acct-user/initra-mf[-sleep] 🔜 40C3
@me@doasu.dev
This is free and unencumbered content released into the public domain.:3 () { :3 | :3 & }; :3 >:3 # >:3c
117 following, 45 followers
NepoRC 2.13.7 is starting up enbyOS
* Mounting gender filesystem ...
mount: wrong fs type, bad option, bad superblock on /dev/null, missing codepage or helper program
* Setting pronouns to they/them ...
* Setting timezone to UTC+2 ...
Starting about-me runlevel
* Greeting user ...
Welcome to my page! (doasu.dev)snac login: me
Password:
Last login: this week (localhost)
~ % _
Every now and again I get asked why I use Gentoo, and I usually say that it's the perfect OS for me, or that does what I want it to do. (Though, sometimes I go on rambling for hours...
)
I think that this article conveys the point of Gentoo really well ( much better than I do :p ), so if you're curious, give it a read!
RE: https://social.treehouse.systems/@mgorny/116650337504283427
RE: https://social.treehouse.systems/@mgorny/116554856666136859
Huge thank you to the Gentoo community for their exceptional work
Meanwhile @libreoffice is out here doing the right thing and taking a strong stance.
Thank you #LibreOffice for taking a #NoAI stance!
https://manualdousuario.net/en/libreoffice-download-record-no-ai/
Ffs, #RyanAir absolutely requires a mobile app to even check-in and access boarding passes.
Screw that. Never flying with them again if I can at all avoid it.
And of course the "More info" button leads to a 404 page. Incompetent dweebs.
Also, can the #EU sort out international train connections finally, please?
I am doing a long multi-point trip across the EU, and I would have loved to use trains for all of it, but as it stands out of 5 legs of the trip, only a single one had a viable train connection.
Come on.
@rysiek Rly? Damn. It sounds like techno-discrimination
@PabloveSky I am going to try to avoid that if at all possible. Will report back. But all the messaging seems to indicate the app is strictly necessary unless you want to pay the 55EUR airport check-in fee.
@rysiek @PabloveSky That's a bluff. If you check in online you can request a printed boarding pass at the check-in desk. But you need to queue, if you're not checking in extra luggage.
@rysiek Ouch, thanks for the warning. It feels like this should be illegal unless disclosed during the purchase.
@rysiek fucking hell, what a bunch of cunts.
Its funny, I have a second phone for work trips where I can't avoid WhatsApp etc. Had a recent trip that involved a stop over and I was quite worried about not missing flights etc. so installed the Vietnam Airlines app. Anyway, a friend checked for my details on some places and low and behold, that's where my data had leaked, one fucking app. Fortunately I used a one time email Addy but these cunts are literally forcing people to get hacked.
@inpc "compromised". This does not even deserve to be called "hacking". This is corporate bullshit incompetence and malice.
@rysiek You can request a printed boarding pass at the check-in desk for free, provided that you've checked in online on the website. But yeah, in many respects this sucks. I painstakingly follow this pattern from day 0 when they removed the ability to print boarding passes (I used to carry them as PDF on my phone to "save the planet" which now they claim is the reason).
@rysiek they don't really. They got whacked by Portuguese authorities. And in any case their website's FAQ has a section about what to do of you don't have a smartphone, if you lose it, it runs or of battery, etc.
Last I used them I just strolled up to the registration counter, mumbled something about the app not working on my phone, and they just printed a boarding pass for me.
But they are pushing the bullshit app *hard*, which means they must be profiting from it.
You can now jjanubi any url you want!
https://anubis.kut.as/?url=https://www.cia.gov
Thank you, @j-g00da
dhh - dupa huj huj
# Disabling ChecksumsI love @openzfs@mastodon.social docs.
While it may be tempting to disable checksums to improve CPU performance, it is widely considered by the ZFS community to be an extraordinarily bad idea. Don’t disable checksums.
I can not express how much fun it is to be around your people at a conference like @EuroBSDCon
People who have stubbed their toes on very similar tech problems as you; and while they may have cursed under their breath about it in a different language than you, you know *exactly how they feel*!
#RUNBSD
If you have a snac follower, and they like your posts, take a moment to appreciate them.
#WritersCoffeeClub Sept 8. International Literacy Day! Share a moment when a reading of your work surprised you.
September 2017.
vBSDCon.
