Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)T

teawrecks

@ teawrecks @sopuli.xyz

Posts
1
Comments
167
Joined
3 yr. ago

  • Debian developer Antoine Le Gonidec resigns after option allowing corporate LLM use without disclosure wins vote

    Jump

  • Fear isn’t the right tool to avoid it, though.

    To be clear, I'm saying the opposite. Fear is the necessary ingredient that causes fascism. Everyone is seeing wealth inequality, and jobs going away, and cost of living rising, and (depending on the country) healthcare costs rising and quality getting worse. There is a lot of fear going around all over the world, and that's the classic environment for a demagogue to step in, point at a scapegoat, and stoke populism. As we've learned in the US in the last decade, it's a short jump from there to fascism, and Germany and Italy weren't the only countries to learn the same lesson back during the great depression. Canada was about to elect their own version of Trump, but instead elected Carney in direct response to Trump's aggression. So I'm confident if it wasn't Germany back in the 30s, it would have been someone else, possibly even the US. We did our own internment camp stuff towards the japanese, and there was famously a pro-nazi rally that was held at Madison Square Garden in NYC in 1939. The fear was everywhere, it just came to a head with Germany first.

    You kinda want to avoid it.

    Hah, understatement. And yet democracies still reliably fall for it. And the reason is they have fears that make them prone to exploitation.

    I still don’t see any reason to completely sell out to some specific big companies...they’re trying to manipulate you.

    Absolutely not what I'm suggesting. The opposite. As I mentioned before, I tend to believe that any models trained on stolen public data should be made public domain, and to whatever extent possible, the companies using them should need to license the resulting models from us.

    Edit: but I'm specifically against any govts taking a stake in these companies, because that's effectively a bailout for them. They would love to unload their risk on the people.

  • Debian developer Antoine Le Gonidec resigns after option allowing corporate LLM use without disclosure wins vote

    Jump

  • I'm really unsure about how to compare timeframes.

    Totally fair. I think the relevant difference is that we're still in the middle of the advent of this technology, we still have a hardware shortage that limits access to it, and it began as we were already entering a crisis era and economic downturn. From our vantage point, we only see the damage caused by the tech, not the potential long term impact, and we are (rationally) assuming the worst.

    But if a new scientific revolution could happen as a result of AI for the betterment of humanity, do we want to accidentally prevent that by throwing it out with fascism?

    (And if that is in our future, the next question historians 500y from now will ask is, could it have even been avoided? Much like we could ask now, was there a timeline where the original printing presses were destroyed out of fear of corrupting the masses? And realistically, how long would that have lasted before a new one was built?)

    3 years ago is an arbitrary point in time. Neural networks were invented 70 years ago.

    But we both know that when people say "fuck AI" they're not talking about anything other than generative AI made in the last few years.

    I don't really get what you mean by fear...I think it's natural to feel a lot of rejection towards all of that.

    Fear has defined the Trump era for the last 10+ years. It is why countries all over the world are tilting to the right. It is the necessary ingredient for fascism to gain traction. It happened during the great depression all over the world, even in the US, and it's happening now.

    I'm not saying it's an unnatural or unreasonable reaction, I'm saying it's causing democracies to be prone to rash decisions. If you recall, this thread is about the use of AI in FOSS projects. The most prominent sentiment by 99% of people in this thread is "fuck AI, burn it down and anyone who uses it". And if that fear-based reaction is what prevents well-meaning, intelligent open source maintainers from being allowed to leverage an objectively powerful phenomenon to responsibly and ethically usher in a new scientific/FOSS revolution, we would be remiss to stifle that opportunity out of fear.

    Edit: and to be completely honest, I don't know who is upvoting me, because it's very uncharacteristic of randos to follow a thread this dense between two people. I get the sense that there are bots in this thread voting for/against their favorite ideologies.

