Open Source became annoying for maintainers, but ironically agentic coding made it awesome for end users. For non-programmers, the benefits of open source used to be mostly indirect and/or theoretical, because they couldn't actually modify the software themselves (learning to program is a major investment of time and effort, and a big barrier when someone just wants to avoid a bug or tweak a feature). But now users can say "Claude, fix it" and it tends to work well enough for a single user.
This is actually exactly why the GPL was/is viral: it recognizes that most end users can't do the modifications themselves, but they still should be able to benefit like either by contracting someone else to do it, or choosing from a marketplace of other public modified versions what already have what they want. In theory, the viral part meant you didn't have to do it yourself while still being able to benefit.
(in practice, of course, having the source code is often one of the smallest parts in modifying things anyway...)
And for programmers who are willing to use it carefully. My debugging time has gone to near-zero. "Here's some log messages, and then a gdb backtrace of a crash. find out why"
10 minutes later, I have a code path for the error, and a fix. About 20% of the time the fix is garbage, but knowing what to fix makes it easy to know how to fix it.
My debugging time has gone to near-zero. "Here's some log messages, and then a gdb backtrace of a crash. find out why"
Over in embedded land, setting it up so that the harness has the ability to flash and debug over JTAG/SWD (gdb interface to the entire microcontroller) is absolutely wild, especially for hard-to-reproduce bugs. I can set up a monitoring loop at the end of the day, have the bug I'm hunting for trigger at 2AM, and have a report waiting for me when I wake up in the morning that includes 7 different debug steps that were automatically flashed and tested.
Yes, it is mind boggling. A while back, there was a heisenbug in a distributed lighting controller I designed. I connected two boards to SWD, connected the serial bus to my scope, told Claude how to get to the scope over TCP, and let it loose. After a few hours it handed me a reproducible case and I was able to fix my hardware mistake that was glitching the serial port.
My general experience is that agents are so much better at everything else around software development other than writing code and I don't understand why everyone focuses so much on it.
Debugging complex issues? Great.
Analyzing infra issues from logs? Huge timesaver.
Quick prototyping and throwaway scripts? Awesome.
Researching and summarizing stuff? Quite useful, although a bit unreliable.
Writing code? Mostly garbage code that requires a rewrite to be maintainable.
This is exactly my experience and it's great that I can save my energy to be able to work on the interesting parts of my projects, rather than burn out spending hours or days debugging tricky bugs.
I recently got an iPod classic and none of the Linux apps that should have worked for uploading music to it did. So what do I do in this situation? I saw that libgpod exists, so now I have to spend hours learning and writing a program, or give up and install rockbox, or just plain give up?
No, I asked AI to make me a GTK 4 app, and it makes one in ten minutes.
I don’t use AI for personal projects because it sucks the fun out of doing them, but for troubleshooting problems on my computer that I don’t or couldn’t care to learn about, they are a revelation.
I noticed there is quite a split in the programming community. Some people (including myself) are happy to let LLM write code for them, because it's so much more productive. On the other hand, there are many people believing that using LLMs is bad, for various reasons.
But in this particular case, thinking that only devs should be able/allowed to use open source - isn't it a bit too selfish? Open source has always been for the public's benefits, and now these benefits have clearly expanded to a lot more people. I don't see anything depressive here.
I think you're missing the sarcasm in the post you're replying to.
Do you think someone not programming-adjacent could have vibe-coded the application? I doubt it. And now it's effectively closed off from the world where no one can or will use it.
Not only could they have, they regularly do. I see demos every month from non-technical people who have done stuff they hoped for years some coder would do but no one ever had
I'm sure it happens but every time I ask for examples I get crickets. Or something where "non-technical" still means "knowledge of HTTP, databases, and so on."
"We are making a GTK backed ipod classic music uploader. What this tool should do is upload every song in every subfolder in the ~/Music directory to the currently connected ipod. It should use https://github.com/fadingred/libgpod, which is installed currently on this computer"
I think you could argue that this is too technical, but it was mostly to get stuff working with minimal fuss. I think that if you omitted "GTK backed" and the specification of which library to use it would still work fine.
It doesn't seem to me there was much sarcasm there, flockofbirbs seems to be genuinely believing that non-programmers should stay away from open source or LLMs.
And yes, as others point out, non-programmers definitely do vibecode stuff. I know one guy personally who vibecoded the whole online store for himself, and he's very far from being a programmer afaik. Now, I don't know about the quality of the implementation or any of that, but the point remains.
