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].
They’re all slopcode apps now. Most companies made it mandatory.
My coworker said it best: “Anyone can build an app now, but nobody wants to maintain one.”
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 made the mistake of installing Starling recently, not realizing how it was made. I contributed a PR to it, write a few issues describing the bugs, and since then there’s been absolutely no activity from the creator.
That’s fine, they are under no obligation to work for free. But I wouldn’t have installed it if I knew it was abandonware.
Are there any obvious signs of it being vibe coded that I’m missing?
In my experience, projects not being very active, especially small ones by a single person, isn’t anything new that has much to do with LLMs, it was always that way for hobby projects. And it was inactive for only about a month now, with the author replying within one day before that. I have a few hobby projects myself and don’t reserve time every month to work on them or check on their repos.
If you have access to their git, watch out for the commit history. Check how much code each commit is introducing and/or editing and how fast they are, and how many times these big commits happen.
If you see a project with big commits in very short interval of times, it’s a sign that it was vibe coded
No obvious signs, nope. It wasn’t until I started using it in earnest that I got suspicious and then when trying to work on the code it became very clear.
This in my experience is the hallmark of a vibe coded project. Individual pieces of code can be perfectly fine, but when you zoom out into an overall structure things get weird. Design patterns changing, same/similar problems solved in different, sometimes conflicting ways.
Been working on a project like that at work. Initially I enjoyed the change of pace, but as I realised that there’s no coherence at all in the project structure my joy turned into frustration.
To me, the most frustrating thing is that you can’t ask someone why something is done in a particular way, because no decision was ever made because no one was there to make a decision. Things just happened.
Seems like your PR was merged by the creator? If you are mad that the creator isn’t doing anything with THEIR OWN project, fork it and fix it yourself… No need to get pissy and make a full complaint post in here about it.
No one got “mad” or “pissy” or filed a complaint. I feel they they were very explicit that it was not a complaint.
It doesn’t sound like OP is mad at the creator, but annoyed at the increased number of projects that are abandoned
It’s hard to judge the longevity of a project, and LLMs can make a project look more stable and professional than it is. It’s fair to feel annoyed if you mess up that call and move your workflow to something that gets abandoned.
Disagreed. You should blast and air everyone’s embarassing trash as loud as possible so no one misses it.
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.
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.
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.
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.
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.
I can only delete windows so many times … sadly my work won’t let me delete the install on my work laptop
I’m not surprised. If you didn’t have to will to properly build it yourself, you won’t’ have the will to properly maintain it.
There was a Ted Talk a while back (I can’t remember who) where they said “I have always wanted to give a Ted Talk… but when I got selected to do this I realized that what I really wanted was to say I had given a Ted Talk”. Meaning they want to be known as someone who had given a Ted Talk, not actually go through the process of writing and delivering the Talk.
People who write open source code do it because they like the process of writing, just like an author enjoys writing books. LLMs are for people who just want to be able to say they have written a book. People who slop-code aren’t actually interested in learning how to code. Which is a fine toy for them to play with, but not sustainable (or reliable for others to use).
At some point the project also just gets hard to maintain with AI, you burn ever increasing tokens the bigger a project gets and AI produces awful code for the purpose of maintenance by hand. Once the bugs start rolling in they realise the AI can’t fix them, they certainly can’t afford to have it constantly load the entire project as context and they don’t know how to fix it themselves and the tech debt produced is immense. So the projects get shut down.
We went from “everyone will know how to code in the future” to “no one know will know how to code” pretty quickly
This isn’t the burn you think it is. Should we start listing actual high use hand coded abandoned foss projects?
I make all my own slop apps now. Bespoke crappy solutions for bespoke crappy problems. Abandmont rates are up, I can attest
I’ve found LLM’s to be pretty decent at writing one-off scripts for boring tasks. “Baby wipe scripts.” Use them to clean up the shit and throw them away.
I’ve rebuilt a few SaaS projects that I use so it’s under my control. Might not have all the bells and whistles, but it aligns with my needs better.
I rebuilt a simple Playstore task app into a multiuser fleet maintenance app for the farm. I’m not putting it in the wild, I don’t need that headache. I build it exactly for our needs and I don’t need to have to deal with users I can’t tell to fuck off to their faces when they get snotty about a bug.
But overall, this kneejerk “everything AI makes is broken” bullshit is starting to get to me. It’s pretty obvious these people either haven’t used it in the last year or so, or don’t know how to. Or they’re just performing for the internet, and actually use it all the time. I tend to think the latter.
Another person already wrote this itt, but the cost in time and money of building something that “works” is very low. That cost increases every time you need to fix something. It’s why I’ve switched to deepseek flash instead of the bigger, better coding models. For the same price as a million tokens of Claude for example I can get a hundred million tokens on flash. It takes three times as long and needs a little prodding but it’s thirty times cheaper.
Even if you’re not dealing with ballooning costs when it comes to upkeep, the present environment around ai programming is filled with perverse incentives that reduce the chance something is open sourced. Why preserve the old work and try to carry it into the future when you can just rewrite it? Why create standardized libraries and approaches to specific problems when it’s only gonna be used once? Why make it open when you don’t have time to deal with requests and bug reports you don’t care about? Why use gpl when you don’t intend to make it public anyway?
Those were all arguments against gpl and open sourcing parts of the Unix codebase too, btw.
Harnesses or agents try to address some of this by sharing improvements but that’s three layers removed from directly turning fossil fuels into cpu cycles to make the same program a thousand times slightly different for a thousand different people.
