Why aren’t people moving away from Github? There’s Codeberg, Gitlab, and radicle. What’s holding them back?
I still host a significant part of my code on GitHub, despite moderately hating Microslop, because:
- Their CI is mature and super fast. My gorgeous instance has selfhosted CI, but some actions only work on GitHub (release-plz has bugs for forgejo), and their CI servers are fast as fuck.
- Visibility. People look for code on GitHub. Maybe maybe maybe they look on code on Codeberg, but without a doubt, nobody looks for code on my selfhosted forgejo instance. I am looking forward to forgefed integration sonce over a year, but it’s not there yet.
Also, GitHub has become an identity provider for many services, such as crates.io to release rust crates (packages).
Search is another key benefit. It’s really nice being able to search the gamut of open source repositories for bits of code.
The problem isn’t AI integration in code editors; the problem is people who let it think for you and blindly accept the results.
It’s great for automating repetitive tasks and setting up frameworks but you’re bananas if you let it commit for you.
It’s why if I have to use AI integration, I’ll specifically prompt it to give guidance and links to interesting articles on how do do things and have it teach me how to do things not just do dump all the code out already completed.
I think part of the problem is Microslop’s Github Copilot in the web, which makes it possible for non developers to quickly and easily create PRs without understanding a single thing about programming, let alone software development.
It wouldn’t surprise me of people genuinely think they are helping.
This is interesting and sounds like how I’ve been using it - basically like customised stack overflow answers.
Would you mind elaborating a little on your approach? Are you saying you provide it with guidance and links or are you asking it for those?
Well I basically tell it to not just do all the code and dump it out to me; I instruct it to explain the rationale, reasoning, and code and then provide external links for additional reading on the subject instead of just doing, I turn it into an instruct model so I can at least expand on my knowledge and then not have to rely on it as much the next time.
Basically, yes, like a Stackoverflow model from the early 2000s.
For instance, something like this: "
When talking about subjects involving programming and coding, the key goal should be instructional and informative to not only include code and samples but also how they work so in the future I can continue and expand on my knowledge. Also suggest places to expand and learn in the future on any programming or development topic. NEVER auto commit or create pull requests in my repositories without asking and waiting for a confirmation first. I prefer to review all code first for learning purposes and QA purposes."So kind of like a personalised learning assistant? I realise it’s different but this inverted instruct approach puts me in mind of Doctorow’s reverse centaur!
Don’t you find that the links you get are hallucinated though? Even if they’re not now you can imagine this collapsing into slop echoes…
I’ve tended to ask for examples to help me bootstrap new projects. A bit like getting customised docs. I certainly haven’t had enough success with generated code to think about automatically adopting it.
Let’s deal with the mountain of AI slop garbage in the same way as what worked before:
We need some kind of reputation system for open source contributors. If you push slop, I don’t want to see your PR. If you consistently make worthwhile contributions, I’ll check out your PR.
How has that worked before? I’ve never encountered the system in practice. Can you give an example and explain how it works?
My fear is that newcomers will be locked out because of the assumption their code is LLM code but it genuinely is theirs. Or that they used an LLM, are willing to learn, think the code is genuinely good, but don’t know why it’s bad.
Very curious how the existing system works.
You mean reputation systems for filtering out low quality submissions? That’s why they exist. Spam filters, reddit, stack overflow, Lemmy, whatever. And people say “what about newcomers who start with no reputation” but it always seems to work out somehow.
My point was that “was this made by AI” is the wrong question. Not everything made using AI is dogshit, nor is everything made by real people always good. And you won’t always be able to tell the difference at all. Rather, let’s focus on the spam problem
Binary reputation systems aren’t good. I can say something that right and it can be downvoted because it goes against a person’s beliefs, because I’m unpopular, because a certain group doesn’t like it, because, because, because. Popularity is not a good measure of quality. Just look at the “publish or die” system. Just because you’ve been cited multiple times doesn’t make your paper right.
Imagine a trans contributor being downvoted just because they’re trans. How is that a good system? Do you expect trans people should only contribute in software projects where trans people are accepted? How are you going to prevent brigading?
I’m not here to debate you. I think it’s clear by now what I think should be done. If you disagree, that’s fine. No system is going to be perfect.
Abandon git, return to subversion.
I felt like I had imposter syndrome at my tech job, and then we got a boss whose only contribution is to make chatgpt goals and objectives. Which he can’t understand them himself, because he gets called out in the meeting for obvious bullshit, and he defends himself by saying he has chatgpt summarize it.