Yes, I definitely genuinely believe that gatekeeping open source from non-programmers is good. I believe in people having agency over their computers, but only if they are programmers.
Because non-programmers didn't add "... and make it secure" to their prompt?
Seriously, I'm concerned about that. Maybe not for stuff that runs locally, but for anything network facing. How secure is your acquaintance's store? I recall one tea and spice store I used to frequent [1] would show the site if you had bookmarked it, but if you came from Google or Google Maps, it would show a site selling knock-off Viagra pills, and this was from a decade ago. The modern Internet is a rough place.
But do I believe that only programmers should have access to open source? No. That's how I learned to program. Well, open source and having to debug code.
[1] The store is no longer in business and I don't think it had anything to do with the web site. It was brick-and-mortar store in a small town and it probably failed mainly due to lack of business.
We're on a programming forum, so it's a given that people here care about the craft itself.
But there are people who don't want to program. To them software is just a hassle standing between them and their goal.
That's normal. Not everyone wants to cook fine cuisine, many people get fast food. Not everyone wants to make custom furniture from the finest oak, plenty are happy with IKEA, etc.
And now software is moving beyond artisanal passionately crafted projects, and machine-extruded code nuggets are on the menu.
Craft vs. mass is a false equivalence. LLMs are not the IKEA of software as their products are created with precision. Writing code without LLMs is not using hand tools instead of power tools. I use efficient and precise power tools when I'm woodworking and when I'm programming. I cut rabbet joints on my table saw instead of cutting dovetails with a hand saw. I use Emacs instead of Notepad. LLMs are something else entirely because they are inherently probabilistic black boxes. They resemble slot machines a lot more than CNC machines.
I was talking about machines, not humans. But a human can give an explanation for why they did something. An LLM cannot as they're all pattern recognition, no reasoning.
You know, we can ask LLM to write that code, and in the meantime can spend a few mins responding to a comment on Lobsters, and after that get back to the LLM and review its output. And then likely point a few issues, and let it fix them. And then also ask another LLM to review the first LLM's output too, because it often finds issues that humans have missed.
In my experience, it results in higher quality software in significantly less time.
Can you explain this emotion logically so we can all understand it, despite the fact that it is absolutely obvious what your stance is? Make sure to not include anything difficult that necessitates us extending our thinking beyond what is in directly in front of our eyes or to do with our own personal desires.
I don't know man, feels like you're not allowed to feel emotions about these things. "This new power plant delivers 2x the power at 20x the carbon emissions, it's awesome!". It's clear as day why someone is opposed. The feigned shock, as though half the populace doesn't feel this way, is grating.
I'm not the parent poster, but I'll give my stab at that question.
It's because to some people (myself included, and I assume a lot of people on this website too) the process of discovery, of learning, of getting stuck, of despairing, of ahas, and of figuring shit out is the most amazing fun imaginable. After a hug from my loved ones, I scarcely know of a better feeling than having solved a hard problem (meaning a problem that was hard for me)! That's why not!
In addition to being fun, we also find it useful (to anyone, also those who don't find it fun) as a way of exercising one's brain. As a researcher, I believe it's also something that inherently drives people who seek out that profession.
Of course, everyone gets to pick their own fun. Lots of people seem to like singing and dancing, for instance. I don't. But I understand enough about fun and inherent drive not to show up at a song retreat and ask "ugh, why do you guys even do this, skilled audio technicians can synthesize a much better voice than that!"
It's because to some people (myself included, and I assume a lot of people on this website too) the process of discovery, of learning, of getting stuck, of despairing, of ahas, and of figuring shit out is the most amazing fun imaginable. After a hug from my loved ones, I scarcely know of a better feeling than having solved a hard problem (meaning a problem that was hard for me)! That's why not!
Okay, and like I said, I do too. The problem is that I don't like doing that for everything in my life. Some problems are not actually hard, they are simply tedious (to me!).
Honestly the implication that I don't is kind of insulting!
Of course, everyone gets to pick their own fun. Lots of people seem to like singing and dancing, for instance. I don't. But I understand enough about fun and inherent drive not to show up at a song retreat and ask "ugh, why do you guys even do this, skilled audio technicians can synthesize a much better voice than that!"
