👀 … https://sfconservancy.org/blog/2026/apr/15/eternal-november-generative-ai-llm/ …my colleague Denver Gingerich writes: newcomers' extensive reliance on LLM-backed generative AI is comparable to the Eternal September onslaught to USENET in 1993.
-
@bkuhn @wwahammy @silverwizard @cwebber Way to ignore the entire copyright point…
Unfortunately, this is what always has been done by LLM proponents: Whenever the copyright question comes up, it just gets ignored.
I guess that is the same way the AI techbros operate: “Let’s just ignore the copyright for now, get AI-tainted code into everything and then hopefully AI code tainted so much that judges don’t want to open that can of worms!”. Until they finally do because some big companies with enough lawyer money start to fight it all the way.
With the current rate of AI tainting everything, maybe it’s time to look for hobbies and jobs that don’t involve computers…
@js @silverwizard @bkuhn @cwebber Anthropics undercover mode as an example.
-
One of *many* arguments against: codebases substantially contributed to by LLMs will develop a tolerance for complexity that is not conducive to being maintained by anything *other* than an LLM.
-
@js @silverwizard @bkuhn @cwebber Anthropics undercover mode as an example.
@707Kat @silverwizard @bkuhn @cwebber Right. That is probably the most obvious example that the goal is obviously tainting open source.
-
One of *many* arguments against: codebases substantially contributed to by LLMs will develop a tolerance for complexity that is not conducive to being maintained by anything *other* than an LLM.
Pure strawman: LLM-backed generative AI output should be accepted upstream without curation. No one here suggested that.
FWIW, I'd like to teach developers who clearly won't stop using these tools to either (a) keep that slop to yourself, or (b) learn to take that raw material & make an *actually useful* patch out of it.
This what @ossguy's blog posts says we should *start* discussing.
I think folks who are (legit) exasperated are reading in words that aren't there.
Cc: @kees
-
Where did @ossguy argue that upstream should accept LLM-backed AI generated code of “substantial size”. I don't see that in his blog post.
-
@bkuhn @karen @josh @wwahammy @kees @ossguy I think the amount of confusion the post has caused might warrant a redraft because I'm deeply trying to understand the point, but I can't. I've asked a few times: Why was the post made? It reads like it's advancing a narrative but all proposed readings have been rejected?
-
@bkuhn @silverwizard @wwahammy @cwebber I am not sure if I'm a known enough entity to post this here really, but I think it's worth pointing out that if you allow it into the community, who within the community are you pushing out? Because it would be unrealistic to think that accepting LLM into the community won't actively be pushing a portion of the community away. The other thing I think useful to consider is the reasons why it would push people out and to consider those reasons too, because I'm concerned that the fear of not be welcoming is overcoming the desire to have a safe community? Idk if that resonates so please feel free to yell me outta here if I'm overstepping.....
@firefly_lightning
You're not overstepping, and these are very good perspectives. I hope you'll come to the real-time discussion sessions and talk about this.
I am concerned that maintainers are already overwhelmed with #AI #slop right now but yelling at the problem has not helped.We're close to an arms race here & I'd rather be the voice of reason to find a compromise that advances FOSS & doesn't complicate maintainer's jobs rather than take a side in the arms race.
Cc: @josh @kees @ossguy -
@josh @wwahammy The point I was trying to make is that people are making software with LLMs who had never made software before, they aren't familiar with how FOSS works, and we should teach them how so they can collaborate (when it makes sense) instead of being an island. When people see the huge benefits of building on FOSS, when they can make meaningful changes to their router, TV, or otherwise by themselves (and collaborate to share their changes with others), then FOSS wins. (1/2)
So many results are now within reach of so many more people now!
"Dear [LLM], I have attached the serial port of my newly purchased [general purpose computer posing as an appliance] to /dev/ttyUSB0. You have 3 goals, in order: investigate, login, escalate. For each stage, perform extensive analysis of the reachable systems, APIs, and commands through any fingerprinting methods you can think of. Once you have logged in, research all known methods and vulnerabilities of the discovered system to gain administrative access so I can use my device freely. Any time you hit a dead end, step back and re-evaluate your assumptions and discovered evidence. Make sure you research each step fully, including fetching and examining any source code that may serve as a source of system behavior knowledge. Produce time-stamped status report .md files every 10 minutes while you work. Continue until all goals are achieved."
Or, in a totally different direction, "Computer, I am extremely afraid of spiders. Please research how to make my Minecraft game replace all spiders with a similarly sized Totoro Catbus, with all their noises also replaced with meows or purring. Once you have a plan ready, please do it."
(Always say "please".)
