FOSS in times of war, scarcity and (adversarial) AI [video]
fosdem.org
fosdem.org
The "end of history" hangover is real. We went about building the modern stack assuming bad actors were outliers, not state-sponsored standard procedure. But trying to legislate good use into licenses? I don't know how you would realistically implement it and to what extent? That solution implies we have to move toward zero-trust architectures even within open communities.
As an example: formal proofs and compartmentalization are unsexy but they're a solid way we survive the next decade of adversarial noise.
I remember reading a quote somewhere that stuck with me. Paraphrasing, "If the architecture of my code doesn't enforce privacy and resistance to censorship by default, we have to assume it will be weaponized".
I am out of ideas, practical ones, lots sound good on paper and in theory. It's a bit sad tbh. Always curious to hear more on this issue from smarter people.
It's also questionable to which extent restrictive licenses for open source software stay that relevant in the first place, as you can now relatively easily run an AI code generator that just imitates the logic of the FOSS project, but with newly generated code, so that you don't need to adhere to a license's restrictions at all.
Zero trust cannot exist as long as you interact with the real world. The problem wasn't trust per se, but blind trust.
The answer isn't to eschew trust (because you can't) but to organize it with social structures, like what people did with “chain of trust” certificates back then before it became commoditized by commercial providers and cloud giants.
It’s changing but not completely.
Back in 1770 there were basically 0 democracies on the planet. In 1790 there were 2. Now there are about 70 with about 35 more somewhere in between democracy and autocracy. So most of the world's population is living under a form of democracy. I know that things are degrading for many big democracies, but it wouldn't be the first time (the period between WW1 until the end of WW2 was a bad time for democracies).
I have no idea how we get from here to a civilized internet, though.
It would require it not to be easy to farm (Entropy detection on user behaviour perhaps and clique detection).
That is hard too though.
which is impossible.
- No code is feasibly guaranteed to be secure
- All code can be weaponized, though not all feasibly; password vaults, privacy infrastructure, etc. tend to show holes.
- It’s unrealistic to assume you can control any information; case-in-point the garden of Eden test: “all data is here; I’m all-powerful and you should not take it”.
I’m not against regulation and protective measures. But, you have to be prioritize carefully. Do you want to spend most of the world’s resources mining cryptocurrency and breaking quantum cryptography, or do you want to develop games and great software that solves hunger and homelessness?
Some code architectures make privacy and security structurally impossible from the beginning.
As technologists, we should hold ourselves responsible for ensuring the game isn't automatically lost before the software decisions even leave our hands.
I think privacy is essential for freedom.
I'm also fine with lots of censorship, on publicly accessible websites.
I don't want my children watching beheading videos, or being exposed to extremists like (as an example of many) Andrew Tate. And people like Andrew Tate are actively pushed by YouTube, TikTok, etc. I don't want my children to be exposed to what I personally feel are extremist Christians in America, who infest children's channels.
I think anyone advocating against censorship is incredibly naive to how impossible it's become for parents. Right now it's a binary choice:
1. No internet for your children
2, Risk potential, massive, life-altering, harm as parental controls are useless, half-hearted or non-existent. Even someone like Sony or Apple make it almost impossible to have a choice in what your children can access. It's truly bewildering.
And I think you should have identify yourself. You should be liable for what you post to the internet, and if a company has published your material but doesn't know who you are, THEY should be liable for the material published.
Safe harbor laws and anonymous accounts should never have been allowed to co-exist. It should have been one or the other. It's a preposterous situation we're in.
Doomscrolling or porn is just too "appealing" to children, like sugar. Children don't have their minds fully developed to be able to say "no" to them.
If in school everybody has a smartphone and does doomscrolling, your children will do as well. Or they'll be ostracised.
We have had several arguments about no social media and we're only 1 out of 6-ish years in to the too naïve to look after yourself on the internet phase, and the eldest already figured out how to download some chat app I'd never even heard of without permission.
Bad “censorship” is involuntarily denying or hiding from adults what they want to see. IMO, that power tends to get abused, so it should only be applied in specific, exceptional circumstances (and probably always temporarily, if only because information tends to leak, so there should be a longer fix that makes it unnecessary).