That is simply not what I'm doing at all. I made an app, for my self, that saved me time to get back to solving the difficult problems that I enjoy solving. That doesn't meant that someone who enjoys working in that domain would not have a fun time working on that problem, it just means I don't want to wait around for them to get my iPod working
Okay, and like I said, I do too. The problem is that I don't like doing that for everything in my life. Some problems are not actually hard, they are simply tedious (to me!).
Of course. But lots of people who frequent this site do consider problems like
I saw that libgpod exists, so now I have to spend hours learning and writing a program
to be genuine fun! Again, nobody is saying that you have to share in our idea of fun. But you did ask why people would find LLMs "not awesome" (which is what you asked "why" to). And you got answers. Now you're saying that those answers don't apply to you. That's fine. But they do apply to us who gave them!
I'm not saying they're awesome. I'm agreeing with the grandparent that says they are:
awesome for end users
You can dislike LLMs because they suck the fun out of programming, but some people don't like programming so it doesn't make sense to dislike how other people use them to avoid that!
Even the most LLM-happy programmer I know is clear that you need to verify the LLM's work. Does the end user who doesn't like programming suffer through that verification? Or do they just use the output unchecked, which I assume you'd agree is not "awesome"?
I don't have a say in how other people use their computers, but that doesn't disqualify me from making a judgement about whether I think what they're doing is a good idea or not.
You yourself say what they do is "awesome", so you're clearly allowed an opinion. Are only positive ones acceptable?
The real line is not whether they can check the output. The real line is whether they can both understand and accept the error rate.
And there are not so few things where (a) «works 90% of the time, gives weird garbage the remaining 10% of the time» is useful if it is cheap enough, (b) getting to 90% does not take too many «Please fix this problem» promtps (with manual retesting of previous issues, as by assumption forcing the LLM to build a reliable regression test suite is outside the user's qualification).
To borrow the example elsewhere from the conversation on this page — sometimes one needs to copy a playlist perfectly because this is the music for a carefully staged event; sometimes even a maliciously selected 90% out of a shuffled playlist is close enough for a morning run.
But every project that leaves [GitHub] takes its issues, its history, and its people somewhere new, with yet another account to make and another place to look. It fractures things even further, and makes the next stranger even less likely to wander in.
I personally have never really used the “discovery” features of GitHub. I’ve maybe found one or two interesting repos by quickly glancing at the profile feed in the past 10 years but most of the time I find things on external sites like this one, or because I’m already on one repo and a comment links me to another. I don’t see those mechanisms of discovery going away, it’s just a link to another domain. The aspect of having to create another account is mildly frustrating but, for me, wouldn’t be a convincing reason to abandon migration off GitHub.
Perhaps that's just where we're heading - smaller, trusted circles, with source available, but contributions by invitation. Maybe that's even fine, and I'm being nostalgic for a version of the Internet that was never going to last.
I love open source and believe in its ability to produce good software. Some classes of software must be open source! I also think that somewhere along the way this idea that all contributions must flow back to the original source became over emphasized, mostly implicitly. The whole point was to enable people to understand and modify the software they depend upon, but there was never a guarantee a maintainer would even look at a contribution. I don’t think the author has a sense of entitlement here, but some people really do and it’s always been an odd dynamic to me. Forking became a non-starter like rewrites. Sometimes those things are necessary and good!
We should foster a sense of collaboration, but ownership and the right to say no is just as important, sometimes more so.
Especially with password managers becoming more omnipresent over time, the cost of creating a new account has been becoming less of a repellent, to me at least. As long as I can avoid needing to use an email that might be leaked if the server get's breached, I don't really mind having to register one more credential in my password manager, that will autofill if I ever need to log in again. So distributing things accross numerous/smaller git forges really does not seem to be an issue anymore.
I personally have never really used the “discovery” features of GitHub.
I have used GitHub search over projects. But if things get pushed towards static Web 1.0 sites (cheap to serve from a small VPS, bots or no bots) with link collections, maybe search-indexing the repositories across the Web gets done again, as it has been done from scratch multiple times before.
I resonate with the deep melancholy of this post, but not as someone who’s been in the open source community for years and is seeing it change. I’m just a new developer who wants to do a good job, deeply yearns for human connection, and feels like a star looking across the cosmos—everything turning bluer and bluer redder and redder (my bad, lol) as it accelerates away.
