This is why you should not install any of the vibe coded apps that get advertised in here regularly. You’re just creating a liability for yourself.
@rimu@piefed.social given the new post tag requirements, please edit your title to include [AIT].
My coworker said it best: “Anyone can build an app now, but nobody wants to maintain one.”
The only slop worth using is slop you make for yourself.
I think I agree. I have a few self projects I’ve vibe coded and used a few months that I’ve considered putting out there open source and if folks wanna try cool if not whatever…but that’s probably how all of those dead projects start. I’m even hosting them on my own private Gitea and updating via Obtanium because I don’t want to bother anyone lmao. Wait crap Google is gonna stop me from doing this in a couple months huh
That is so accurate. I have a couple of LLM-coded applications running, either because a solution wasn’t available, or existing solutions were beyond the scope of what I need, and would idle at up to 1 GB of RAM instead of 10 MB. In situations like these, being able to get a quick solution thrown together is such a boon.
Can confirm, am running a lot of services on a VPS that make my life more convenient. I prompt from my phone and get useful stuff out. I am a software engineer, so I can course correct ir when necessary. I only use opencode for this, not Hermes or Openclaw
Vibe coding a simple project is easy, but a crapshoot, at the current state of LLM development. Vibe maintaining anything at all is basically impossibly currently, you need a competent developer for that.
I agree that people cannot vibe code well unless they are a developer. Knowing the difference between slop and using a tool to automate bulk writing code is crucial at the current state. And I don’t ever know if it’ll get better because you need to know why you want to build something someway.
And I don’t ever know if it’ll get better because you need to know why you want to build something someway.
The major issue I’m seeing with junior (and even intermediate) developers is that they trust that the AI will always do things the correct way and don’t question its approach, and they don’t develop proper debugging skills and just rely on the AI to attempt it.
To get decent quality output out of an AI model, you need to have critical thinking skills, at least basic knowledge of the overall architecture for whatever you’re trying to build, and enough knowledge to question the model when it does something wrong.
Blindly trusting AI is why so many old security issues are coming back - stored/reflected XSS, SQL injection, exposing databases directly to the internet with no password, things like that. Newer frameworks mostly got rid of them, and now AI is bringing them back. It’s a fun time for red teams at least.
As someone somewhere between junior and intermediate developer, I will say vibe coding on my personal projects has accelerated my learning so much on the proper way to code things so much more than constantly bugging my seniors who don’t have time to properly review and critique my code (because surprise surprise! We’re understaffed). At least now I can ask Claude to explain its approach and fact check it myself, and the times I’ve had it run too loose, I’ve gotten practice debugging code I’m unfamiliar with since Claude eventually hits a point where it fucked something up and has no idea how to fix it. But obviously the caveat is you have to approach it as a learning tool not an automation tool.
I’m a developer since 20 years and been trying out vibing godot and I expected it to have troubles with Godot but at least getting basic programming paradigms right but it has been more the other way around. I’m constantly policing it for hardcoding or creating unmaintainable messes where the base classes have exceptions for each child instead of them owning their own logic.
I’m in some deep shit because my manager put colleague in charge of a project and he chose to slopcode a solution I don’t understand. Now our client wants to understand it and the code is absolute shit. His only solutions are “did you ask Claude”
Disagree. It’s perfectly viable. I’ve maintained several projects for over a year myself
Are you maintaining them for your own use, or do they have other users who you are supporting?
I have a few projects, some are just for me. And others are available for use. Mostly plugins for other projects, couple hundred people using them.
They all have good cocd pipelines with testing, code validation/ static code analysers. It’s trivial to maintain. Not big projects by any means but 20-50k loc
Can you name the one with the largest user base? How do you distribute it? What’s the ecosystem?
Are these projects… in the room with us right now?
I’m not the one throwing out baseless accusations
Just baseless claims.
I have nothing to prove
Then why would anyone believe what you said?
Why would they believe the other guy?
It’s perfectly viable.
That is you making a claim.
I’ve maintained several projects for over a year myself
That is you trying to prove said claim with anecdotal evidence. If you had nothing to prove, you’d have made the right choice and stfu.