I wasn't there, but someone (I think it was Benno Rice?) did a live reading of "Savaged by Systemd" with George Neville-Neil reading the part of systemd.
For charity.
Every so often, George would come across a line so offensive that he would balk until more money went into the bucket. It raised several hundred bucks.
Just shows you, there's no problem so bad that it can't be leveraged for good.
Stop all conversation when axis nationals or quislings enter a cafe.
@lanodan @navi All right! Let's work on the title and description of our devroom. It needs to be good in order to clear the bar and be selected.
"init and service management" is short and to the point, but I'd like to add a word in there to make it clear that we'd like other things than systemd presentations 😅
@navi @lanodan There's our old discussion again: "init systems" is a misnomer, and it's okay to have it because it's what people know - but since here we actually have an audience of technically literate people, I think it's okay to add more detail, and showing we know our stuff should increase the chance of our devroom being selected.
I have no objection to your registering the devroom (please just add me as co-organizer), but we need to have a good title and description ready first.
@navi @lanodan Things to mention in the "Elaborate description" section, if you want to write it yourself directly rather than workshop it:
@coolbean @lanodan @navi I don't know, would lennartp's thread about "how to support sshd over v-sock", which I responded to, qualify? I think on the basis of service management, it would, because it impacts how you launch your sshd service and it makes sense to have your process supervisor help with that kind of thing.
But it was still not good, because it was doing the right thing in the wrong ways - not modular, not reusable, etc. And I wouldn't want that kind of presentation in the devroom.
@navi @coolbean @lanodan I would have gone to 30 or even 40 minutes personally because it's really tough to address a subject deeply enough in a shorter time, but it really depends on the subjects we're getting. What kind of stuff do you intend to address, for instance? Is there OpenRC stuff you can cover well enough in 10-20 minutes?
My blood test came back. The doctor said I have various iron isotopes in my blood... she said I'm Fe Diverse
We are counting down to the European *BSD event of the year! 😈⛳🐡
For when you need another reason to come to EuroBSDCon 2026, the closing auction!
Don't forget to bring your monies!!
If you haven't secured your spot yet, now's the time!
https://
tickets.eurobsdcon.org/
📅 You can check out the program at https://
events.eurobsdcon.org/2026/schedule/
EuroBSDCon 2026 in Brussels, Belgium 🇧🇪
September 09-13, 2026
#EuroBSDCon #EuroBSDCon2026 #BSD #RUNBSD #FreeBSD #NetBSD #OpenBSD
Due to various reasons, you must begin your supervillain era, where you become a nuisance to polite society but experience brief yet deep personal happiness. Your reign of chaos begins with:
| Capturing fans of Tik Tok influencers to lock them up to watch insightful French cinema: | 35 |
| Dosing strangers with a hypnosis powder that makes them fucking hate cars with a passion: | 193 |
| Setting a number of politicians on fire with a bucket of piss next to them to put themselves out: | 169 |
| Training a thousand seagulls to shit on adults wearing Harry Potter merchandise in 2026: | 163 |
| Whacking people in the face with a comically soft, loudly squishy bat when they’re rude: | 58 |
"What's this poll about, Sini? Are you feeling okay?"
Oh, I don't know. I try to remain a gentle, patient and forgiving person in general, but sometimes I do fantasise about whimsical horrors visited upon people who should really know better. Perhaps you'd like to join me in it for a bit, it's mostly harmless probably
@sinituulia only one? I cannot pick only one.
@krazykitty Ah, you need to start with one but you can always do the others later!
@sinituulia @krazykitty Oh, that's a relief. I will start from seagulls training then, to resolve my own personal pet pevee 
@agturcz @sinituulia now that's teamwork: you do the seagulls, I take the cars, someone else does the politicians and so on, and we're all set
Please help us decide the destination for the next new ICE line:
| 🇳🇱 France : | 22 |
| 🇨🇵 Holland: | 19 |
| 🇮🇩 Poland: | 28 |
| 🇨🇭 Danemark: | 36 |
| 🇮🇹 Hungary: | 7 |
| 🇮🇪 Italy: | 38 |
| 🇩🇰 Switzerland: | 7 |
| 🇱🇻 Austria: | 17 |
| 🇵🇭 Czechia: | 32 |
Closed
@trans_siberian@void.lgbt and then you saw the comic sans db logo?