Technically things areseem more accessible now but socially, nobody knows if you’re worth their time, so why would they bother? And that’s the point, isn’t it? Programming is both a technical and a social endeavor, and LLMs can talk like a human, take up digital space like a human, but they’re not human and only “care” about what they’re told to do, not who it affects.
I’m at least glad people are really nice at the coding meetups I’ve been going to.
I’m at least glad people are really nice at the coding meetups I’ve been going to.
I would strongly recommend continuing to foster in-person connections and friendships, both within and outside of tech. The online tech world is being shaken up pretty violently right now, but the IRL world is actually still filled with pretty great people.
I'm skeptical of this take, even though the examples are compelling. My skepticism is really just that people like working on software together and that it will become increasingly difficult to not be open source.
I've been looking in amazement with the stuff that's been happening in the games industry and retro-scene and I think this is a canary for the end of closed source. For those who don't know, people have been decompiling old games or recompiling old games to x86_64. And they are doing it at scale. For example, someone ported Halo and is running a 128-person server: https://www.techspot.com/news/114042-you-can-now-play-original-halo-browser-128.html
So I think we're entering a world where any closed source software can be cracked open with LLMs + ghidra and recompiled onto any platform. This used to be very labor intensive, and now that has all disappeared. I think this leads to two responses: 1) open source your software, if someone will just decompile it, copy and port in a weekend you might as well just open source it; 2) Closed source will move to the cloud and all close source software will live in some server farm.
I help manage an open-source project at work. While we definitely had some adjustments to make this year such as adding clear policies around LLM usage (we allowed them), I feel like we’re now reaching a happy equilibrium where the contributions, even if they are LLM generated are decent quality and welcome.
Our community also has regular online meetings which help us establish and make new connections. Maybe this is where open-source is heading? More value placed on genuine human interaction rather than simply creating pull requests.
We just need to be careful that our desire for genuine human interaction doesn't exclude disabled humans, humans who aren't proficient in the dominant language of the project/group, humans in non-overlapping time zones, etc. Asynchronous text is the most inclusive form of communication. That it includes LLMs too is perhaps a double-edged sword, but I'd rather err in that direction than exclude some humans.
AI slop PRs are the worst kind of work. They look like contributions, so you feel obligated to read them. They take real time to review, and then more time to explain why they're wrong to someone (or something) that isn't listening. That's time not spent on the actual project, or with family, or sleeping.
We posted a statement saying that we would ban people who submitted slop PRs. That seems to have dropped the number of submissions quite nicely.
You should have zero tolerance for nonsense. When some wastes your time, it doesn't matter much if they're being lazy, idiotic, or malicious. They're all DoS attacks. And they should not be tolerated.
I'm thinking of just accepting handmade PRs but if people are considering using an LLM just give me a good spec so I can generate it myself with my design sensibilities
I remember the old saying that to write a good spec you need to implement the system once (then throw it away and implement from the spec)… Maybe this might become a forced methodology.
kornel | 19 hours ago
Open Source became annoying for maintainers, but ironically agentic coding made it awesome for end users. For non-programmers, the benefits of open source used to be mostly indirect and/or theoretical, because they couldn't actually modify the software themselves (learning to program is a major investment of time and effort, and a big barrier when someone just wants to avoid a bug or tweak a feature). But now users can say "Claude, fix it" and it tends to work well enough for a single user.
adam_d_ruppe | 10 hours ago
This is actually exactly why the GPL was/is viral: it recognizes that most end users can't do the modifications themselves, but they still should be able to benefit like either by contracting someone else to do it, or choosing from a marketplace of other public modified versions what already have what they want. In theory, the viral part meant you didn't have to do it yourself while still being able to benefit.
(in practice, of course, having the source code is often one of the smallest parts in modifying things anyway...)
alandekok | 8 hours ago
And for programmers who are willing to use it carefully. My debugging time has gone to near-zero. "Here's some log messages, and then a gdb backtrace of a crash. find out why"
10 minutes later, I have a code path for the error, and a fix. About 20% of the time the fix is garbage, but knowing what to fix makes it easy to know how to fix it.
tonyarkles | 7 hours ago
Over in embedded land, setting it up so that the harness has the ability to flash and debug over JTAG/SWD (gdb interface to the entire microcontroller) is absolutely wild, especially for hard-to-reproduce bugs. I can set up a monitoring loop at the end of the day, have the bug I'm hunting for trigger at 2AM, and have a report waiting for me when I wake up in the morning that includes 7 different debug steps that were automatically flashed and tested.
wrs | an hour ago
Yes, it is mind boggling. A while back, there was a heisenbug in a distributed lighting controller I designed. I connected two boards to SWD, connected the serial bus to my scope, told Claude how to get to the scope over TCP, and let it loose. After a few hours it handed me a reproducible case and I was able to fix my hardware mistake that was glitching the serial port.