  • Debian developer Antoine Le Gonidec resigns after option allowing corporate LLM use without disclosure wins vote

    Jump

  • Consider that you're comparing hundreds of years of the printing press without much lived-context, vs about 3 years of LLMs in the context of the country you live in. It's not the case that when the printing press was invented, that suddenly everyone had access to a bunch of free, unbiased information. It was a complex piece of machinery, it took decades to be adopted widely, and when it did, it wasn't in the hands of the proletariat at first. The people wielding it most effectively were kings. And in the present day, if you live in China today, odds are you are not nearly as pessimistic about the advent of AI as people in the west, primarily because your government has actively taken steps to make you feel safe. Most recently, China made it outright illegal to replace workers with AI. Say what you will about China, but I think americans would like to see something even 1/10th this effective right now.

    I otherwise can't disagree with your summary of the current sad state of AI. The only part I would contend with is that, while it can be misused to "make people more stupid", when wielded correctly it is also a great tool for education. It is able to point you in the right direction of the information you're looking for faster than any other algorithm. Like with any new information, you have to be trained in the ways of skepticism. You have to know how to vet the output for yourself. But it's so much faster at finding that relevant information for vetting than combing through reams of irrelevant text, or renting time from a domain expert, if you're privileged enough to have access to one. Used correctly, LLMs are an incredible accelerator for learning.

    And while not all open models are full open-training, there are many viable open-weight models that you can both fine-tune on your own content, and run locally so that no one is MITMing your prompts; and they do rival the best that openai/anthropic/etc. have to offer. There is currently a whole ecosystem of normal people iterating on various open models outside of the purview of the oligarchs. Yes, we're in a hardware shortage right now, which super sucks, but we're still talking on the order of thousands to 10s of thousands to have your own model running, not hundreds of thousands to millions.

    Outside of LLMs, we are seeing breakthroughs in healthcare. A year ago, doctors were leveraging AI to significantly speed up chemo treatment schedules, leading to a measurable number of lives saved. And just last week we saw Moderna announce a personalized cancer vaccine pass phase 3 testing! The ability to create a personalized vaccine that teaches your immune system how to fight your cancer cells in spite of them constantly mutating was thought to be impossible even 5y ago. Meanwhile, the work is still ongoing, but AI seems promising in its ability to approximate solutions to previously intractable problems in fusion power generation.

    Issue isn’t the maths… It’s the fascists being in control of it. We need to get rid of those.

    Yep, that's all I'm claiming. To me, the active ingredient for fascism is fear. The internet could be full of all the slop propaganda that grok can cook up, but if people aren't afraid of their neighbors or afraid of losing their livelihood, it's all rendered powerless. We can take 100 specific steps to combat fascism, but to quote one of my favorite metal albums, "the fear is what keeps us here".

  • Debian developer Antoine Le Gonidec resigns after option allowing corporate LLM use without disclosure wins vote

    Jump

  • The printing press has been a crucial tool for fascist propagandists ever since its inception. Does that make the technology fascist?

    I just got done reading The Demon Haunted World, and can't help but draw a comparison. It's perfectly reasonable to be afraid for the future of democracy right now, but it really feels like people think AI is powered by literal sprites, dancing around on puppet strings at the whim of the fascists who summoned them. It's math. Math isn't ideological. It is simply a fact of the universe that an LLM is possible.

    Yeah, we have to make sure it is applied responsibly, because yeah it's a very powerful phenomenon. Yes, they were made using stolen information, and I fully believe that all LLMs and training data should be fully open sourced for public use as a result, and making any profit from them is unethical. It's stealing from Peter and Paul to sell to Peter and Paul.

    But one day fascists are bound to fail, and you know what will still exist in the world that comes next? AI/LLMs/some future iteration of a transformer-based neural network. It's not the first technology loved and/or invented by fascists, and it won't be the last. But I think in a panic, people are willing to burn down anything in an 50ft radius of any fascist. Keep the knowledge, ditch the fascism, stop acting out of fear.

  • What is your weirdest self-hosting problem you've had to solve? For me: Mouse inside the server.