AI has made me feel better about my own capabilities, because I realize most people are idiots. AI makes them clearly visible to everyone, and theyre too dumb to realize how obvious it is. Its a trap for the lazy, the 40% that do nothing at their job, and they are too clueness to understand what they’re doing.
most people are idiots. AI makes them clearly visible to everyone, and theyre too dumb to realize how obvious it is
It has been kind of amazing. I’m sure they’ll catch on, eventually.
I am pretty strict on my standards but every day I spend jobless I dissappoint my family and green squares on github are the main thing recruiters look at.
On the bright side I found out if you change date in linux and commit, the green squares will fill in retroactively.
May or may not write a script.
You might want to look into
git commit --date
Why aren’t people moving away from Github?
I write open source software because it’s fun, and I publish on Github because it gives me a stronger professional profile.
So yeah for me it’s the potential difference between putting food on the table or not. Github stars suck for many reasons but they do help you stand out.
I recognise your name, but not sure from where… It for sure isn’t github BTW, because I don’t use it, but it does look familiar.
Anyway, if you happen to be known due to something you wrote, that’s great but would your popular project be used less if it weren’t on github, even as a mirror?
And do you believe having a public github profile when you started out (/ whatever project(s) you’re known for weren’t popular) made the difference in your hiring procedure? Did you ask? And if it did, would it have made a difference had you linked to a repo on codeberg or Gitlab?
How moving to other source control systems would help? People can still generate code via LLMs and submit changes for review
Microslop has an editor in Github that makes it dead easy to create PRs with Copilot. It’s very low friction. On another platform, you have to download the editor first and install some agentic coding plugin (or download an agentic editor) and pay for it to get started. Much more friction.
Less popular, less traffic - double edged sword.
Makes sense, wouldn’t call being less popular a true con if all you care it’s indexed for search and is accessible
How do you know those projects are shutting down and not moving?
I hope this is a good question:
What happens in a couple years when all this code that’s been written by Copilot and the like, Microsoft then turns around and says, “OH YEAH, BTW THAT WAS GENERATED BY OUR AI SO NOW WE OWN YOUR APP!” Look, most social media ToS says anything uploaded to their sites is owned by them now, royalties-free.
Right now it’s no big deal to any AI company because more code means more training for the AI, but will we get to the point that they’re happy with code output enough and then turn around claiming they own those? Plus, any successful apps are then basically free/no cost contributed projects?
Bonus: Also, what happens when AI is trained on AI-written code that was initially wrong by AI? Is the system doomed to never really improve because of so many inaccuracies?
That is an excellent question and I wouldn’t put it past them. Only time will tell.
They are already throwing a fit because Chinese companies are distilling their models, so them trying to copyright their copyright infringement machines is definitely plausible. And with a US system as it exists today, it might very well work. Hopefully the rest of the world will have told the US to fuck off by then…
Right now it’s no big deal to any AI company because more code means more training for the AI, but will we get to the point that they’re happy with code output enough and then turn around claiming they own those?
At least in the US:
The vast majority of commenters agreed that existing law is adequate in this area and that material generated wholly by AI is not copyrightable.
So it seems unlikely that they would be able to claim any ownership.
As for the rest of your comment (the parts around ownership): you always own the copyright for any copyrightable work you create, including code. When you post on a website, according to the ToS of that site, you’re licensing your comment/code/whatever to the website (you need to for them to be able to publish your work on their website).
Some (many, most depending on what you use) websites overlicense your work and use it for other purposes as well (like GitHub), but in the US the judges have basically ruled that AI companies can pirate whatever works they want without any attempt to license them and still be fine, so the “overlicense” bit is more of a formality at this point anyway.
What happens in a couple years when all this code that’s been written by Copilot and the like, Microsoft then turns around and says, “OH YEAH, BTW THAT WAS GENERATED BY OUR AI SO NOW WE OWN YOUR APP!” Look, most social media ToS says anything uploaded to their sites is owned by them now, royalties-free.
courts have already ruled that AI can’t own copyright. if it’s not generated by humans, it doesn’t generate copyright.
How does that interplay with the whole “Corporations are people” and if the corporation owns the llm, it could theoretically claim ownership of what the llm generates? (To be clear I agree with the decision that ai shouldn’t get a copywrite and don’t think corporations are people but I am genuinely curious)
dunno. I’m not an IP lawyer. however i’ve read a few case summaries where they’ve tried to say putting inputs into an LLM gives copyright and the courts ruled it didn’t. if it doesn’t work for humans, it doesn’t work for corporations either.
granted, the area of law i work in, there’s lots of difference between legal outcomes if you’re merely set up as a partnership instead of a corporation so who knows.
LLMs have made it so that it takes longer to determine whether content is bad than it takes to make bad content. The solution SHOULD be to demand that people examine the content themselves and turn it into high-quality content, but that’s not going to happen so long as it is possible for anyone to submit pull requests. The only solution that will actually work is to restrict who is allowed to submit content to your projects.