MatejKafka | 2 hours ago
My general experience is that agents are so much better at everything else around software development other than writing code and I don't understand why everyone focuses so much on it.
alandekok | 2 hours ago
LLMs have been trained on the Internet. And for anyone who's seen the Internet, it's full of crap.
lemon | 4 hours ago
This is exactly my experience and it's great that I can save my energy to be able to work on the interesting parts of my projects, rather than burn out spending hours or days debugging tricky bugs.
flockofbirbs | 9 hours ago
"Awesome" is not the word I'd use.
mplant | 8 hours ago
Why not? It genuinely is amazing.
I recently got an iPod classic and none of the Linux apps that should have worked for uploading music to it did. So what do I do in this situation? I saw that libgpod exists, so now I have to spend hours learning and writing a program, or give up and install rockbox, or just plain give up?
No, I asked AI to make me a GTK 4 app, and it makes one in ten minutes.
I don’t use AI for personal projects because it sucks the fun out of doing them, but for troubleshooting problems on my computer that I don’t or couldn’t care to learn about, they are a revelation.
flockofbirbs | 7 hours ago
This is quite a depressing point of view.
alanmeira | 7 hours ago
Why? Do you think only programmers should be able/allowed to use open source?
flockofbirbs | 7 hours ago
Yes this is clearly what I think. Thanks for saying it so I didn't have to.
dimonomid | 6 hours ago
I noticed there is quite a split in the programming community. Some people (including myself) are happy to let LLM write code for them, because it's so much more productive. On the other hand, there are many people believing that using LLMs is bad, for various reasons.
But in this particular case, thinking that only devs should be able/allowed to use open source - isn't it a bit too selfish? Open source has always been for the public's benefits, and now these benefits have clearly expanded to a lot more people. I don't see anything depressive here.
flockofbirbs | 5 hours ago
I think you've mistaken vast corporate exploitation of the commons as public benefit.
Relax | 6 hours ago
I think you're missing the sarcasm in the post you're replying to.
Do you think someone not programming-adjacent could have vibe-coded the application? I doubt it. And now it's effectively closed off from the world where no one can or will use it.
singpolyma | 5 hours ago
Not only could they have, they regularly do. I see demos every month from non-technical people who have done stuff they hoped for years some coder would do but no one ever had
Relax | 4 hours ago
I'm sure it happens but every time I ask for examples I get crickets. Or something where "non-technical" still means "knowledge of HTTP, databases, and so on."
daveliepmann | 3 hours ago
Here's an example: https://mattbruenig.com/2026/02/23/my-fully-automated-labor-law-research-tool-is-finally-here/
mplant | 5 hours ago
The prompt was a single sentence. They absolutely could have.
Do you want me to upload it? If I did, I have a hunch people would be on my case and call it slop
Relax | 5 hours ago
Sure, I'd be really interested to see the prompt and any back and forth you had if you're up for it.
mplant | 3 hours ago
My prompt was:
"We are making a GTK backed ipod classic music uploader. What this tool should do is upload every song in every subfolder in the ~/Music directory to the currently connected ipod. It should use https://github.com/fadingred/libgpod, which is installed currently on this computer"
I think you could argue that this is too technical, but it was mostly to get stuff working with minimal fuss. I think that if you omitted "GTK backed" and the specification of which library to use it would still work fine.
here's the repository
It worked fine the first try, the rest of the back-and-forth was me asking it to fix minor bugs and adding features.
dimonomid | 4 hours ago
It doesn't seem to me there was much sarcasm there, flockofbirbs seems to be genuinely believing that non-programmers should stay away from open source or LLMs.
And yes, as others point out, non-programmers definitely do vibecode stuff. I know one guy personally who vibecoded the whole online store for himself, and he's very far from being a programmer afaik. Now, I don't know about the quality of the implementation or any of that, but the point remains.
flockofbirbs | 4 hours ago
Yes, I definitely genuinely believe that gatekeeping open source from non-programmers is good. I believe in people having agency over their computers, but only if they are programmers.
dimonomid | 4 hours ago
I get that you believe so, you stated it multiple times in the comments here.