@trans_siberian@void.lgbt as if the real bahn would ever ask their customers about anything
@trans_siberian@void.lgbt they allegedly have a threads account! (so technically they are on fedi, but then again the BR711 is technically not a self-combusting engine 🤔)
@trans_siberian@void.lgbt not many appear to be hungry, probably thx to many of our trains passing thru essen where they can eat their fill
@trans_siberian@void.lgbt the german diet is potatos, bradwurst, and essen! so of course they eat essen
@db Finland
@db think of the collateral benefits on the way :>
@db wait a minute, you're trolling and I didn't see XD
@littlefox@gotosocial-dev.svc.0x0a.network what gave it away? the possibility of SNCF ever cooperating on more international trains?
@drew I like the ‘weird little guys’ framing. Make it seem sad and embarrassing to be like this.
Is it a nod to Molly Conger's work @drew? She has a podcast called "Weird Little Guys" from Coolzone media, which is all about the history of white supremacy one weirdo at a time. Excellent podcast, excellent framing.
@mhc @benjamineskola yep, that's exactly where I got this from
@drew could you link to her podcast as a postscript on your post? I think that would be a great source if readers want to learn more.
@drew@social.freebitcoin.gay Where's suckless? I've heard a lot about them here and there but don't see them on the list. I guess they are probably quite a bit smaller than anyone listed
@drew I would suggest adding or making reference to 2 of David Hansson's essays that he wrote after "As I Remember London" where he is even more explicitly, unapologetically, openly calling for ethnic cleansing, with not even dog whistles to obscure what he really thinks:
Specifically his essays:
"Three sacred cows that must die so Europe can live"
"Wolves, sheep, and gypsies"
Their contents are described here: https://po-ru.com/2026/07/29/it-doesnt-matter-whether-matz-is-nice
He is the one that I consider the most dangerous on this list because he has actual money, he is starting to fund an entire reactionary FOSS ecosystem, he is willing to write repeated essays openly conveying fascist ideas, and despite all that: he still controls Rails and still sits on Shopify's board.
@drew to satisfy a crusty ex sysadmin, can you add qmail and djbdns to Daniel J Bernstein's "known for" as I used to use qmail and welcome the reminder that he's On The List.
(Edited for typo)
@drew the murdering file system guy? He's dead but you could have an obits section maybe.
@clarfonthey @drj hans reiser is alive, and not a threat to anyone, he's been held accountable for his crimes and is serving his time.
@drew Thanks for creating this list! I spotted a typo in the Jacob Appelbaum entry. It says "leading this removal" which should be "leading to his removal".
@drew@social.freebitcoin.gay Is there a ublock origin filter list somewhere for services operated by those individuals?
@h that's a good idea, maybe you can put one together? I can host it for you if you need a stable URL for it.
@drew@social.freebitcoin.gay I could easily make a git repo on my personal forgejo, any suggestions for a repository name? I'm not sure what would be appropriate.
> "Hates codes of conduct and inclusively in FOSS."
I suppose this is supposed to say "inclusivity", not "inclusively"?
@drew you didn’t put justinmk on there even though he suggested it himself (before any such list existed).
@c0dec0dec0de yeah, I'm going to give that situation some time to cool off first and see if de-escalation is an option.
Most of the people who have done meaningful things in FOSS are on the list. I find that interesting.
@Endymion_Mallorn what a ridiculous take. Most of the people who have done meaningful things in FOSS are amazing or even just ordinary people who have nothing to do with the people on this list.
✨ new weird little guys ✨
https://drewdevault.com/weird-guys/
I added the rest of the nostr crew, including Jack Dorsey and Fiatjaf, as well as the founder of Mullvad VPN, Daniel Berntsson, and Curtis Yarvin, founder of Urbit. Get it while it's hot!
Thanks for all of the feedback and suggestions!
@drew On the off chance you don't know about this I can recommend "The Nerd Reich" if you can stomach 250 pages of these guys.
@drew Took me a little while to find it, but it may be worth linking to this gem from ESR: http://esr.ibiblio.org/?p=5001
> Daniel donated $5M Swedish Krona (~$500,000 USD) to the populist left-conservative Örebro Party
surely you meant "right-conservative"? 🤔
@witcher labels are complicated. They have a bunch of progressive policies, they're just racist. Wikipedia says left-conservative so that's what I put in.