So many things are ruined when friction is completely removed.
Meaningful friction my beloved
I knew physics class sucked… Always ignoring friction to do calculations…
hey i like spherical cows
Time to go back to email lists. Anyone can come along sure. But it’s only going to be the most determined.
AI agents can trivially send email. I don’t know what the solution is, other than to disallow submissions from people who already have a good reputation with the project. Even that’s not a good solution, because there really isn’t a way to get a reputation that is open to humans, but can’t be easily gamed by bots.
Even projects on other repo systems have shut down. Too many AI submissions for them. LLMs are integrated so deeply into certain IDEs that some developers I’ve seen literally did not know they were using them (no, they couldn’t tell me why they thought writing a prompt in the IDE wasn’t hitting an LLM).
It’s a systemic issue that GitHub exacerbates but it’s by no means limited to it.
I can see people not realizing the LLM autocomplete was an LLM. But not the prompting.
And even then, that’s some fancy ass autocomplete if it’s not LLM powered…
I mean, using an LLM inherently asks you to not think and the kind of person to use them, intentionally or not, is obviously not the smartest cookie in the toolbox
This statement is so banger for 2026 that I’m asking for permission to quote it.
It’s adapted from something my grandma used to say. You’re fully welcome to it
Ironically, github is a bit better at detecting the slop (for now) since the default settings put claude et al as co-collaborators on commits or the project itself.
Only if they’re contributing through GitHub and not through local AI coding apps like Opencode or Claude CLI
It’s not that I doubt you, but do you have some sort of source for this? I’m interested in the metrics.
They don’t provide specific numbers, but here’s GitHub themselves acknowledging it’s a big enough problem that they’re planning on implementing a feature they’ve ignored community requests about for like 10 years.
Thanks! That was just the kind of thing I was hoping for.
In a nutshell, the network effect. At an individual level, if someone wants to leave GitHub, they absolutely can. But unless they’re a repo owner or a BDFL, the project(s) they were working on would still be on GitHub. And that means they can’t access the GitHub PR process for development, or open tickets for new issues, or any other number of interactions, except for maybe pulling code from the repo.
On the flip side, at a project level, if the project owners agree that it’s time to leave GitHub, they absolutely can. And while they could convince the primary developers to also leave with them, the occasional contributors might still be left behind on GitHub. Moving away from GitHub could potentially cut the number of contributors down by a lot. And what’s guaranteed is that the project will have to retool for the new space they move to. And if it’s self-hosted, that’s even more work to do, all of which is kinda a distraction from whatever the project was meant to do.
The network effect is the result of the sum being more useful than its parts. When the telephone was invented, a single telephone on its own is entirely useless, because nobody else has one to use. But with ten telephone, one person has the potential to call any of 9 other people. With 10,000 telephones, that’s over 9000 people they could call, or those people calling them. At a million phones, the telephone is well entrenched into common usage. Even when more and more people despise making phone calls, the telephone is still around, having changed forms since the 1980s into the modern smartphone.
Why? Because networks are also stable: if a few thousand people give up their smartphones per year, the utility of the telephone is not substantially changed for the grand majority of telephone users. The threshold to break the network effect varies, but I hazard a guess that if 1/3 of telephone users gave up their numbers, then the telephone’s demise would be underway. Especially in the face of modern replacements.
I would regard GitHub as having a network effect, in the same way that Twitter should have collapsed but hasn’t. Too many local governments are invested into it as their sole social media presence, and in doing so, also force their citizens to also subscribe to Twitter. GitHub is not a monopoly in the sense that anti-trust laws would apply. But they are a monopoly in that they own the platform, and thus own the network.
Also, it might be helpful to think less in terms of switching and more in terms of transitioning. Rather than moving from Github entirely all at once, start by making a Codeberg (for example) account and moving some of your things there.
This is indeed practical advice on how to do a transition, but my answer was primarily in response to the OP’s question, which was about the reasons why people don’t even try to transition. I don’t at all suggest that a slow transition is in any way invalid. But to even get to that, the reasons for not transitioning have to be overcome. And for any sizable project, community cohesion is going to come first, or else the result will be less than the sum of its parts.
There’s a lesson to be learned about making tools that need to be used with subtlety and making them more approachable than a fisher price hammer, and then getting shocked that people use them inappropriately.
I’m sure *this time* humanity will learn it’s lesson.
Literally computers
Literally everything! From nuclear energy to fire. The fiftieth time the caveman that invented fire used it, he said “eh, what’s the worst thing that could happen if I don’t do all the same stuff I did last time to keep me safe” and his whole fucking village burnt down.
Laziness.