Oh here we go
Sounds interesting.
Can you link those projects please?
Hardly an issue with generated code. You could say the same about projects before LLMs were widely used for code generation: “most projects are abandoned within months of release”. The difference now is the scale and how some people feel about it.
Most hand-coded projects are “abandoned” within months of release, but this often happens because the project is done, not because the project outgrew the capabilities of the LLM and the vibecoder is too incompetent to fix the accrued technical debt themselves.
Nah, most projects where just abandoned. Most projects never reached maturity and where abandoed in like weeks if not days. Look at the amount of 0 to 10 star github repos that are open source and had no commits since ever. LLM’s just removed the hurdle to start something “simple” so now the problem just grew in size.
Look at the amount of 0 to 10 star github repos that are open source and had no commits since ever.
Plenty, possibly most, of those do exactly what they were built to do and nothing more or less. I have a couple of dozen such repos myself.
Hmm, I think we are running both on assumptions. This would need some data to be analyzed to actually have a clear picture.
Disagree. Most projects are never “done”. Whether that is defined by the user or by the users, there are nearly always things to work on. Maintainers just lose interest or don’t bother. This is as old as open source and it has nothing to do with LLMs, it’s only aggravated by it.
There is a crucial difference though, vibe coding enables people who are neither interested nor capable of understanding what is involved in maintaining a piece of software. Before vibe coding it didn’t even come to this situation: if you can’t write software you don’t need to maintain it. While there were bad coders and bad maintainers before, the numbers have now increased dramatically because everyone can pretend to be a software dev these days.
I’m generally anti-ai but this is kind of exactly what vibe coding is for.
Someone has a problem right now and there is no tool to fix their problem how they want it fixed, so they throw some shit together for personal use and maybe someone else can use it if they want idgaf.
Why would they maintain that? It does what they need it to, when they need it. Usually these tools are very basic.Usually these tools are very basic.
If you read this community in the last weeks you’d know that slopcoders don’t stop at basic projects.
I think that compresses the nuance too much. I think of vibe coding like this:
Back in the day, the C64 bedroom coder wasn’t trying to “disrupt” anything. They were making something that worked for them, modifying example code (usually…poorly) and then sharing it. That’s how the demo scene started. I was there Gandalf, 3000 years ago.
The pride was in the craft, not the pitch deck.
The difference between that and the “we shipped” crowd isn’t that the C64 coders don’t want recognition - it’s that they weren’t substituting the performance of shipping for the thing itself.
Or using a woodworking analogy; you want people to appreciate the dovetail joint, not the Instagram reel of you cutting it.
If you made the dovetail joint, it’s obvious. Does which tool you used particularly matter? Shouldn’t I independently assess the quality of the joint regardless?
It’s very trendy right now to be anti-ai everything, and I get that, but we’re at risk of missing the forest for the trees here.
AI is a force magnifier and it’s not going anywhere. Take the same due diligence with projects you see here you do with other areas of your life - if it sounds too good to be true etc etc.
I hate being the wowser police on this, but from where I sit both the “FuckAI” and “we just shipped… curious to hear” crowds have the same issue.
I’m in the “Fuck AI” crowd even though I currently heavily use LLMs, specifically for the reason that slopcoders, techbros and memelords destroy the planet.
I’m in the same boat as those people and that’s specifically why I hate it.
Then you’re not actually against AI, you’re against hustle culture, influencers and unsustainable practices.
I’m 300% against that myself.
What I don’t like is when nuance gets compressed for sake of point scoring, tribalism or as an end run for due diligence. Not saying you did that BTW.
I don’t disagree necessarily, but they shouldn’t be uploading / sharing one-off projects they don’t intent to maintain.
That is a bit of an open source philosophy difference.
Is it better that everyone has open source everything so that anyone who finds the one-off useful can benefit from it?
Or is the software actually not provided “as-is” like the license states and on releasing open source software the community deserves regular updates?
I think that second option is a very entitled path. We are not entitled to the continued used free labor of a random person on the internet.