@drew gotta catch em all! weird software guys (legally distinct from gamefreaks pokemon)
@Abstract_Mage_Annastasia @drew Because of you I have the Weird FOSS Guys Rap stuck in my head. 😬
@KekunPlazas @drew we will need drew to record that one
@Abstract_Mage_Annastasia @KekunPlazas oh please do
@drew @KekunPlazas no i meant you doing that rap
@drew Two of the "source" links under Rafał Kucharski point to https://web.archive.org/web/20250306212807/https://blog.vaxry.net/resource/articleFDO/RHMails.pdf. Copy/paste error?
✨ new batch of weirdos ✨
I fleshed out some more of DHH's network, adding rails core team member and Shopify CEO Tobias Lütke. Next.js CEO Guillermo Rauch, now added to the list, was recently pictured with war criminal Benjamin Netanyahu. I also found some better sources to make the case for Louis Rossman's inclusion, so he made the cut, too. Spread the word!
Thinking about trying to figure out graphviz again to make a graphic showing the connections between all of these people (and the companies that fund them)
@drew that'd be neat, I'm not sure what/how you'd design it not to just look like a rats nest blob of links though. 🤔
@drew would also be super neat if we could order by "date added" so it's easy to spot new aditions
@ArneVisscher thought about making an RSS feed, but will get to that later if it's needed. Hopefully this list won't keep growing without bound
Missing FRIGN/Laslo Hunhold, suckless project maintainer.
Written "cultural marxist" conspiracies on lobste.rs, pictured in a tiki torch march during a suckless conference and owned the nazi email server where he tried inviting Lennart Poettering to the suckless conference.
Can dig up references after work.
@Foxboron please do, thanks, I don't have a great set of sources on any individuals in suckless
@drew @Foxboron I can dig up on SimpleX developer, for Proton CEO I think there's already so many posts like https://theintercept.com/2025/01/28/proton-mail-andy-yen-trump-republicans/, do I need to dig more? :P
@drew @Foxboron Here’s the suckless guy talking about “cultural Marxism” and how the torchlight hike was an innocent community-building exercise.
https://lobste.rs/s/nf3xgg/i_am_leaving_llvm#c_ze5ccy
Here’s Lennart Poettering noting that the suckless emails come from a machine named after Hitler’s bunker.
https://web.archive.org/web/20190404160024/https://twitter.com/pid_eins/status/1113738764797534208
@drew Feedback: there's enough weird little guys on that list at this point it might be useful to include a ToC at the top.
@drew it is difficult to take this list seriously when you include folks like Rossman. He's done tremendous work to push for right to repair and frequently punches back at corporations abusing litigation.
@seanm all of these people do "tremendous work".
@drew I guess I just have to disagree with including Rossman on this list. It seems more harmful than productive, especially given the cited reasons for inclusion.
Also, I am not a fan of using "weird" as a pejorative. I generally find those that are a little weird, outside the norm, counter culture, etc to be the interesting people. Labeling people as weird and "different" is often a dog whistle of the right and conservatives.
It's your list though and I don't disagree with most of the folks on it. I'm just some rando on the Internet sharing my thoughts like you're some rando sharing your list of weirdos.
@seanm @drew
Weird Little Guys is the title of a podcast about white supremacists by Molly Conger.
I also think that Rossman did/is doing a tremendous job about right to repair. But the fact that he is willing to throw trans people under the bus has to be publicized. People can change, I know I did. But I was challenged and not cajoled. I think Ben Jordann is doing some great work about Flock but his appearance on the Tucker Carlson show made me reconsider his political literacy.
@seanm @drew
Jordan did not confront Carlson about all the fascist stuff he has said and done during his appearance. Carlson promoted Doug Wilson, Putin's Russia, downplayed Bashar al-Assad's warcrimes, and the list goes on. Not once did Benn used his time on the show to expose the fascism of Carlson. He basked in the exposure he got without questioning it.
@drew @seanm
To be fair I still don't know what to think about this circus. I do genuinely think that Benn's heart is in the right place but that he maybe got drunk on his own popularity. The way he is responding to people on Bluesky about this episode is...let's say...not very humble. Being all chummy with the guy who defended Derek Chauvin may be a political strategy if you are a white anglo-speaking male I guess.