    Jump

  • Optical or roller ball?

  • How can i teach myself to contribute to these projects?

    Jump

  • Probably the easiest place to start is appropriately reporting a bug that you're impacted by.

    Take note of the version of the affected program you are using (likely depends on your distro). It could be that the bug has been fixed in a more recent release, and you don't want to create noise for an issue that has already been resolved and hasn't made it to you, or is specific to the build for your distro.

    Once you've figured out where the issue must lie, make a report with detailed steps on how to reproduce the issue. Don't ghost the thread, if a dev asks followup questions, try to help give feedback.

    If you're lucky, and they post a commit to fix the issue, take a look at it and try to understand what it is doing. Over time, you'll start to get a feel for how open development works.

  • Datahoarders, how many Linux ISOs do you actually have?

    Jump

  • The only reason to do this would be to seed them. But my puny upload bandwidth makes that infeasible.

  • Proton VPN's anti-censorship Stealth protocol finally lands on Linux

    Jump

  • Yeah, I do wish proton was more open about their tech. I do know that proton allows you to make free accounts without giving them identifiable info (which makes them a popular choice for scammers, which is annoying when services just block their email domain thinking it's always nefarious). I don't recall what the payment options are, but I'd be surprised if they didn't have any crypto/anonymous ways to pay.

    I still haven't made use of the proton VPN, mostly because, yeah, I don't care for running a proprietary client for something like that, but also the last time I looked into it they did offer wireguard configs that would do the same thing. The client is just intended to be a familiar cross platform interface. No idea about this Stealth mode protocol, though. Seems like that would have to be run locally.

    If you do pay for their email, I recommend bringing your own domain. 99% of the time the proton.me domain works fine, but for the 1% of the time when a system has a problem, it does make me wonder if there's somewhere in their system blacklisting their domain.

    with the cooperation with law enforcement proton has stated to do

    I'm curious what statements you're referring to. For a company to continue operating, they have to comply with local law enforcement to the extent the law requires. They don't have a choice. If you asked mullvad, they would say the same. But the point of all these services is that they (claim to) do their best to not keep any records by design, so that when law enforcement comes with a subpoena, they have as little as possible to hand over in cooperation. But yeah, as you said, at the end of the day we just have to trust that that's how they operate.

  • Proton VPN's anti-censorship Stealth protocol finally lands on Linux

    Jump

  • The nice thing about proton is that they offer a suite of tools comparable to Google, but all privacy respecting (email, docs, drive, etc). So they're an easy recommendation for people dependent on those services. I almost never use their VPN, but it is neat that I have the option of a paid VPN along with the rest.

    Also I only just noticed the other day that proton offers multi-hop VPNs now, so that your entrance and exit nodes are under different jurisdictions, is that just the standard for VPN services these days? I thought that was a pretty neat feature.

  • Linux Mint’s Next Release Will Fully Support Wayland in Cinnamon

    Jump

  • It's supported.

  • AURpocalypse now: a look at the recent AUR attacks

    Jump

  • The AUR should not be thought of as a package manager repo. It should be thought of as a pastebin for pkgbuild scripts, i.e. build instructions. Running them without looking is the equivalent of blindly copying shell commands from stackoverflow.

    If you are thinking "I want to install this package I found, it doesn't exist in any repo, but their build instructions are complex and don't have instructions for arch," a pkgbuild is a great resource. At the very least you can read someone's pkgbuild to see what dependencies and build steps worked for them (in the same way that you can disect a shell script line-by-line to understand what it's doing).

    The only official way to use the AUR is to manually download a pkgbuild file and use manually run makepkg to execute it. All the other tools that turn it into a convenient repo source (ex. yay, paru, pamac) are unofficial.

  • AURpocalypse now: a look at the recent AUR attacks

    Jump

  • Does anyone know if yay gives me the ability to hook my own tool in to review pkgbuilds before accepting them? They argue that they don't want to just give attackers access to a scanning tool, because all they'd do is just iterate on their pkgbuild until it reports "not detected". But if yay gives me an easy way to hook in whatever tool I want, the attacker can't be sure what tool to defeat. If thousands of people all run various tools, surely a few of them will spot the anomaly quickly.