Yes, but there’s a difference between having your source and binaries available on github and submitting to an app store like flathub.
I don’t really agree. If you have a tool to do a thing, share it in case someone else could benefit, just be open about what it is.
If someone isn’t motivated enough to actually think and create something, what makes you think they’ll be motivated enough to maintain that thing they didn’t create?
I would say that the free tiers of those coding tools got cut back so much, that some of those slop coded stuff wouldn’t keep doing it any more.
Claude is on a 5 hr timer, and you get very few responses. Qeen Code was free for a second, that was nice.
Speaking from my own slop coded project viewpoint.Yeah, but OTOH, do we cheer for centralisation of technology for thems that can affords to pay?
Locally run models are available for free. My laptop can run a tiny model 1.8b, so it’s useless, and the family gaming pc could use some smaller side medium ones like 8-10b ones.
Right now the bubble is still insane, but I believe the it population will able to afford better chips in couple of years.Maybe but right now, according to the Steam hardware survey, the most commonly owned VRAM tier is 8GB - sitting at around 27% of surveyed systems, with 16GB closing the gap fast.
Even with MoE and llama.cpp tricks, you’re not running frontier anything at decent context length without significant fiddling on that - it’s possible on 8GB but you’re operating with almost no headroom, 16GB might let you scrape by with a Q4 quant MoE.
The very best local coding models (arguably the Qwen3-Coder-30B-A3B MoE or Qwen 3.6-27B) need 16-17GB at Q4 quantisation, and building a system from scratch to run one is probably a $3.5K proposition. With the cost of living crisis hitting everywhere, that means the table ante gear is beyond the reach of many. Even a decent GPU is north of $1K in many local markets.
I adore small LLMs, and know a lot of tricks to leverage them, but 14B is the bare minimum for what I would start to consider competent.
We’re boned until 2030ish, when gear gets cheaper (allegedly).
Part of me thinks there’s a dark conspiracy at play here. Give people affordable access to frontier LLMs, make self hosting hardware cost prohibitive, then jack up subscription prices.
I think there’s a way out of that mess, but it needs people to stop chasing “bigger, better” and start chasing “actually, how can I use what I have to do X instead of needing bigger and better?” but that needs talented devs and a mind shift.
ICBW and YMMV.
2030 is only a couple years away ;).
IMHO, $4-5k for a computer that can code as much you want is a bargain. If you have a product that will sell.
Conspiracy is right, the bubble is real and they will keep it pumping til it burst. Llms are not the second coming. They are great tools, but they are not solution for everything.
This economical milestone will be really “interesting” in the next 10 years. Interesting as the Chinese curse for wishing you interesting life.That assumes prices actually come down.
On purely dollar costs, $5K for a coding / AI only PC doesn’t amortise vs subscription or API access. Not for occasional shenanigans anyways.
I’ve done the calcs, because I wanted to justify the indulgence and the numbers dont stack up. Yet.
Of course, I based that on current rates. If basic subs go to $50/month, exclude coding agents, and API skyrockets…well fuck, I’ll go back to modding and retro gaming. I still remember how to code in ARexx, BlitzBasic etc and I keep threatening to make an Amiga or C64 stuff…there are other rabbit holes :)
PS: Lemmy spying on me; YouTube just recommended this
I expect this gap will be filled by cheaper models like GLM as they catch up to frontier models in their capabilities.
I believe they will catch up in no time.
But not every one will be able to pay anything for them.
I’m not paying because it is a hobby project (a huge one) and it is totally free with optional donations. That’s not much ROI.
Even if they weren’t abandoned, you shouldn’t use them.
Well, yeah. There are lots of reasons. But basic self-interest is something we can all agree on.
Easy come, easy go
On the flip side, if you’re vibe coding an app you should seriously consider whether it’s something you want to open source or make available publicly. There’s a social contract that comes with that.
I have 2 self hosted slop apps I build and maintain myself. I think people would genuinely get great use out of them.
