I’ll jam with my friends, make a game in a weekend, but I have no interest in dealing with the egos of big FOSS projects.
However, I am seeing vibe coding create a new type of community. In the music space you have tons of people vibe coding new OSes for music hardware.
Akai has even promised to officially support custom firmware on the MPC Sample Keys! Custom firmware has existed for decades, but maybe a tiny handful of people could make it. Now just about everyone can!
I don't know if it's good for your CV. I have done a ton of open source before AI was a thing, and the truth is that nobody cares. I still have the basic coding test in interviews, even if the team interviewing me is actually relying on my open source work in their project.
And AI making it easier to contribute to a project, I think it's even less valuable today. Do it if you want to, that's all!
As a user, I like it when I own the devices I buy and can install and modify the software I want. It's not always the way it is, and that's life. I won't go to other projects and tell them that they must change something because I want them to.
If I keep my code proprietary, you are beholden to the shape of the community I dictate.
The license is first, it precedes, any notion about community.
Fabrice Bellard doesn't need to personally administer to the ongoing development and community functions of QEMU or ffmpeg, it is sufficient that he released the code as open source, and others carry the torch from there.
And if people disagree about the form those communities take, they can split them as they see fit. As happened with libav and ffmpeg for a number of years. This too is fine, and is only possible because of the license.
First you're just disagreeing about the relative importance of "community" vs the license used. This does not seem like an important distinction, blog rhetorical structuring aside, since the value of each is really "it depends" (in the extreme, neither community or license matter if no one ever sees or hears of the code, for instance).
Second:
> If I publish my code under an open source license... If I keep my code proprietary...
That isn't even the distinction that the author is making. They're only talking about communities around open source code. They spell out the dichotomy quite explicitly:
> This is why I don't care as much anymore about a project being under a copyleft free software license or a non-copyleft free software license.
Im sympathetic to the perspective in some ways but I think it's wildly impractical to actually try to keep AI out of software development.
I suppose an anti-ai optimist might hope for a future where AI written software proves too problematic over time and things return to their ideal vision of no AI in most software.
I really, really dont think that's in the range of possibilities. So what would be left?
- pockets of anti ai software
- these people slowly disappear or the window moves ("ai in software BUT ...")
- ai use is completely and thoroughly normalized
But I believe that the anti-ai crowd will end up with contributions by people who pretend they don't use AI. Which may be fine if what you want is that "contributors own their work to some extent" (i.e. they need to understand something in order to pretend they didn't use an AI), but for those who don't want AI because of the inequalities it creates or the carbon footprint, then... well it doesn't work.
I think this is a certainty. Future generations will be born into a world where education, technology, and jobs have corrected for AI's existence. They will never know a world without it or have the experience of the current growing pains.
We've seen this plenty of times with new tech in the past, and as disruptive as AI is, I can't imagine the cat going back in the bag.
1. All modern LLMs are (in my eyes) inherently morally problematic. They were trained on stolen information, and where they didn't break the law they corrupted social contracts. Their manner of operation is also problematic: disadvantaged communities are being exploited at a rapidly increasing disproportionate rate in the name of "more compute".
2. I am fundamentally disinterested in the productivity mindset that LLMs lean into. I don't care to "produce more"; the value of my work does not lie in the amount of money that can be gotten in exchange for it. This is an easier position to hold as someone who does not hold a typical corporate job, but even if I did I would still resist because an increase in my output would not correlate to an increase in my salary.
At the root of it all is people. I care about people (their existence, their happiness, my connections to them, their connections to each other), and any use of LLMs is (in my mind) anti-people, so I want no part of it.
It's bizarre to me that over what feels like less than a year, enforcement of intellectual property and copyright laws has become the most important moral principle. Suddenly, it's your moral duty to enforce Disney's trademark of Mickey Mouse.
I reject this moral framing completely. Information wants to be free!
Further:
> They were trained on stolen information, and where they didn't break the law they corrupted social contracts.
I completely reject this framing. Learning on something has always been part of the social contract (public libraries are a great example).
100% agree. That's what it is all about. That, and our time.
> and any use of LLMs is (in my mind) anti-people
Hard disagree. A lot of good can be done for the people with this tech. Blanket rejection of an entire category of technology has not worked out well for the disadvantaged communities in the past. We have plenty of history.
I think about how disappointing it is when a new project makes it to the front page of HN and I click through and it's just... vibe-coded slop. Looks pretty at first, but it's clear that there isn't really much human underneath. Lots of obvious interface problems or just a distinct lack of a personal touch. It's empty, and it makes me sad.