    Edit: it looks like they've added this exact functionality in response to the attacks: https://jguer.space/blog/2026-06-15-yay-v13

  • [META] Are paid for, closed source projects, being advertised on this community, appropriate?

    Jump

  • That sounds like you're describing someone who is only making a lemmy account because they see potential customers they want to advertise to.

    That's the exact reason I don't want someone to make a lemmy account.

  • Arch Linux AUR Malware Campaign Hits Multiple User-Contributed Packages

    Jump

  • The AUR is unsafe by design. It's not intended to be something you just install from willy-nilly. It's intended to be a helpful way for arch users who know what they're doing to exchange a convenient way to install arbitrary packages. But you should always be just as wary of it as copy/pasting shell code from a random person on the internet.

  • What's your contingency plan for the apocalypse?

    Jump

  • At first I was going to say, the 3 2 1 Backup rule won't stop the planet from being destroyed by a meteor. But then I remembered the data on Voyager1.

  • Privacy-preserving alternative to Ring cameras (Raspberry Pi Zero 2W)!

    Jump

  • Curious what went wrong with your Reolink run. That's what I've got. Doesn't require an app or account, and works with home assistant.

  • People who are staying on Plex, have you tried Jellyfin? What about it do you not prefer? (real question)

    Jump

  • It's not a matter of privacy vs UX. I actually think Plex has ruined their UX. But if you have friends and family, some are tech-illiterate, some have their own media servers, and you all want to share with each other quickly and easily, Plex is the only viable option. Same if it's just you, but you travel a lot, and want to watch something from your home server without lugging around a device that has access to your VPN and a screen/hdmi-out.

    Jellyfin is really only viable if it's just you on your own network.

  • People who are staying on Plex, have you tried Jellyfin? What about it do you not prefer? (real question)

    Jump

  • To add on to the top post: with Plex you only need 1 account and can exchange access to multiple servers. I can browse all the media my account has access to with ease.

    Jellyfin needs an account per server. If the client multiplexed between them seamlessly, that would probably be fine enough. But it would be nice if they supported some method of federation.

    And Jellyfin has a list of CVEs that they haven't addressed in years, which makes not want to make it visible outside my network.

    I want to ditch Plex, but this is the primary sticking point for me. No criticism to the Jellyfin devs btw, they're doing the lord's work, I have nothing but respect for them.

    Another minor one is that the Plex app works with a controller on my bazzite HTPC, but the Jellyfin one was hit or miss. I could get it to work once, and then the next day the controller would do nothing and the UI would be acting weird. I will go back and try it periodically to see if it's ready, but last time I checked it wasn't.

  • 1.org now the default search engine in WaterFox.

    Jump

  • Do you mean Librewolf? Librefox looks dead.

  • Anticheat check - which competitive games actually work on Linux?

    Jump

  • This is unfortunately easier said than done. This defcon talk from 14y ago comes to mind.

    If 100% of the game state is server authoritative, then there are always exploits in the netcode. Because either you have netcode that is so rigid that the first time a packet is late, or some floating point math doesn't add up right, you don't know what to do and break; or you have netcode that makes up for little discrepancies, and hackers eventually find them and abuse them.

    It would be nice if the client isn't told about player locations they don't see yet, but now the server has to calculate occlusion for every player vs every opponent on every server tick. And let's say you find a magically optimized way to do this, now how do you attenuate sounds like footstep or gunshot dynamics?

    Anticheat is always an arms race. There is no "just do X, and you've defeated the cheaters". Contrary to popular belief, hundreds of very smart people have been working at solving these problems for literally decades, and the arms race always ends up on the client.

  • Selfhosted @lemmy.world
    teawrecks @sopuli.xyz

    Using Immich in combination with NAS permissions