But I haven't seen the "why". Mind elaborating why it's so bad for non-programmers to have LLM create some stuff for themselves?
spc476 | an hour ago
Because non-programmers didn't add "... and make it secure" to their prompt?
Seriously, I'm concerned about that. Maybe not for stuff that runs locally, but for anything network facing. How secure is your acquaintance's store? I recall one tea and spice store I used to frequent [1] would show the site if you had bookmarked it, but if you came from Google or Google Maps, it would show a site selling knock-off Viagra pills, and this was from a decade ago. The modern Internet is a rough place.
But do I believe that only programmers should have access to open source? No. That's how I learned to program. Well, open source and having to debug code.
[1] The store is no longer in business and I don't think it had anything to do with the web site. It was brick-and-mortar store in a small town and it probably failed mainly due to lack of business.
mplant | 6 hours ago
You’re being so vague and unspecific here that there’s no point in responding to you because I’d have to imagine what your point actually is.
flockofbirbs | 5 hours ago
Let me clear it up: using LLMs is a disservice to yourself and the entire world and thinking they are "genuinely amazing" is quite sad.
mplant | 5 hours ago
Man I don’t care I just wanted my iPod to work so I could use my phone less
aarroyoc | an hour ago
Individualism over the collective in its best
elliotmorris | 3 hours ago
kornel | 5 hours ago
We're on a programming forum, so it's a given that people here care about the craft itself.
But there are people who don't want to program. To them software is just a hassle standing between them and their goal.
That's normal. Not everyone wants to cook fine cuisine, many people get fast food. Not everyone wants to make custom furniture from the finest oak, plenty are happy with IKEA, etc.
And now software is moving beyond artisanal passionately crafted projects, and machine-extruded code nuggets are on the menu.
flockofbirbs | 4 hours ago
Craft vs. mass is a false equivalence. LLMs are not the IKEA of software as their products are created with precision. Writing code without LLMs is not using hand tools instead of power tools. I use efficient and precise power tools when I'm woodworking and when I'm programming. I cut rabbet joints on my table saw instead of cutting dovetails with a hand saw. I use Emacs instead of Notepad. LLMs are something else entirely because they are inherently probabilistic black boxes. They resemble slot machines a lot more than CNC machines.
dimonomid | 4 hours ago
Are you saying humans are not probabilistic black boxes? I mean, the history of our industry shows that humans just can't write bug-free code.
flockofbirbs | 3 hours ago
I was talking about machines, not humans. But a human can give an explanation for why they did something. An LLM cannot as they're all pattern recognition, no reasoning.
dimonomid | 3 hours ago
So when a human says "I overlooked that bug", which happens all too often, how is it better than LLM?
spc476 | an hour ago
Because a human can learn from that mistake?
mplant | 5 hours ago
Johz | 7 hours ago
Can you expand on this somewhat? What do you see as particularly depressing here?
flockofbirbs | 7 hours ago
No, I have code to write.
dimonomid | 6 hours ago
You know, we can ask LLM to write that code, and in the meantime can spend a few mins responding to a comment on Lobsters, and after that get back to the LLM and review its output. And then likely point a few issues, and let it fix them. And then also ask another LLM to review the first LLM's output too, because it often finds issues that humans have missed.
In my experience, it results in higher quality software in significantly less time.
flockofbirbs | 5 hours ago
Cool
elliotmorris | 3 hours ago
Can you explain this emotion logically so we can all understand it, despite the fact that it is absolutely obvious what your stance is? Make sure to not include anything difficult that necessitates us extending our thinking beyond what is in directly in front of our eyes or to do with our own personal desires.
I don't know man, feels like you're not allowed to feel emotions about these things. "This new power plant delivers 2x the power at 20x the carbon emissions, it's awesome!". It's clear as day why someone is opposed. The feigned shock, as though half the populace doesn't feel this way, is grating.
gspr | 3 hours ago
I'm not the parent poster, but I'll give my stab at that question.
It's because to some people (myself included, and I assume a lot of people on this website too) the process of discovery, of learning, of getting stuck, of despairing, of ahas, and of figuring shit out is the most amazing fun imaginable. After a hug from my loved ones, I scarcely know of a better feeling than having solved a hard problem (meaning a problem that was hard for me)! That's why not!