I agree with you that children should be protected from beheading and extremism; also, you should be able to easily avoid that yourself. I disagree in that, IMO, anonymous accounts and “free” websites should exist and be accessible to adults. I believe that trusted locked-down websites should also exist, which require ID and block visceral media; and bypassing the ID requirement or filter (as a client) or not properly enforcing it (as a server operator) should be illegal. Granting children access to unlocked sites should also be illegal (like giving children alcohol, except parents are allowed to grant their own children access).
This is still techno-optimism. The architecture of your code will not to that. We are long past the limits of what you can fix with code.
The only action that matters is political and I don't think voting cuts it.
Never estimate how state actors can use violence (or merely the threat of it) to force people to do things. The only way to respond to that is not through code or algorithms or protocols, but through political action (whether it be violent or non-violent)
Nothing is xkcd-538-proof, in absolute terms. Violence is always possible.
But having a tool that is more resistant to authoritarian overreach by being geographically distributed does make it harder to pull these attacks off.
example of what is not possible to fix with code?
Maybe there's 3 of us, and the 4th part of the password/secret/private key is on a server of mine somewhere. If I don't check in for x duration, it wipes itself.
Yeah it means my Monero is gone now, but at least my attacker didn't get it.
Combine that with rate limiting and a dead man's switch.
Cute tech can slow them down until they go through the effort of controlling most of TOR's exit nodes and point the Eye of Sauron at you.
allows small person to be more free by giving assymetric power to heshe.
Power. Real power. The power to kill you, take your property, harm your family, tell lies about you on the news, etc.
I've always been surprised by the naivety of tech people with respect to this question. The only possible solution to power is power itself. Software can be a small part of that, but the main part of it is human organization: credible power to be used against other organized holders of power. No amount of technology will let you go it alone safely. At best, you may hope to hide away from power with the expectation that its abuse will just skip over you. That is the best you could hope for if all you want are software solutions.
some exact piece of hardware or some exact activity of power?
think of it as tdd. we check few simple exact cases before generalising.
Text files don't have power. Appealing to old power institutions to give them power is not the way to create new power either. Legacy systems with entrenched power have tended to insulate those at the top, killing social mobility and enabling those institutions to act against broad interests.
Open source has always been a force of social mobility. You could learn from reading high quality code. Anyone could provide service for a program. You could start a company not bound by bad decision makers who held the keys.
Open source always outmaneuvers inefficiency. Those who need to organize are not beholden to legacy systems. We need technically enabled solutions to organize and create effective decision making. The designs must preserve social mobility within to avoid becoming what they seek to replace. I'm building the technically enabled solutions for at https://positron.solutions
This is the real issue. FOSS was born out of a utopian era in 60's-2000s' where the US was still a beacon of hope. That is fundamentally impossible in todays world of ultra-shark-world-eat-you capitalism and global race to the bottom.
If it didn't already exist, FOSS would not be able to get off the ground today. FOSS couldn't start and survive today. Its survival is in jeopardy.
FOSS doesn't need the US or 1980s counterculture to succeed. It just needs cheap disk space and someone willing to share their code. The price of storage and internet continues to fall, and I think FOSS will be fine as long as that continues.
But the article was kind of about how to control access, the licensing, and bad actors. And that is out the window, anybody can steal your code regardless of the license, and North Korea can use it in missiles if they want, nothing can stop it if it is openly shared.
We reached the limits of societal coherence and there’s no way to bridge the gap
I think you are on to something there. Noise is really signal divorced from tone. Our current consensus protocols are signal based. They demonstrate control, but not rightful ownership. Pairing a tone keypair with a matching signal keypair in a multisig configuration would be compatible with current networks, but also allow a bottom-up federated trust network to potentially emerge?
But the part about FOSS being used in a project not aligned with the creator's values seams hypocritical:
IMO FOSS is a gift to humanity and as such:
"A gift should be given freely, without obligation or expectation, as a true expression of love and kindness"
If this is suddenly being weaponised against us, I don't see how that's not a problem.
If it took some people in the FOSS space this long that it also includes people, companies or purposes they disagree with, then I don't know what to tell them.
That's like saying "I have the freedom to kill you".
Saying that you can create something, then you reserve the 'freedom' to limit what everyone else does for it really doesn't fall under the word freedom at all.
The software creator (human or AI) must give the user of its software the same freedoms it has received.
If it has received the freedom to view the original, readable, source code, then users should have the freedom to view the original, readable, source code.
