past.

I think you're being pedantic to the point of losing all credibility. And you're gatekeeping in the worst possible way.

Your argument that using an LLM means you didn't "make" a thing could just as easily be used to say that if you used a framework or a library, you also didn't "make" a thing, because you didn't write all the code yourself and you didn't understand everything.

Would you say the authors of YT-DLP didn't "make" anything because they are a wrapper over FFMPEG and CURL?

If you produce something useful (even just to yourself), you "made" a thing.. It has a purpose, and it's useful. Who cares how you made it or how long it took?

For example I've made plenty of bash scripts over the years that were effectively thin wrappers that combined other tools, but they did something novel and useful that saved me tons of time, and I've shared those as gists or in other ways.

Some of these scripts took me mere minutes to make. I didn't use an LLM to make them, but the "heavy lifting" definitely wasn't in my script. No one ever showed up to berate me and try to tell me that I didn't actually produce anything useful in any of these cases.

So this "in the past" that you imagine where people were held to some bullshit standard for what is accepted as "making" a thing?

That past doesn't really exist, and if does where you are, it sucks.

People do care - lots of people do. Maybe you don’t, and maybe the argument the other poster made wasn’t fair, but you make just as spurious claims when you say „who cares how you made it or how long it took?”

Really? You care about some stranger's work and how they created it?

Pro tip: you shouldn't.

Care about your work if you want, but "caring" about how someone else makes things and how they talk about it is just being a judgemental busybody and again, gatekeeping.

As for my comment, I stand by it.

If I made something genuinely useful, then it should not matter if it took me 5 minutes or 5 days to make it.

You'll never convince me that more time spent making something (all other factors being equal - yes, quality matters) gives it more value or credibility.

Thanks for the tip haha. I am personally divided on it, but it’s more complicated than just quality or output for me.

I will apologize if I came off combative - was in the backseat of a car so may have been a little direct.

I think people care, all for different reasons. You care too - just when it comes to quality as you mentioned (and I’m sure other factors).

In this case many people correctly pointed out that there were clear indications of a lack of direct cognitive effort on behalf of the operator - including code quality, comment / git history, and including copy that he didn’t write. Now that doesn’t necessarily mean the LLM which did most of the work (or not? How can we know without an auditable prompt history) didn’t produce quality work.

When it comes to something that was made in 5 days AI assisted - the idea may be novel, but without a full prompt history it does become difficult to assess quality of cognitive effort on the operator at least.

In any case - it’s hard for me to know what to call the author. And I do think this applies to traditional media as well, so Im definitely not saying it’s all one sided (Bah! AI!)

That's fair, I guess I was really reacting to the narrow gatekeeping-like case the OP was advocating, where time spent/invested - or only doing things a certain way - was correlated to legitimacy and being able to claim you "made" a thing.

OK, clear enough. But that's an unusual definition of "build" that I don't think you should necessarily expect others to share.

The first entry for "build" in the OED is: To construct, put up, erect (a house or other dwelling or building); to construct (a permanent static structure such as a wall, bridge, or architectural feature, a road or railway, etc.). Also: to commission, finance, direct, or oversee the construction of (something of this type).

This debate is irrelevant. It doesn't matter. Having the idea and the skills to dictate a stable piece of software are all that matter.

There is no point in taking pride in typing out lines of code by hand. The entire world is moving on.

The problem is not the emittance of code by a statistical black box.

It’s not even knowing how it works or even what it used to achieve the desired result.

You wrote it or not, that’s your code. You are the one responsible. The machine has no brain. Use yours.

That may be, but if I come back to the old project I didn't touch in months, code may as well be written by an alien.

I can return to my 5+ year old code and start hacking in 30 minutes or so. I believe it boils down to documentation and comments.

Of course there's a point in it, don't be a boor.

Any act of making can be treated as product, as craft, or as art. You'll find yourself taking all three stances at different points. Any engineer or artisan or artist can choose to pour their attention and love into any layer of the production process.

I like ceramics. I like wheel throwing. Once upon a time, the pottery wheel was a newfangled technique for rapid and regular production of commodity goods. Same with coil building, same with slip casting, same with standardization of glaze recipes, gas and electric kilns, all of it.