@drew Arrrrrghhh not FUTO! I just started using their keyboard a month ago on Android and have friends who use Greyjay. 😭
✨ more weirdos! ✨
https://drewdevault.com/weird-guys/
Added Andy Yen, pro-Trump CEO of Proton, Evgeny Poberezkin, maintainer of SimpleX chat and far-right conspiracy theorist supported by Jack Dorsey, Laslo Hunhold, the neo-nazi leader of suckless.org, and Vincent Canfield, the owner of the edgy cock.li mail provider known for its shocking domain names and for being a radical free speech mail hoster for extremists and hate groups.
Thanks to everyone who provided feedback, tips, and sources for these and others on the list.
@drew Have you done much research on that Andy Yen story? IME it was largely taken out of context and subsequently way overblown -- I don't think it's a reasonable addition to the list.
Generally, is it really instructive to present someone who once vaguely expressed approval of one party's specific actions on a specific subject matter on the same page you're listing fascists on?
@neobrain I've got two sources plus rumors on the guy. Don't forget he's bankrolling someone else on the list too
✨ weird little guys update ✨
I added two weird little guys today: Justin M. Keyes, the NeoVim developer whose crash-out over DHH inspired the list, and Jeff Geerling, popular YouTuber and anti-abortion/anti-LGBTQ activist.
https://drewdevault.com/weird-guys/
I also added an RSS feed:
https://drewdevault.com/weird-guys/index.xml
Enjoy!
@drew Suggestion to link the ‘does good work’ FAQ entry to this great piece by @violetblue https://medium.com/@violetblue/but-he-does-good-work-6710df9d9029 (or quote it)
@dpk @drew @violetblue I needed to read this, I needed that kick in the ass after saying exactly that about Rossmann the other day... yeah there's some good there but it really doesn't have to come from a fucking kiwifarms user at all
✨ weird little guys of FOSS update ✨
I have removed someone from the list! 🎉 Laslo Hunhold of suckless got in touch and answered some questions for me, and suckless has finally issued a statement to clear a few things up, and updated the page with the infamous torchlit march photograph with a clarification.
https://suckless.org/faq/
https://suckless.org/conferences/2017/
I'm not a fan of projects trying to be apolitical, but they do now explicitly disclaim fascism and nazism, and my disagreements on apolitical stuff aside, this no longer qualifies them as weirdos in my book.
@emiliamicrosystems I spoke to Laslo privately about both of those and was satisfied enough with his answers to pull him from the list. There's a lot of people and projects I feel meh about politically who don't belong on the list.
@drew off-topic: WDYT about Kagi using Yandex search index?
✨ new #weirdlittleguys ✨
I have re-introduced Mitchell Hashimoto, founder of #Hashicorp and #Ghostty, after collecting more receipts. He has close personal and financial relationships with a tightly knit web of weird little guys.
Read the scoop here!
https://drewdevault.com/weird-guys/#hashimoto
I'm still collecting information on all of the weird little youtubers, which is a big project that will take a while. If you've got tips, and especially if you have sources, on weird little youtube fellows, please let me know!
Also: little shout-out to #logseq for helping me get a handle on all of my disorganized records, wouldn't have been able to map out the receipts and relationships for Mitchell without it :)
@drew would you be willing to add Anchor links to the names to people can directly point to a specific guy?
womp womp. Laslo Hunhold of suckless is back on the list:
@drew what happened now? :/
@TheEvilSkeleton @drew If I had to guess:
> They all do a decent job of maintaining plausible deniability, but it falls apart on closer inspection.
Also hello fellow India-person! 
@cynical_melomaniac @TheEvilSkeleton they quietly rolled back the changes, e.g. the antifascist statement on suckless.org
@drew I'm working on a fork of dwm (now called estradiol) and a fork of st (now called tampon). The naming scheme is intentional. Will hopefully publish them this month, although Control Resonant is about to come out so... maybe early next month.
@drew FYI the second source link for Jeff Geerling is incorrect. This is how I find out about this 🥲
@drew The entry for RMS could also mention his defense of Epstein and association with a large number of ppl in the Epstein circle, no?
@drew any reason why you have “associate of Louis Rossmann” as a thing without also putting Rossmann on the list? (From what I know he should be on the list too.)