In addition to being fun, we also find it useful (to anyone, also those who don't find it fun) as a way of exercising one's brain. As a researcher, I believe it's also something that inherently drives people who seek out that profession.
Of course, everyone gets to pick their own fun. Lots of people seem to like singing and dancing, for instance. I don't. But I understand enough about fun and inherent drive not to show up at a song retreat and ask "ugh, why do you guys even do this, skilled audio technicians can synthesize a much better voice than that!"
mplant | 3 hours ago
Okay, and like I said, I do too. The problem is that I don't like doing that for everything in my life. Some problems are not actually hard, they are simply tedious (to me!).
Honestly the implication that I don't is kind of insulting!
That is simply not what I'm doing at all. I made an app, for my self, that saved me time to get back to solving the difficult problems that I enjoy solving. That doesn't meant that someone who enjoys working in that domain would not have a fun time working on that problem, it just means I don't want to wait around for them to get my iPod working
gspr | 3 hours ago
Of course. But lots of people who frequent this site do consider problems like
to be genuine fun! Again, nobody is saying that you have to share in our idea of fun. But you did ask why people would find LLMs "not awesome" (which is what you asked "why" to). And you got answers. Now you're saying that those answers don't apply to you. That's fine. But they do apply to us who gave them!
mplant | 3 hours ago
I'm not saying they're awesome. I'm agreeing with the grandparent that says they are:
You can dislike LLMs because they suck the fun out of programming, but some people don't like programming so it doesn't make sense to dislike how other people use them to avoid that!
gspr | 3 hours ago
Even the most LLM-happy programmer I know is clear that you need to verify the LLM's work. Does the end user who doesn't like programming suffer through that verification? Or do they just use the output unchecked, which I assume you'd agree is not "awesome"?
mplant | 2 hours ago
I can't imagine anything I'd care about less as if it's for their own personal use.
gspr | 2 hours ago
Then we're just very different people.
mplant | 2 hours ago
Why do you care how people use their computers? And how does it have anything to do with your original argument that it takes the fun out of things?
gspr | an hour ago
I don't have a say in how other people use their computers, but that doesn't disqualify me from making a judgement about whether I think what they're doing is a good idea or not.
You yourself say what they do is "awesome", so you're clearly allowed an opinion. Are only positive ones acceptable?
k749gtnc9l3w | 19 minutes ago
The real line is not whether they can check the output. The real line is whether they can both understand and accept the error rate.
And there are not so few things where (a) «works 90% of the time, gives weird garbage the remaining 10% of the time» is useful if it is cheap enough, (b) getting to 90% does not take too many «Please fix this problem» promtps (with manual retesting of previous issues, as by assumption forcing the LLM to build a reliable regression test suite is outside the user's qualification).
To borrow the example elsewhere from the conversation on this page — sometimes one needs to copy a playlist perfectly because this is the music for a carefully staged event; sometimes even a maliciously selected 90% out of a shuffled playlist is close enough for a morning run.
landon | 15 hours ago
There are some who believe that this was a bug, not a feature.
flockofbirbs | 12 hours ago
We seem doomed to learn over and over again that there is no global town square.
landon | 6 hours ago
There is, for a time, then it moves on, and our town's is gone.
nickmonad | 19 hours ago
I personally have never really used the “discovery” features of GitHub. I’ve maybe found one or two interesting repos by quickly glancing at the profile feed in the past 10 years but most of the time I find things on external sites like this one, or because I’m already on one repo and a comment links me to another. I don’t see those mechanisms of discovery going away, it’s just a link to another domain. The aspect of having to create another account is mildly frustrating but, for me, wouldn’t be a convincing reason to abandon migration off GitHub.
I love open source and believe in its ability to produce good software. Some classes of software must be open source! I also think that somewhere along the way this idea that all contributions must flow back to the original source became over emphasized, mostly implicitly. The whole point was to enable people to understand and modify the software they depend upon, but there was never a guarantee a maintainer would even look at a contribution. I don’t think the author has a sense of entitlement here, but some people really do and it’s always been an odd dynamic to me. Forking became a non-starter like rewrites. Sometimes those things are necessary and good!