Or even not in software, I think of all the people who have been so quick to offload their thought processes to a statistical model. What do you mean you don't know what you want to eat at this restaurant so you asked Claude and got recommended something that isn't even on the menu?
And that's just individual stuff. I am genuinely fearful of the integration of AI into, say, the legal and judicial processes, or governance, or whatever else. People give the models wayyyy too much credibility considering they (the models) cannot give an accurate confidence interval for their results, and I think it's going to make things much worse for disadvantaged communities to a degree that the communities will not be able to overcome by using the technologies themselves.
Yes, I think this is a key point and I'm not surprised to hear it. It's a fundamental problem with AI. It's not something that will disappear as it becomes more useful or makes less mistakes.
How do you feel about a dependency that was written by AI? It seems like things will get progressively harder and harder to maintain a real boundary, outside of just writing everything yourself. And perhaps hosting.
For the time being, I'm content to simply not use it myself. I don't think it's reasonable for me to foist my morals onto those around me, in the same way that my very close vegan friend still talks to me when I order a whole milk latte. I don't think there's a way to exist in modern society without incurring some degree of (likely indirect) immorality, so it's all about setting boundaries for myself where I can and accepting where I end up.
Then you simply would be let go.
But, of course, this is also part of the immorality of the AI industry.
Whether they will keep doing so is another matter entirely.
The train will keep going for another 6 months at least.
> where the anti-ai crowd ends up in a few years.
I think if LLMs turn out to be unviable in a few years, then anti-AI crowd (hello!) can stand there proud as they did not deskill themselves and kept themselves more independent than most of their peers.
I admit that the above sounds more like a delusion at the moment, but hey one must dream.
I'd prefer to concretely define the term before answering this. It's loaded. I think "bubble" is generally thrown around way too liberally anytime something is goes up a lot.
It's possible that valuations will crash. I think it would be temporary though. I think something like the internet tech stock crash of the early 2000s is possible although I think the fundamentals are much better here. I would wager it wont crash anything like that. I mean, would you liken Anthropic to Pets.com? And keep in mind even with that crash the internet is more ubiquitous now than ever, and the major players then are not necessarily the major ones now. Anthropic and OpenAI can go under.
> What happens if Anthropic and OpenAI go under?
What would the implication of that be though? The AI models arent going away. The tech is still there.
I guess the main features I'm thinking of when talking of a bubble is 1) demand (especially at the real price) and ultimately return of investment is not there and 2) circular financing is keeping the frenzy ongoing despite that. This by definition is not profitable and sustainable and if that becomes the mainstream belief then nobody would lend the industry money, rendering it bankrupt.
> The AI models arent going away. The tech is still there.
I'm under the impression that AI data centers (which are very different from other data centers) are only good for a handful of years at most, because of the GPUs limited lifetimes, and also are expensive to run (electricity) which makes AI fundamentally different than DotCom as the infrastructure built back then which is fiber cables was very much not like that. And you need AI data centers for both inference and to train new models. Which is why I asked you if you think we could keep using the same models we have right now, in which case we'd need much less compute and get lower maintenance costs to keep this going even after a crash.
Was Pets.com burning tens of billions of dollars per year? This comparison is not favorable to Anthropic.
Do generally agree that the models aren't going away. Neither OpenAI nor Anthropic are financially viable without government intervention, but it's not hard to be profitable in this sector and it will be even easier to be profitable after the unprofitable guys stop spending hundreds of billions in fake money inflating hardware prices.
But open-source software is far more about the process than the product. And therefore longevity of resulting software can be expected to be far greater. Foundations of our software stack - libraries, source control operating systems, compilers, databases and numerous other pieces require people capable and willing to do a long-term care of the software - hence, the community that need to last decades. Linux and PostgreSQL is 35 years old, git is 21, OpenSSL 28.
I believe human-only software development will remain for the foreseeable future, despite the growth of LLM-based ones, just like mass-produced grocery bread didn't displace the artisanal bakeries.
This seems to be more on the tool-wielder than the tool.
Linux itself is receiving LLM-assisted code (a quick search saying 20% at the moment, although I don't know how accurate that is)
In enough cases, I'm sure many of us can think of some, some of the relatively recent "community" stuff has been counter-productive to an open source project, abused as an opportunity to enforce on clueless bystanders whatever fad orthodoxy of the day, or in the worst cases campaign viciously against specific demographics under the banner of doing the opposite.
Talented engineers who just want to do an open source project don't want any of that.
Separate from all that, I'm 100% with no Discord and no GitHub, both of those are total cancer.