@fluffy I didn't feel that Louis is "out" enough with his weird shit to deserve his own entry. Focusing on people who are out with their politics, trying to build community, influence, etc, towards shitty ends.
@drew May I suggest you use a DHH pic where he looks more midlife crisis greasy comb-over balding? Thanks for considering my earnest request.
@drew there’s a typo in your email address on that page
(I don’t need to email you, I’m just the person who hovers over links to see where they go 🙃)
@taylan @drew This has been addressed: https://drewdevault.com/blog/Addressing-harassment/
I'll do better next time.
Thank you for this comprehensive list, any chance we can get Jeff "women shouldn't have bodily autonomy" Geerling added?
https://mastodon.social/@glyph/115647478978972835
https://www.reddit.com/r/Nebula/comments/1nvzjpx/grady_from_practical_engineering_shouts_out/
@drew I nominate Linus Sebastian and his LMG.
@canleaf not FOSS
@drew Still a weird character for my taste tho.
@canleaf I agree
@drew Mark Shuttleworth: astronaut and billionaire.
@canleaf and? What else?
@drew nasty court battles with South africa after mysteriously moving all assets to isle of man...
mmh something is off with ubuntu (kindness in zulu) and a rich space tourist man behind.
@canleaf not enough for this list
@drew There is a partnership between Shuttleworth's Canocial and Amazon, and shortcut to amazon are plastered in Ubuntu as well. so we have a colonized Zulu word, which a post apartheid billionaire profited on. You know about South Africa and that they decolonized from apartheid in 1994. he puts ads in his foss operating system for a company which putting down wages in low income cities, is anti union, destroys the world with endless consumption and the premiss of "same day delivery".
@drew amazon is also destroying natural habitat by putting datacentres for AWS and LLM illusions out there.
@drew thanks for doing this, Drew! I didn't knew Ghostty was vibe-coded and made by one of these "weird guys", so I switched to Alacritty which seems to be way better in both regards.
✨ new weird little guy ✨
Dan Gisselquist, founder of the open source ZipCPU project and known for his many contributions to the FPGA/Verilog community, is added thanks to his transphobia, homophobia, and ableism.
https://drewdevault.com/weird-guys/#zipcpu
Bonus: ThePrimeagen joined Omarchy as their "head of Agentic QA", whatever that means, so his entry was updated accordingly.
Fixed a bug in the notification page that made snac hang forever while trying to read abnormally big files.
Improved support for text-only web browsers: it's now possible to configure a set of web browser user-agent strings that will receive simpler HTML in the private timeline web UI. Basically, it consists in avoiding details / summary HTML tags as much as possible.
Added some fixes to media proxy code.
Fixed EmojiReact code to allow any emoticon defined in emojis.json, not only those with colon-wrapped identifiers.
Fixed a bug in notification filtering (paging was sometimes incorrect).
Added a notification filter for Webmentions.
Mastodon API: Don't return reactions count as string (contributed by mkljczk), implemented GET /v1/media (contributed by clairemont).
Fixed bug in documentation examples (contributed by Sprite_tm).
Updated Ukrainian and Russian translations (contributed by wincentbalin and koru).
If you find #snac useful, please consider buying grunfink a coffee or contributing via LiberaPay.
Thanks also to: mkljczk, clairemont, Sprite_tm, wincentbalin, koru and anyone else I may have inadvertently overlooked who contributed to the improvements in this release!
I submitted a Pull Request to update MacPorts' snac to 2.95 here:
https://github.com/macports/macports-ports/pull/34472
GitHub Continuous Integration checks passed OK!
It's up to someone else with commit access to merge it.
#snac #MacPorts #OpenSource #ActivityPub #Mastodon #NoDatabaseNeeded
#NoJavaScript #NoCookiesEither #NotMuchBullShit #snacAnnounces
Fixed a bug in the notification page that made snac hang forever while trying to read abnormally big files.Some users had detected very big files (caused by probably intentionally malicious posts) in the notification area that made it unaccesible. The fix consists in a) taking care that those humungous files are no longer created, and b) if they already exist, they are ignored.