I tend to see the ai programming defense in the same way you described yourself and the “ai is bullshit” crowd. People defending this stuff generally either aren’t looking where they’re walking or haven’t had to use it long.
As a decently prolific user of ai programming, it’s turned computer code into disposable plastic wrappers. Even if they don’t persist and make a giant garbage raft in the ocean or calcify our pineal glands we still wasted a bunch of energy on them.
I don’t see that it increases the costs. I’ve written software and PMd other software builds in a previous life. Software has bugs and either I fix it slowly because I have to wait for winter to have time, or the AI does it while I drive a tractor for a couple bucks in tokens. AI can architect fine now and if you use your head, you can cross-check the codebase to ease maintenance with other LLMs, then come in and give it a final QA yourself.
I have a bugtracker I build into any software I build, an agent watches for new entries by users and does some preliminary work before bringing it to my attention and then every day I get a report that I can say what gets worked on while I have coffee. Honestly, I don’t care if the agent decides to refactor the whole mary-anne, it pings me perioidically in Matrix and I steer it around like I do my hired men asking which field to cultivate next.
At this point I have almost entirely vibe-coded software that’s been dogfooded for almost 2 years that works great and massively improves our operation. And I have a dozen other ideas ready for me to give some attention to in order to get them off the ground. It’s an exciting time. I love it.
Yeah same. I make loads of little things to bodge my own problems but would never be moronic enough to try and capitalise on something that took 60mins to make.
Exactly. Oh I have this small problem, or solution I want to solve. Great I can put something together quickly and move on with life. I am not selling it, and it’s not even needed for ‘open source’ because its often incredibly niche problems that I am trying to solve for, that I wouldn’t be able to do elsewise.
I like Cory Doctorow’s take: AI is good for single-use, personal code to solve an immediate problem, and terrible for long-term, production projects. I imagine there’s a bunch of neophytes out there who use AI to create their first project, find out that github exists, and thinks someone else must be having the same problem they just solved, so why not release it to the public?
It’s late and so maybe I missed it, but I didn’t see the part of the article that compares the abandon rate of slopcode with the overall abandon rate. Not saying that the premise is wrong or anything, but you can’t tell how bad something is unless you can compare it with the norm.
I would guess that the key difference is that vibe coded apps can get to a more or less working state a lot quicker, while other apps are likely to be abandoned before it’s done.
Though in either case I’d always be careful with new projects. If it’s just a single guy that’s been working on something for less than a year and only have a handful of GitHub stars, I probably wouldn’t install it.
Also it’s again the false sense of security pf “if you don’t use vibed apps you’ll be fine”, making people forget basic security procedures.
I, for instance, had a service vulnerable and discontinued without noticing for months. It was something 100% made before LLM was a thing. Still had unpatched vulnerabilities and the project was abandoned. It was my fault for not checking more often is the services I host are safe or not.
So what? The important datapoint is the number of new successful projects, not the number of the abandoned ones.
You meant “the number of new projects that will soon be abandoned”.
Vaporware is never not a thing.
I thought vaporware was announced or promised code that never materializes, or shows up much later than claimed. Often used by big companies to squash competition, who typically have very real solutions available.
That’s one example, yes. My point is we like to invent new ways to call out the same shit. Vibe coded projects that get abandoned within half a year are just another example of vaporware, just a new take on the concept.
This is the main reason why vibe coding, even if it produces good code, is still a major problem. It encourages people with the goal of making software, but without the actual will and motivation to keep supporting that software to pump out software and publish it.
It’s like all the faceless AI-automated YouTube channels we have now. It’s not that these people had no way of doing it before, it’s just that it’s easy and might make them some money, or make them feel like they accomplished something until they get bored and move on.
There’s something to be said for convenient and easy to use things, but they’re a double edged sword, because they also directly target people with the least emotional investment to use them, as a side effect of that convenience.
but without the actual will and motivation to keep supporting that software to pump out software and publish it.
TBH, most software will never be used my many so needs no support. Also, I think lack of long-term support is not the same kind of problem. It used to be any more. Back in the day when the original author dropped support it was a major investment to get someone else up to speed. Now fixes and enhancements can be done by LLMs as well, given a somehwat competent software developer.
But in general: The newer the project and the more bells and whistles it has, the less I personally would want to make it an essential piece of my workflow.
Isn’t that no different than the millions of open source projects that have few authors, little interest and are abandoned for the next shiny thing? At least in my mind with the current state of LLMs, if there is an open source project that you want to update for yourself, you should be able to do that pretty easily.
Isn’t that no different than the millions of open source projects that have few authors, little interest and are abandoned for the next shiny thing?
The end result isn’t inherently different, but the basis is.
Someone who programs a FOSS project with their own programming skill is, by definition, more invested, and more willing to spend time making that project exist. Someone who is mostly just willing to tell the LLM to do it for them is, by definition, less invested in the project.
It is more likely that someone less invested in a project will abandon it.
At least in my mind with the current state of LLMs, if there is an open source project that you want to update for yourself, you should be able to do that pretty easily.
Sure, that’s possible, but the main concern is that most people either don’t have access to more capable models (which will otherwise require them spending money for their software to keep getting things like security patches, vs projects that have maintainers that are more invested in them that aren’t abandoned in the first place which are effectively free unless you choose to donate), and you get forks in development path. It might fit one user’s goals, but now you have, say, 100 people who all tried reviving abandonware with their own LLMs, all with different features, security, import/export mechanisms, etc.
That’s not to say it’s bad that people can use LLMs to “revive” abandonware, but I don’t think we should be encouraging people to create and publish software that’s highly likely to be abandoned in the first place.