But I know people who choose to quarry and purify their own clay, or to mix their own glazes, or to fire their pieces in a hole in the ground dug by hand with a fire they built by hand. I know people who make dozens of identical copies of just a few forms, after they spend months iterating on carving intricate molds. I know people who will spend a whole season working on a single hand-built bust.

And none of these people have ever, to my knowledge, expressed scorn for any of the others, for focusing on a different step in the process.

In the original sense of the word, vibe coding means neither reading nor caring about the code: https://simonwillison.net/2025/Mar/19/vibe-coding/

Some would use the term "Agent Engineering" for what OP did.

There's a trend where people go around claiming they built something and just happen to conveniently omit everywhere that everything was "built" using an LLM and the author doesn't even understand what the code does. That's the crux of my objection.

FWIW I agree with your post and perhaps we need a separate name for people who just instructed an LLM to build something, but then took the trouble to understand how the thing actually works.

It seems to me that this is not the case here. The author used LLM to build the whole thing and left it at that. Which is completely fair, I'd do the same if I just wanted to scratch an itch with the minimal amount of effort, but then I wouldn't go around claiming "I built it".

If I take a photo and then use GIMP to edit it, then did I take a photo or just direct my camera at something and then steer my computer to make changes to it? After all, it's not me drawing each pixel manually. It's vibe-photoed!

No. It's vibe-photoed when you use AI models which are trained on countless hours of actual human work. Otherwise you apply deterministic tools at your will and according to your skill level.

Shit take. You’re trying to play the bs semantics game and can’t even pull it off

The official client is very annoying.

I rewrite docs myself on my own projects because they read as very LLMish, there's an intermediate stage which feels slightly less bad where I ask it to remove most of the comments, especially LLM signposting ones and then "rewrite any remaining comments and docs in a dry ukgov/gds style" which makes things slighyly less worse.

Should add a full page click through slop warning to each page on the site. Also give credit to the LLM that made it in the "Brought to you by" section at the bottom.

I'll pick a human-written monstrosity over any "lean" consloption out there any time of day.

> Yes, the code is generated. Yes, the docs are generated.

> I also wouldn't call this vibe coded either.

It's vibe-coded. Get over it.

There is a lot of gray area between vibe coding and zero AI. Get over that too.

End of the day this person is solving a good problem and sharing the work. This is what we want.

Also: nagging about the specifics of the README's tone of voice, while the 600-1000MB RAM usage drops to 100-250MB seems a little childish to me.

It's still open source! Just like in the pre-AI era they are welcome to submit a PR and improve it in the way they like it. The main dev seems very open to accepting help.

Those numbers only look good in comparison.

I don't see you publishing something which further reduces the memory footprint?

Given the project is open source, have at it! See if you can help improve it instead of complaining. This is a significant improvement and a cool project!

I'm not complaining and it is a cool project (from what I can tell). But I think it's fair to say 200 MB memory usage is a lot without providing source code as a proof. Just by feel when you recall what were computers capable of when they did not have 100s of MB to spare.

I remember when I was scrounging memory sticks to get to 512 mb for my windows xp setup. And I was able to play Unreal: Return to Na Pali[0] on that.

[0] https://unreal.fandom.com/wiki/Unreal_Mission_Pack:_Return_t...

I run actual Winamp for my music streaming, and windows task manager claims it's using 4.1 MB.

Even that seems pretty high, but whatever, but given y'all are spending hundreds of mbs, I think I'm ok. Yes, spotify has features my subscription music service doesn't have... but the one I use lets me stream with Winamp.

Just like that comment you typed is keyboarded or keypadded. Or did you voice it?

A new term I encountered is “human-stewarded”.

I like the direction, but stewarded feels just a little too awkward to type or say to catch on. Perhaps "guide coded"? But it will be really hard to dislodge vibe coded from being generally applied to this concept, I think, in part because it is a catchy little term.

I came to like "vibe coding". Finding the right vibe in the words with the agent to steer them into the direction you want.

Whether the result is slob, depends on the programmer.

The fact than an app like this can be vibe-coded is actually mind-blowing. I'd love to see your prompt history here.