If it has received the freedom to modify the source code, then users should have the freedom to modify the source code.
Etc.
It's not hard to follow for people who want to do the moral thing.
It's VERY hard to follow for people who want to make money (and ideally lots of it, very quickly).
Permission is hereby granted, free of charge, to any person obtaining a copy of this software [...] to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software [...]
The freedom to run the program as you wish, for any purpose (freedom 0).
5. No Discrimination Against Persons or Groups
6. No Discrimination Against Fields of Endeavor
[1]: https://mit-license.org/ [2]: https://www.gnu.org/philosophy/free-sw.html#four-freedoms [3]: https://opensource.org/osd
While "use for any purpose" has been included, I think considering purpose is a natural extension of this concept. Suppose there were some software project that aimed to practically eliminate the ability for users to share and use free or open software as it is today. Is it more free to allow such a project to be unrestricted from using other software, even if that project would lead to the destruction of free software otherwise?
If what I'm doing is interesting or unusual, LLMs will firstly not recognize that it's different, secondly will screw up when blindly combining it with stuff that isn't different, and thirdly if it's smart enough to not screw that up, it will ignore my work in favor of stealing from CLOSED source repos it gains access to, on the rationale that those are more valuable because they are guarded.
And I'm pretty sure that they're scraping private repos already because that seems the maximally evil and greedy thing to do, so as a FOSS guy I figure I'm already covered, protected by a counterproductive but knowingly evil behavior.
These are not smart systems, but even more they are not wise systems, so even if they gain smarts that doesn't mean they become a problem for me. More likely they become a problem for people who lean on intellectual property and privacy, and I took a pretty substantial pay cut to not have to lean on those things.
It is pretty obvious to me, that being blase about whomever using FOSS for adversarial reasons is not very "open" or "free". Somewhere in the thread there is an argument about the paradox of intolerance and I don't really care to argue with people on the internet about it because it is hard to assume the debate is in good faith.
My point is this: Throw away all your self described nuance and ask this yourself whether or not you think any malicious, war-monger, authoritarian, or hyper-capitalist state would permit a free and open source software environment? If the objective of a business, government, or billionaire is power, control, and/or exclusivity then, well, your lofty ideals behind FOSS have completely collapsed.
Training on my code / media / other data? No worries, just make sure the weights and other derived artifacts are released under similarly permissive license.
At least in the US.
Quite what happens if another country ordered, say chatGPT, to be released under the AGPL since it was trained on AGPL code, who knows.
He's also 72, we can't expect him to save everyone. We need new generations of FOSS tech leaders.
Something about Richard stallman really is out of this world where he made people care about Open source in the first place.
I genuinely don't know how people can relicate it. I had even tried and gone through such phase once but the comments weren't really helpful back then on hackernews
https://news.ycombinator.com/item?id=45558430 (Ask HN: Why are most people not interested in FOSS/OSS and can we change that)
> user freedom, not creators freedom
In his view users are the creators and creators are the users. The only freedom he asks you to give up is the freedom to limit the freedom of others.
I love FOSS, don't get me wrong. But people should be able to say: I made this, if you want to use it, it's under these condition or I won't share it.
Again, imho the GPL is a blessing for humanity, and bless the people that choose it freely.
Don't bless the people that think you are evil for not applying the GPL to your creation.
This is not true, though. As a copyright holder, you are allowed to license your work however you wish, even if it's under for example GPL-3.0-or-later or whatever. You can license your code outside of the terms of the GPL to a particular user or group of users for example for payment.
Really, it's only when the user agrees to abide by the license that you'd have to give access to source code when asked, for example.
> I love FOSS, don't get me wrong. But people should be able to say: I made this, if you want to use it, it's under these condition or I won't share it.
And they can. Whether that wins one any friends or not is another matter.
Creators are not creators, they're also users. There's a very solid chance that a better world for everyone would be achieved if freedoms for all users would be bullet proof. Every user should be able to modify and repair all their hardware and software without creator involvement.
Sure, we can copy software, so it's not like they are taking your house. But "they" may be taking your livelihood.
Ok, objectively perhaps the world would be better, but we can't know. And opinions don't mean anything. What matters is individuals and being fair to them, whatever society grows from that is just what we have.
That said, if we ever go multi-planet, and there is a planet with no copyright and everything is GPL, I'd check it out and imagine I'd feel quite at home there.