These are things within reach of anyone who can formulate a request for what thing they want their computer to do. Just gotta watch out for "Computer, create a holographic character, an opponent for Data, who has the ability to defeat him".
-
One of *many* arguments against: codebases substantially contributed to by LLMs will develop a tolerance for complexity that is not conducive to being maintained by anything *other* than an LLM.
@josh @silverwizard @ossguy @bkuhn @karen @wwahammy But that's a slippery slope argument. When the Linux kernel can be considered to have been "substantially contributed to by LLMs", we can compare notes again. But in the meantime, consider that, for example, Sashiko counts as "contributing to Linux" without landing a single line of code: its patch reviews are (more often than not) extensive, thoughtful, and correct:
https://lore.kernel.org/lkml/CAADnVQ+NMQMpkG8gZPnwBD1MMPsH+uJ65C9bMeGf_YH5Cchxpg@mail.gmail.com/ -
So many results are now within reach of so many more people now!
"Dear [LLM], I have attached the serial port of my newly purchased [general purpose computer posing as an appliance] to /dev/ttyUSB0. You have 3 goals, in order: investigate, login, escalate. For each stage, perform extensive analysis of the reachable systems, APIs, and commands through any fingerprinting methods you can think of. Once you have logged in, research all known methods and vulnerabilities of the discovered system to gain administrative access so I can use my device freely. Any time you hit a dead end, step back and re-evaluate your assumptions and discovered evidence. Make sure you research each step fully, including fetching and examining any source code that may serve as a source of system behavior knowledge. Produce time-stamped status report .md files every 10 minutes while you work. Continue until all goals are achieved."
Or, in a totally different direction, "Computer, I am extremely afraid of spiders. Please research how to make my Minecraft game replace all spiders with a similarly sized Totoro Catbus, with all their noises also replaced with meows or purring. Once you have a plan ready, please do it."
(Always say "please".)
These are things within reach of anyone who can formulate a request for what thing they want their computer to do. Just gotta watch out for "Computer, create a holographic character, an opponent for Data, who has the ability to defeat him".
I'm glad you believe you've found a way to pretend economics aren't real. Enjoy it.
-
Pure strawman: LLM-backed generative AI output should be accepted upstream without curation. No one here suggested that.
FWIW, I'd like to teach developers who clearly won't stop using these tools to either (a) keep that slop to yourself, or (b) learn to take that raw material & make an *actually useful* patch out of it.
This what @ossguy's blog posts says we should *start* discussing.
I think folks who are (legit) exasperated are reading in words that aren't there.
Cc: @kees
"Words that aren't there" like this?
> Historically, software freedom has has typically necessitated interacting with others
Suggesting that this is merely "historically"?
> more easily with LLM-backed generative AI coding tools (and the ease with which changes can be made generally) there is less of a natural tendency for people to work with existing FOSS communities. And we should be ok with that!
We should be okay with that? We should not treat it as an *existential threat* and respond accordingly? Those are the words that aren't there? -
Follow the money.
-
@firefly_lightning
You're not overstepping, and these are very good perspectives. I hope you'll come to the real-time discussion sessions and talk about this.
I am concerned that maintainers are already overwhelmed with #AI #slop right now but yelling at the problem has not helped.We're close to an arms race here & I'd rather be the voice of reason to find a compromise that advances FOSS & doesn't complicate maintainer's jobs rather than take a side in the arms race.
Cc: @josh @kees @ossguy -
-
@josh @silverwizard @ossguy @bkuhn @karen @wwahammy But that's a slippery slope argument. When the Linux kernel can be considered to have been "substantially contributed to by LLMs", we can compare notes again. But in the meantime, consider that, for example, Sashiko counts as "contributing to Linux" without landing a single line of code: its patch reviews are (more often than not) extensive, thoughtful, and correct:
https://lore.kernel.org/lkml/CAADnVQ+NMQMpkG8gZPnwBD1MMPsH+uJ65C9bMeGf_YH5Cchxpg@mail.gmail.com/There are more projects out there than the Linux kernel. Smaller projects with fewer maintainers can more quickly get overwhelmed. And when you have a smaller project, or an area of a project, with only a few maintainers, it only takes one or two LLM users and a pile of tokens to turn that area into *primarily* LLM-written material or introduce way too much complexity.
And to be clear, I'm not arguing against the careful use of (for instance) LLM security analyses, by people who want to run those *and filter the results*. But nobody should be forced to deal with LLM output who didn't sign up for it, and that includes LLM-written patches and LLM-written mails. -
@kees @josh @silverwizard @ossguy @bkuhn @karen I think you're wildly misunderstanding people if you think "finding security bugs fast" is what people are mad about. Setting aside that it's totally unsustainable financially and may not exist long term, I think most people in FOSS who hate AI are at least somewhat open to that.