Oh no, that's closed source, lol

Don’t worry. Folks are entirely too concerned about the distinction between human made and LLM made. In a short handful of years, it will be incredibly rare to find something an LLM hasn’t been involved with

Agree, everyone thought I was crazy in 2025 when I was preoccupied with this question about moral 'authorship' in the AI era, but I think the culture will catch up quite soon, whether it likes it or not

https://sjsebastian.substack.com/p/thinking-through-the-mach...

I have no doubt that in the future AI code will be superior to human code. But today is not that day, and these projects that have never yet had human eyes on them are mostly shoddy pieces of work.

> I agree LLM copy is annoying and will rewrite them myself.

Remember when people used to care about making a good first impression?

That hasn’t existed in software generally for like 20 years, to be fair

> Yes, the code is generated. Yes, the docs are generated. I agree LLM copy is annoying and will rewrite them myself.

I find this kinda funny. Fix the obvious part, better not look at the code.

One thing I like about Spotify is the recommendations. I like to do what i call “beat digging” which is basically exploring the platform for new artists and tracks?

Does this still have that?

>Author here.

Not author. LLM manager, technical producer (if we're borrowing game development language), advanced AI user; not author. Authors are those whose code was used to train LLM(s) you were using while producing this fine piece of software. I'm saying this having neither positive nor negative emotions. Just clarifying.

I agree that "author cares" and "author doesn't care" is a very important distinction.

But to me vibe coded is about reading/editing the code. If all your code modifications are by prompt, it is vibe coded by definition.

Having used Claude a lot for the past four months, I get exhausted from its style. Every sentence has been buffed and polished and enhanced and it never just plainly says a thing. It's as if you have to eat every meal with every possible flavor enhancer in existence. At one point I will totally lose it if it mentions "smoke test" one more time.

It is also totally not what I want for docstrings in code or Powerpoint drafts, where consistency in language is critical.

Have a bit of food dropped into the MSG container. Now eat the whole container

Like I said in another thread, everything Claude (Code) writes sounds like it’s congratulating itself of how smart it is. While not actually being that smart.

It writes like me when I was 14 and thought I was very smart.

I believe AI technical prose peaked with Opus 4.6. I still use it (mostly for that purpose) and I think it's legitimately great. I'm hoping they are able to reverse the trends that benchmaxxing and RLAIF have wrought on Claude's non-code output with the future models.

Now that the public anger is growing i really hope Anthropics next models seriously address this. Or for that matter, i hope everyone (OpenAI/etc) starts focusing on this too.

Intelligence is important but it feels as if the LLM is only capable of conveying information while drenching it in vomit.

I use Claude daily and I have never heard it say 'Smoke Test' once.

Interestingly enough, https://fastpotify.rocks/winamp/ shows what I dislike about LLM-generated text: From "The buttons do what they say [...]" to "[...] a different format and are not supported" is just a huge wall of text without any visual guidance.

It's as if the person writing these lines didn't care about the result. One could argue that people not caring about written documentation isn't anything new, but in the past, these documentations also had little text. So they were easier to digest. Now, we get great walls of text that are hard to read. I cannot say that I like that.

Not running code from someone who has so little self awareness that they publish documentation like this tbh.

Doesn't matter if it's open source.

What sucks is that clearly the author has prompted the model somehow to generate this text, and this prompt probably contains the actual meat that the model just "expanded on" while adding it's own fluff, and if the author just shared the actual verbatim prompt instead, the documentation would be 90% shorter but also 200% more "informative per character used".

Just share the prompt you used, regardless of how low quality or bad it is, rather than this word spaghetti of 100s of words saying basically nothing, such a waste.

That isn't going to be very helpful is it?

> can you build a website to promote the app

> can you add a page showing off the winamp mode

Still more useful than what exists now.

That would actually be a reasonable thing to do with a model that is good at writing though. I'm sure if you used that method on Opus 4.6 you'd get a pretty decent result, because Claude models used to be quite good at that sort of technical synthesis. Asking a model that is bad at technical writing (e.g. Opus/Fable 5) to write your docs is a bad idea.