With GenAI that's starting to happen with anyway.
The only way forward is the abolishment of copyright.
Similarly, I could imagine carving out an exception when training on copyrighted material without licence, as long as the resulting model is open-sourced.
Training is fair use. The closed models wouldn't be impacted. Even if we assume laws gets changed and lawsuits happened, they just get settled and the closed source models would progress as usual (see Bartz v. Anthropic).
Meanwhile if somebody wants to go all "GPL AI" and only train their models on GPL compatible code, they'd just be restricting themselves. The amount of code they can train on shrinks drastically, the model quality ends up being garbage and nothing was won.
Further, assuming laws got changed, those models would now be incredible easy to attack, since any slip up in the training means the models need to be scraped. Unlike the big companies with their closed models, Open Source efforts do not have the money to license data nor the billions needed to settle lawsuits. It would mean the end of open models.
Essentially LLMs are recontextualizing their training data. So on one hand, one might argue that training is like a human reading books and then inference is like writing something novel, (partially) based on the reading experience. But the contract between humans considers it plagiarism when we recite some studied text and then claim it as your own. So for example, books attribute citations with footnotes.
With source code we used to either re-used a library as-is, in which case the license terms would apply OR write our own implementation from scratch. While this LLM recontextualization purports to be like the latter, it is sometimes evident that the original license or at least some attribution, comment or footnote should apply. If only to help with future legibility maintenance.
I agree with you.
Imagine a parallel Earth where there was a free OS that the majority in the world used called GNU/Felix.
Felix (it/its), who wrote GNU/Felix and who was the project’s strong but kind leader, one day had a head injury that somehow decreased its empathy but raised its IQ.
Subordinates of Felix on the council of leadership noticed that it was adding features that would track all user data to use in some nefarious plan.
In this case, most would agree that for both the freedom and good of all, Felix should no longer lead this effort.
However, they would want to be sure that even the Will Bates’ great company Bikerosoft didn’t lead the project either, because despite its wonderful and ubiquitous Bikerosoft Office apps and Ezure cloud tools and infrastructure, it was a profit-based company.
https://itif.org/publications/2024/07/19/the-us-china-tech-c...
My worry: that it suddenly becomes treason to commit to or pull certain repositories, that certain repositories become nationalized, or that other nation states do the same. Heinous forks and theft of code en masse for fear of being shot by a drone streaming video over ffmpeg.
Of code becoming a regulated munition, and tools that rat you out if you are designing resistance software.
There has never been any inherent political or economic value in open source software. Those things come from deliberate decisions by authors and users such as licensing and mass adoption.
Open source is not synonymous with the GPL and most businesses try to avoid open source software when implementing their core competency.
What you mean here? Businesses often implement their own core code, but they don't deliberately favour closed-source software.
I suppose this is relevant to a subset of HN audience who attend FOSDEM. Even the talk abstract is worth discussion as it highlights an important side effect of FOSS goals and the current state of the world.
This is untrue. It does have sense of truth and ethics. Although it does get few things wrong from time to time but you can't reliably get it to say something blatantly incorrect (at least with thinking enabled). I would say it is more truthful than any human on average. Ethically I don't think you can get it to do or say something unethical.
The lie should be clearly blatant one, something that a reasonable person would never do.
you can't. an LLM chatbot has no concept of truth, or reasonableness, or blatantness. you're willing to accept a distillation of tokens that may or may not form a truthful statement based on probability, with no intent or understanding behind it.
> NGI Zero, a family of research programmes including NGI0 Entrust, NGI0 Core and NGI0 Commons Fund, part of the Next Generation Internet initiative.
with the Next Generation Internet thing at the end receiving money/financing from the political supra-state entity called the EU [1] . So I guess said speech-holder is not happy because political entities which are seen by the EU as adversarial are also using open-source code? Not sure how war plays into this, as I’m sure he must be aware of the hundreds of billions of euros the EU has allocated for that.
[1] https://ngi.eu/
(Disclosure: I once received NGI0 funding.)
There's no "intimate" knowledge required in order to be aware of the EU spending tens to hundreds of billions of euros on the war close to its Eastern border, it has been one of the main topics of discussion in the media for a good time now. Unless this speech holder has lived under a rock since February 2022, which doesn't seem to be the case (he was the one mentioning the "war" thing).