If you want a project to be community driven, adding an agreement that empowers its leaders to punish and exclude any community member for any reason, including reasons that have nothing to do with the project itself, is the opposite of that.
Codes of conduct result in dictator-driven projects, not community-driven projects.
And besides that, someone can always make their own community around FOSS software. Fork and move on. But the new community will compete with the old, and it's possible that the fact that people stay with the old community means that they actually prefer that community
developer communities can also be valuable but i think increasingly people are finding they can be quite toxic. the chillest communities i've found have actually been ones primarily made up of non-devs vibecoding. no egos, no endless arguments, holy war battles, or bikeshedding. they just build stuff and share. it's kind of nice acutally.
I disagree. I think AI centralises power and reinforces inequalities, and that's before we talk about cybersecurity and carbon footprint.
It is not clear at all that the benefits outweigh that.
What is clear, though, is that it's a very useful tool to write software.
You need to build a thing first. Community doesn't matter if you don't have cool software. License doesn't matter if you don't have cool software. I'm seeing people forget this principle a lot recently.
The best thing you can do to promote free software is to make good free software. I was recently invited to sign a European digital sovereignty petition (which I am not eligible to sign). Know what helps much more with that? Making European digital sovereign systems. Instead of petitioning the government to switch MS Office to LibreOffice someone would have to actually put in the effort to find out why the government doesn't use LibreOffice, and fix that. (It's probably not because it doesn't run on their computers!)
You have to use copyleft licenses or else the software quickly becomes unfree, if it's actually worth anything. This doesn't happen to most non-copyleft free software because it's not worth very much, not because it doesn't happen.
A community of users doesn't have much value. Any forum or Discord guild or mailing list is secondary to the actual software and it actually being used. Plenty of successful software has no "community" - sudo, for example. Plenty of software with a "community" is not successful. Some software is successful and has a community, like Blender.
A community of developers is only valuable if they're doing valuable things. The Fediverse chat group that shared the link to the European digital sovereignty petition - no useful Fediverse software seems to be coming out of the people in that group. It's all about ivory-tower protocols that won't possibly work in the real world, and stuff like that. No actually working software.
We also have to stop over-estimating the worth of free software just because it's free. Good free software is good software which is also free software. No software is bad, bad software is also bad. Some random code you wrote and published somewhere on the internet usually isn't very useful, and that's probably why it doesn't have a community. Don't try to create a community first - that won't make the software any better. You have to make good software, that can spread by word of mouth, first and then you can have a community after that. Think about the number of random little programs on the internet, now think how many are installed on your computer - not a very big percentage, but the ones that you do have installed, many people have installed. Those are the useful ones.
In most projects the original creators remain the main developers no matter how many other contributors join. Late contributors having a significant effect on a project is the exception, not the rule, and when it happens it's actually an undesirable power inversion since the original creator retains unearned influence - see the PolyMC incident. There is no way to become the chief developer of a project other than having created it. If you think you can rewrite the whole thing and make it better, don't "join the community" - make a fork (which makes you part of the community, but that is not the main goal!).
If you're actually doing the majority of the work, offering a dual license is not unethical. We are forced to live in capitalism whether we like it or not, and we should be trying to squeeze money out of capitalist entities - lord knows they're good at squeezing money out of us. To not play that game is to lose by default. This isn't incompatible with being nice to people who are not money-squeezers, if you dual-license. A community fork can legally continue as AGPL-only. (You don't necessarily need a full CLA either, you just need people to agree their code will be in your commercial version. Programmers treat licenses and CLAs like satisfying the borrow checker, but that's not how they actually work.)
I wonder how people work with omarchyslop (the linux distribution spam-slopped by DHH). Do they have any real community outside of paid fanboi accounts? If so, how do they interact? I feel that there is no real exchange of e. g. code since claude slop just autogenerates that. It must be strange for them.
One perspective that you might want to conside, and which is discussed by Oxide folks in this podcast episode [1], is that agents are a kind of non-deterministic compiler which is still, to some extent, capable of producing useful output. Consider this: even before agents, ruby source code might have been seen slop to someone writing a strongly typed language.
Lastly, there is a community for users of that distro if you care to look for it (I am not among them).
Unless you use AI.
Same people, same mute message; every time, with no delay. "Accept me, then I'll see if I want to accept you".
Fuck off.
More evidence of tribalism, combined with exceptionally precise confessions running as whataboutisms
Pelagicaustral is on the money.
It would just be nice if people understood the difference between wanting a club with a gate versus a community. I am impressed, though. I never thought I’d see a vomit diarrhea load of bumper stickers, but a digital form at the end of the article.
It's a revival of old web whimsy. Sorry for your loss of it.