Ultimately what seems more important that how good the model is at task X vs task Y, seems to be that the human accepting/rejecting/iterating on the text can actually identify "good" from "bad" text, and makes the effort to fix it. Otherwise it doesn't matter much what model you use, it makes boring, bland and bloated prose.

It's a weird world where they put less time into writing that doc than you did reading it.

The sad thing is that may not be true. If you use LLMs to write docs and actually read it, you can spend far longer trying to convince the LLM to write in a sane way than to just write it yourself.

Semi related, but i've caught myself nitpicking small things from the code of an LLM only to realize i'm putting in more work trying to convince the LLM to make the code the way i want rather than just modifying it myself lol. It's as if i'm trying to educate it in the ideal way, as if it would learn and not repeat the mistake.. but it rarely does of course.

It seems to me that interacting with LLMs has a large pile of cognitive traps for humans.

I prioritized shipping over carefully crafting docs. I thought nobody would read them and they'd rather enjoy the software. I was wrong. I'll rewrite them.

I think you got a but too much negativity, from a crowd that became more easily irratated because their AI sensor became very sensitive by e.g. reviewing a lot of AI generated code and comments.

Fair enough. But then, how should I enjoy the software, if the documentation to its' usage is buried beyond a whole paragraph? To be clear, I do not judge the usage of AI per se. But you can instruct AI to write a clear manual. It just takes time. :)

Then, just do not add text that adds no value. Less is more.

Documentation is first part people see when presented. I personally hate AI written text even though I do agentic engineering a lot as well. Any user facing documentation must be reviewed, if you snoozy by reading it - remove or rewrite.

I do not care about paragraph long comments in code, because I’m not gonna read them, but I still follow my internal guideline and cut all this verbosity to level where it readable using hooks and skills.

So basicly like you said if you care about product do not add placeholders. It’s take away first impression.

I never thought ubiquitous human-written marketing wank was any better.

The difference is that the marketing stuff used to be only on corporate websites. Now it's everywhere, because pretty much everybody hates writing, and the "pretty much" can be left off when the writing is a job or some other form of obligation.

> I feel like if you can’t be bothered to write the code, at least document it yourself and write the marketing copy so I know you understand the product.

Do you not understand it?

This person doesn't owe you any work. He's providing a product for free and this is a reasonable enough description of the product.

> Did the LLM generate some amazing rm -rf somewhere, or another blunder that wrecks data I might care about?

I'm pretty sure at this point hand-crafted code is more likely to contain security bugs. The mythical "rm -rf" random LLM insertion is not a real threat. Every time I see stories about this kind of issue, it's almost always preceded by basically prompt poisoning.

> The mythical "rm -rf" random LLM insertion is not a real threat. Every time I see stories about this kind of issue, it's almost always preceded by basically prompt poisoning.

It is real, typically happening when dumber models (haven't personally seen any of the SOTA ones on reasonable thinking/reasoning level do this) try to clean up stuff via bash and forget to assign a variable. Things like "rm -rf sessions/$id" without double-checking $id is actually defined, would do "rm -rf sessions/" and remove the entire directory. Add in more variables in a path like "rm -rf $HOME/projects/$user/$repo/$tmpdir" and the dumb models people think they can use for local coding, promptly removes a lot more than expected.

Not sure how people aren't already running these agents in an environment/sandbox where stuff like this wouldn't matter, locally or not, but some LLMs do truly accidentally delete stuff for people, without "prompt poisoning", I had it happen myself while testing Qwen3.8-27B (BF16) locally just a week or two ago, exactly as described above.

Just recently, Opus 4.8 almost deleted a bunch of important files (I caught the issue before) because it failed to consider that two-way-sync would delete files from my NAS if deleted locally.

Fable has done similar for me because it assumed intent that I wanted permanent deletion after moving folders. The bash script it wrote failed to copy the files and then deleted the originals. Not catastrophic in my case though

> This person doesn't owe you any work. He's providing a product for free

A product being free doesn’t make it exempt from criticism. It’s not about “owing” anybody work.

> and this is a reasonable enough description of the product.

It’s not a great idea to signal a lack of effort, nor is it wrong for users to pick up on that signal. This is the post-llm equivalent to spelling or grammatical errors.