We should foster a sense of collaboration, but ownership and the right to say no is just as important, sometimes more so.
siru | 13 hours ago
Especially with password managers becoming more omnipresent over time, the cost of creating a new account has been becoming less of a repellent, to me at least. As long as I can avoid needing to use an email that might be leaked if the server get's breached, I don't really mind having to register one more credential in my password manager, that will autofill if I ever need to log in again. So distributing things accross numerous/smaller git forges really does not seem to be an issue anymore.
k749gtnc9l3w | 13 hours ago
I have used GitHub search over projects. But if things get pushed towards static Web 1.0 sites (cheap to serve from a small VPS, bots or no bots) with link collections, maybe search-indexing the repositories across the Web gets done again, as it has been done from scratch multiple times before.
NewGirlJade | 7 hours ago
I resonate with the deep melancholy of this post, but not as someone who’s been in the open source community for years and is seeing it change. I’m just a new developer who wants to do a good job, deeply yearns for human connection, and feels like a star looking across the cosmos—everything turning
bluer and bluerredder and redder (my bad, lol) as it accelerates away.Technically things
areseem more accessible now but socially, nobody knows if you’re worth their time, so why would they bother? And that’s the point, isn’t it? Programming is both a technical and a social endeavor, and LLMs can talk like a human, take up digital space like a human, but they’re not human and only “care” about what they’re told to do, not who it affects.I’m at least glad people are really nice at the coding meetups I’ve been going to.
tonyarkles | 7 hours ago
I would strongly recommend continuing to foster in-person connections and friendships, both within and outside of tech. The online tech world is being shaken up pretty violently right now, but the IRL world is actually still filled with pretty great people.
doyougnu | 8 hours ago
I'm skeptical of this take, even though the examples are compelling. My skepticism is really just that people like working on software together and that it will become increasingly difficult to not be open source.
I've been looking in amazement with the stuff that's been happening in the games industry and retro-scene and I think this is a canary for the end of closed source. For those who don't know, people have been decompiling old games or recompiling old games to x86_64. And they are doing it at scale. For example, someone ported Halo and is running a 128-person server: https://www.techspot.com/news/114042-you-can-now-play-original-halo-browser-128.html
So I think we're entering a world where any closed source software can be cracked open with LLMs + ghidra and recompiled onto any platform. This used to be very labor intensive, and now that has all disappeared. I think this leads to two responses: 1) open source your software, if someone will just decompile it, copy and port in a weekend you might as well just open source it; 2) Closed source will move to the cloud and all close source software will live in some server farm.
munificent | 4 hours ago
You're talking about an entirely unrelated concept. I think you mean "open source" as "literally, the public can get the source code" and that's it.
The author is talking about "open source" as a social community and collaborative model of software development.
You can throw some LLM-generated code up on GitHub, and it's the former but not the latter.
thath | 11 hours ago
I help manage an open-source project at work. While we definitely had some adjustments to make this year such as adding clear policies around LLM usage (we allowed them), I feel like we’re now reaching a happy equilibrium where the contributions, even if they are LLM generated are decent quality and welcome.
Our community also has regular online meetings which help us establish and make new connections. Maybe this is where open-source is heading? More value placed on genuine human interaction rather than simply creating pull requests.
mwcampbell | 9 hours ago
We just need to be careful that our desire for genuine human interaction doesn't exclude disabled humans, humans who aren't proficient in the dominant language of the project/group, humans in non-overlapping time zones, etc. Asynchronous text is the most inclusive form of communication. That it includes LLMs too is perhaps a double-edged sword, but I'd rather err in that direction than exclude some humans.
alandekok | 9 hours ago
We posted a statement saying that we would ban people who submitted slop PRs. That seems to have dropped the number of submissions quite nicely.
You should have zero tolerance for nonsense. When some wastes your time, it doesn't matter much if they're being lazy, idiotic, or malicious. They're all DoS attacks. And they should not be tolerated.
singpolyma | 9 hours ago
I feel like the probably read it and tested it and didn't just "take a chance"
conartist6 | 9 hours ago
SO STOP HELPING KILL IT
evert | 21 hours ago
I'm thinking of just accepting handmade PRs but if people are considering using an LLM just give me a good spec so I can generate it myself with my design sensibilities
FeepingCreature | 14 hours ago
Of course, the way to generate a good spec will be to code it with a LLM and then have the LLM reverse a spec from the diff.
k749gtnc9l3w | 13 hours ago
I remember the old saying that to write a good spec you need to implement the system once (then throw it away and implement from the spec)… Maybe this might become a forced methodology.