…but then I’m inviting critiques and feature requests and am roped into supporting them so it’s not just a big pile of shit that wastes everyone’s time. And I don’t want to spend my limited free time making common sense improvements to improve it for others. I want to write a lazy Claude prompt with insufficient context, get it barely doing what I need, and then spend the rest of my time eating crayons and similar pastimes.
Just to nit pick (though maybe not) is there a social contract? That usually implies in exchange for X, you get Y.
If the thing is provided as FOSS, what does the dev get, contract wise?
Isn’t that how we end up with devs walking away entirely due to “I downloaded your project, you owe me xyz, you fuck”? I’ve seen that happen more than once and it’s a real factor in projects being abandoned, even before slopcode.
Speaking for myself only: when I share something, it’s usually something I made for myself that I think others might enjoy or find useful.
As the dev, I’m happy to look at suggestions or reports, with no guarantee that your idea will be implemented. If it is, I credit it and you.
I also refuse PRs, because if I am developing for me and sharing, then I’m not developing a product for sale to spec or running a democracy. I don’t know you, you don’t know me and you likely don’t know what the long term road map or invariant constraints are, so I’d rather just not. I realise that’s not a commonly held position but it’s in the same “limited time” category.
I’m happy for you to fork it, ask questions and spin up your own tho - that’s why I like AGPL-3.
Between all that, I’ve been able to avoid the excesses of both sides but YMMV.
People can code publicly for themselves for sure but there is clearly a subset of people who feel like their slop is worthy of community attention and resources in the same way as traditional projects that have been proven over time. It’s the same as the Ai art slop paddlers thinking their work is as valuable when they don’t even know what’s it in. To be frank, largely parasitic people using largely parasitic tools. A lot of that community doesn’t hold those values you eloquently stated.
I agree with that…but I wonder if we only ever hear from the so called parasitic ones?
Suppose a kid in Africa uses Claude to code an app that tracks the spread of a disease in their community and predict the next outbreak site based on x,y,z.
That’s technically slop code too. Do they get a pass because the cause is virtuous or not crowed about? By the letter of the law…no. But by the spirit of the law, probably yes.
I guess the difference is, how much leeway do we have for genuine enthusiasm vs parasitism. It’s hard to tell sometimes on social media where too many people are doing preening displays in public - and we’ve probably all been guilty of that.
As a rule, if a thing interests me, I’ll read the post, hit the repo, and dig around the files. If I see obvious use of llm in the code (like the stupidly verbose comments that LLMs like to pepper throughout), that usually means that the person either didn’t look, didn’t know to look or doesn’t care. That’s bad.
Bad intro post + bad readme.md + weird commit history + weird AI comments = I’m out.
Honestly, I’m usually out after the first one or two these days.
Yeah that’s a good point, I guess our current social app structure isn’t designed at all to filter signal and noise. The art world has galleries which kind of do that at the expense of barrier to entry. Maybe things like Flathub and repos end up seen that way.
On the flip side, if you’re vibe coding an app you should seriously consider whether it’s something you want to open source or make available publicly. There’s a social contract that comes with that.
This is the attitude of an existing programmer who is using these tools. What I’ve found on here is a specific pattern that keeps repeating:
- A post is sharing/advertising a project. The poster is two hours old.
- The poster is extremely coy about how their project was made despite obvious slop in the post body. They’re a bit clueless about anything that looks like
a social contract that comes with that
and think there’s something strange or accusatory/interrogatory when people ask questions they think are too difficult or technical 3. Some Lemmy users have generally polite but fundamental critiques and questions the poster can’t answer or thinks must be gotchas 4. The poster has a crash out about us all being mean/unappeasable/anti-AI/luddites/Linux users/godless commies and is usually the only one downvoting comments, even ones that read like genuinely 5. The post and account are deleted
There’s a clear disconnect. You’re talking about the homelab community which is a bit different but I specifically remember someone making an accessible Android UI and being extremely frustrated at people asking for the entire code to be released, and at people saying there’s not enough features there for that poster to be looping in advertisements on a fucking home page UI.