> I'm pretty sure at this point hand-crafted code is more likely to contain security bugs.

LLMs produce code which is fairly ‘locally optimal’, so in good hands they can absolutely be used to build correct software. They also allow those with little to no experience* to cobble something together very quickly and easily while creating a ton of security holes. I think you might be underestimating how common the latter is.

* not to suggest this applies to the OP necessarily. Rather some extra wariness is necessary nowadays since the floor has been lowered so much.

> A product being free doesn’t make it exempt from criticism. It’s not about “owing” anybody work.

You can criticize it but your criticism is hallow

> It’s not a great idea to signal a lack of effort, nor is it wrong for users to pick up on that signal. This is the post-llm equivalent to spelling or grammatical errors.

Why should you put in effort to something that should be effortless? Why should I waste my time writing some inferior copy by hand?

> LLMs produce code which is fairly ‘locally optimal’, so in good hands they can absolutely be used to build correct software.

Humans produce code which is fairly ‘locally optimal’

Also, interesting you use ‘ as a quote rather than '. I notice these quotes are used with LLM generated text, as I can't find this on my keyboard.

> Why should you put in effort to something that should be effortless?

Why should it be effortless?

> Why should I waste my time writing some inferior copy by hand?

Have some faith in yourself :) If the LLM writing was superior then surely nobody would’ve noticed?

And to the question: because lack of care in one area implies that that the same approach might extend to other areas. Nobody’s forcing you, but it definitely sends a particular message.

> Humans produce code which is fairly ‘locally optimal’

I’m taking this to mean you find the phrasing vague. LLMs are good at avoiding the sort of fine-grained security bugs that you see in some human written code.

> Also, interesting you use ‘ as a quote rather than '. I notice these quotes are used with LLM generated text, as I can't find this on my keyboard.

Funny, but FYI a fair few mobile keyboards use non-ascii punctuation

If it can transform into Winamp I don't care if the code is LLM generated, typed by infinite monkeys or found in some ancient scrolls.

Jokes aside, how can you be sure that the official Spotify app (or any app, or Windows) was not written by a LLM without supervision?

The official Spotify app has some skin in the game. Random dudes self promotion project, not so much.

I suspect the official app would actually be more consistent and not as bad to use if it was generated by an LLM.

Not OP but I don’t really care if it was or not. All I know is that I’ve been using the Spotify app 10+ years and it works well. I don’t know if this random app works, and if it hasn’t been supervised I doubt it does.

Gave it a quick try today, it works pretty well and is noticeably faster than the official app.

I too enjoy wearing the RealPlayer skin of the 90's. Makes me look very chic.

If the last twenty years of horror movies have taught me anything it is that maybe you can't be a real player, but you can always wear their skin!

LLMs seem to have been fine-tuned to use metaphors, presumably because they read like vivid creative writing. The over-use of metaphor in LLM-produced text is exhausting.

Slop code is already hard to tolerate, but any hint of slop in text will immediately make me chuck it into the bin / close the tab, no exceptions. My lifetime is finite, I'm sure as fuck not gonna spend time reading something that a person could not even be arsed to write themselves. I'm only annoyed that it takes sometimes a few seconds, sometimes a few minutes, to detect that I'm actually not reading a person's thoughts but only a machine simulacrum of such.

I highly recommend anybody to take up a similar rule, in any context. Slop in a website? Close. Slop in an email? Bin. Slop in a code comment? Skip.

Uncanny valley. LLMs are great at producing text that reads almost right.

Did the landing page text get updated? Didn’t click as ai generated to me.

> class BuffaloBill

Your comment has too much intensity as well. Honestly it too looks like slop. I agree with what you said, but nevertheless your writing is likely AI generated here, it’s very obvious.

There is a tribe of people who carve bots out of trees using their hands. https://en.wikipedia.org/wiki/Dugout_canoe. It stupid to waste time doing something a computer can do reasonably well. IMO

Icymi, "carving bots out of trees" is a wonderful and apposite autocorrect. There's layers to this thing, man!

Generally speaking, it's carving bots out of the environment, trees included, for sure.

Reply on news.ycombinator.com