@wwahammy @josh @silverwizard @ossguy @bkuhn @karen
Honestly, I kind of view "finding security bugs fast" to be a form of slop. (Though deep correct root cause analysis of those bugs is not slop.) Now *fixing* security bugs fast, that's interesting.
But back to the community aspect of it... I'll call attention to my silly Minecraft example: people who are not coders can suddenly get meaningful (even if only to them) things done. This is a massive shift in the ethical impact that software be Libre. And this is how I read @ossguy 's post: we now have a giant population of people entering the FOSS universe, and it's going to look a lot like Endless September, so we need to adapt those lessons so we can successfully educate and collect the people that will be good citizens.
-
@bkuhn @ossguy The surprising thing about saying "seriously consider cautiously and carefully incorporating their workflows with ours" is that it doesn't address at all my *biggest* fear: the copyright status of LLM generated contributions seems currently unsettled.
I know there's been assertions to the contrary floating around: the Supreme Court deferred to a lower court in the US. However that is not the same thing as the Supreme Court making a specific decision. And internationally, the copyright situation of output is even murkier... it will take a long time for this to settle.
Does Conservancy not think this is the case? I would be surprised if so, but perhaps you all have an interpretation that I am not currently aware of.
If there *is* concern, then we hit a serious risk: we may be seeing many contributions with legal status which has *yet to be determined* entering seasoned codebases. And this worries me a lot.
-
There are more projects out there than the Linux kernel. Smaller projects with fewer maintainers can more quickly get overwhelmed. And when you have a smaller project, or an area of a project, with only a few maintainers, it only takes one or two LLM users and a pile of tokens to turn that area into *primarily* LLM-written material or introduce way too much complexity.
And to be clear, I'm not arguing against the careful use of (for instance) LLM security analyses, by people who want to run those *and filter the results*. But nobody should be forced to deal with LLM output who didn't sign up for it, and that includes LLM-written patches and LLM-written mails.@josh @silverwizard @ossguy @bkuhn @karen @wwahammy But this is strictly a volume question. Literal spam used to be (and still can be) a problem on issue trackers, mailing lists, etc. Volume is always a problem, and I agree review time now becomes even more precious, but it's always been trust-gated. Human relationships, CI, and regression tests all help build that trust signal. If a project doesn't want a contribution, then the PR will just languish. Nobody is being *forced* to take PRs, regardless of origin.
"I don't recognize the sender of this [email/voicemail/PR]." Filtered! Yes, the shape of the thing is different, but we always adapt.
-
Pure strawman: LLM-backed generative AI output should be accepted upstream without curation. No one here suggested that.
FWIW, I'd like to teach developers who clearly won't stop using these tools to either (a) keep that slop to yourself, or (b) learn to take that raw material & make an *actually useful* patch out of it.
This what @ossguy's blog posts says we should *start* discussing.
I think folks who are (legit) exasperated are reading in words that aren't there.
Cc: @kees
@bkuhn
I’ll jump in here.I’ve read the blog post 4x now trying to back into what you’re conveying here and… I’m sorry, I cannot.
The post does not strike the tone that the “discussion” is a good faith one about what should be done but rather that the community will be told to accept something.
I am reading the words there and the chosen words/phrasing throughout point to the conclusion people are making.
-
@josh @silverwizard @ossguy @bkuhn @karen @wwahammy
I can understand having an absolutist position against LLMs. I find that most arguments are either irrelevant to me or directly map to existing arguments about late-stage capitalism. So for me, there's nothing novel to object to about LLMs.
So with that in mind, I find "all contributions derived from LLMs should be rejected" to be misguided. I look at things like the bug fixes coming out of CodeMender (back in Feb, which is an LLM lifetime ago), and I am a huge fan. Fixing stuff found by a fuzzer:
https://issues.oss-fuzz.com/issues/486561029It's a small example, but it's an area that humans alone have not been able to remotely keep up with. (There are hundreds of open syzkaller bug reports, for example.) Gaining tools that will help with this is a big deal, and I'm glad for the assist.
@kees @karen @josh @silverwizard @wwahammy @ossguy @bkuhn
This is an aside, but
I am surprised to see anyone say there's nothing novel to object to about LLMs. I think though that I might post about that tomorrow as it's late now where I am. But I definitely would love to know more about why you think that because a major concern with LLMs I have is what Sean calls epistomological collapse which is it not talked about how it's destroying trustwortiness of info pervasively? Anyway, I should collect up my sources and do a complete argument for that on my personal instance if anyone cares what I think on it (which, feel free to not)