I get the impression that primarily-slop coders on some level think they’re doing programming, because of how you can get functional prototypes of code that is way above what a total beginner can write on their own. They think having code that compiles (whatever it’s usually Python there’s no compiling) means the hard part is over. They don’t seem to understand that the questions and concerns about vibe coding aren’t moral complaints but genuine concerns about liability, running code even the author doesn’t understand, and a complete cluelessness about what they should be doing to evaluate the code besides prompting it to be “good with no mistakes”.
That Android UI project seemed like a little thing a few people could install on their grandparents’ phones. It’s normal for the author not to understand every little thing. But being totally clueless and being offended at the suggestion, being entitled to put ads in it to get 0.0016 USD per year per grandma in exchange for taking up a quarter of her screen forever, not understanding why this looks scummy, why refusing to release 60% of the code looks scummy, why half the questions are being asked at all.
Again this would not be a problem if this is now the expected norm for a significant portion of any projects you find online. A lot of projects are the first genuine effort of someone out there and they’re not perfect but they didn’t feel like the unceremonious implosion of the entire philosophical concept of personal computing.
And I’m fucking shit at writing good code and I’m pissed.
The pre-LLM-effort-was-a-filter argument holds up, but I think what effort was really filtering for was why someone built the thing. High effort filtered out “this seemed fun for a weekend” projects. LLMs just surfaced that those were always the majority.
The better filter is: does this project serve a specific audience that genuinely needs it, or is it a demo of what you can do with Claude?
What I look for now:
- Specific problem for a specific group of people (not “general-purpose LLM wrapper”)
- Open core (MIT or something that lets the community carry it if the author walks)
- Revenue model or institutional backing (someone has to keep the lights on)
- Evidence the author understands what they shipped, not just LLM output committed wholesale
We build CircuitForge, self-hosted tools for navigating opaque systems (job markets, government benefits, insurance). The architecture is deterministic-first: eligibility checks, validation, and data pipelines are rule-based and grounded in structured data, so the LLM is drafting from a clean, repeatable foundation rather than hallucinating into a void. That also means we can run smaller, specifically fine-tuned models instead of throwing a frontier model at everything and hoping for the best. Smaller models run on consumer hardware, which cuts hosting cost and shrinks the privacy risk surface significantly. Humans approve before anything acts. Pipeline layer is MIT and lives on Forgejo. There’s a full devops stack, a real business model, and I use these tools every day. We’re also actively collaborating with other devs and always looking for contributors.
The people using these tools actually need them. That’s the commitment signal that doesn’t evaporate when the novelty wears off.
Replying to this post with an AI slop comment takes some balls, mate
Ha, the irony isn’t lost on me. But the comment was mine, not generated. The project does use LLMs as a tool (that’s the point of it), but “uses LLMs” and “is slop” aren’t the same thing. The repo is public if you want to check the commit history and structure rather than take my word for it.
Dammit…did you have to pivot this into a pitch?
This reads a little like a LinkedIn post.
Fair point, I’ll own it. Dropping a project link in a thread about slopcode is going to read as a pitch, and I needed to promote or nothing will ever happen.
That said, I figured showing a concrete example was more useful than just adding another opinion to the pile. If you want to pull the repo (https://git.opensourcesolarpunk.com/Circuit-Forge/peregrine), the structure is there to review. There’s a deterministic pipeline underneath the LLM layer (eligibility checks, form validation, deadline tracking all run without LLM involvement), CI with test coverage, and a fine-tuned model approach that keeps inference on local hardware where possible.
The hard part wasn’t the LLM layer, it was the plumbing around it that keeps the LLM in an advisory role instead of a decision-making one. That design constraint is what I wanted to show, not just “look, I made a thing.”
Also, fair warning: my day job involves writing docs and reports for corporate clients, so the LinkedIn voice leaks in whether I want it to or not. Working on it.
As always the social infrastructure (community, user base) is much more important for the success and use value of a piece of software than the exact technology used to make it (lang, framework, helper tools).
We went from “everyone will know how to code in the future” to “no one know will know how to code” pretty quickly
Just don’t install anything new after 2022. Easy way to get actual code audited by human beings.