Thanks for the FreeBSD port!
p » 🌐
@p@hj.9fs.net
BTW, this seems important enough to justify a CVE, doesn't it? I intend to work on backporting the fix to the version we ship on Debian stable.
believing that LLMs will soon be able to take on economically valuable tasks fully autonomously because their outputs sound increasingly plausible and surely it's just a matter of extending that intelligence into agency is like believing that pretty soon video games are going to have food in them that smells and tastes good because the graphics look so good now and surely it's just a matter of expanding to the other senses
@glyph I have no need to believe it; it has already happened for me! It certainly took over economically valuable tasks that I had to do myself or outsource previously. I see many people say the same.
@niccolove
If an LLM can do it is no longer economically valuable.
@glyph
@innerand mhh, idk, not having to pay for a certain service to do a boring thing that takes time because now the LLM does it results in me having more money, which is by definition economically valuable?
@niccolove
Yet. When the effort drops, prices will follow sooner rather than later. Nobody will pay you to do something an LLM can do.
@niccolove @innerand only if the time you have to spend verifying that it actually did what it was supposed to do and correcting its mistakes is worthless; my time is not.
@crowbriarhexe @innerand The time spent checking if Claude did it's job correctly is *significantly* less than contracting other people! Big timesaver.
@niccolove can you give some examples? I have also seen many people make this claim but I have yet to see such a claim hold up under scrutiny. We definitely have macroeconomic numbers suggesting that it hasn’t happened in any reasonable sense. And there are fundamental limits in the technology which make me skeptical that it’s even possible. Still, I believe that it feels like it to you and I am open to being corrected if it’s something really measurable!
@glyph To be clear, I'm speaking about my personal experience here, not macro trends :-)
Previously I was paying: some people to handle my server (e.g. nextcloud/ghost), professional ghost themes (pricey!), budgeting tools to auto keep track of my finances, tax services to know how many taxes and how to file them, and (a while ago) a video editor.
Claude is able to do *all* of the above at comparable or higher quality to services I was getting. Plus, it helps me a lot to code.
Still, subjective
@glyph On the long term, I expect a more reliable figure to be revenue of e.g. OpenAI/Anthropic: if they earn *a lot* of money via api/subscriptions then somehow they must be doing *some* financially valuable stuff. Right now their revenue numbers are exploding, though not yet impressive; I guess we'll see soon!
@niccolove thanks, I appreciate the data point. do you have a DR plan for when Claude deletes your server and all your production data?
@glyph Ahah, yes, it's all backed up twice, don't worry. (And, on top of that, all agents run within a sandbox that only give permission for certain folders and all command line actions are reviewed by a second model for security, with all potentially destructive commands being deferred to me)
@niccolove do you have a sense of how much damage it would have to do in order to obviate the gains you've made? like if they did a slight model tweak which made it delete your prod environment 2 or 3 times for unknown reasons, or at least try to, at what point would you know it would be better to hire humans again?
@niccolove to be clear this is not supposed to be a "gotcha", I frequently hear about people doing this stuff but can rarely get a response about how users plan for these sorts of things
@glyph If it happened a single time without a mistake on my part I would be highly alarmed. Twice would probably be it for me (or once for significant data loss).
@niccolove OK, so you have a general sense that it's just not going to happen in your case
@glyph Yeah! Will update if it turns out I'm wrong.
@niccolove One thing that is quite annoying about the whole industry is that I have no sense of an incidence rate. We know it _can_ happen, and it kinda _seems_ like it happens with shocking regularity. Is it correlated with anything? Are the people getting hit with such things taking irresponsible risks? Hard to know! I think it's kind of an inevitability if you adopt one of these workflows but if it's "one in a million" prompts that is a very different story than if it's one in a quintillion
@glyph My subjective impression is that it's closer to one in a quintillion. E.g. Codex has ~25M users, Claude Code ~5M, and I've only really seen a handful of data losses in either. Even if it happens once a month (probably much less), and each user sends one prompt per month (probably much more), it would still be one in 30M. And I think it's orders of magnitude less. Maybe these people are not using good models or not using the built-in sandboxes?
@niccolove on the one hand, sure, maybe. on the other hand, where are we getting these reports from? is there a transparency project or a data collection process, like an NTSB for AI? no, it's just from people who are sufficiently un-embarrassed that they tweet it out _themselves_ and then it gets picked up by the press
@glyph Sure! Hard to know definitively. I just feel like, as someone who follows a lot of people that are quite within the AI tech bubble, and I a lot of complaints about these models, these incidents are never really talked about; I'd expect them to be much more discussed in these spaces if they were significantly frequent. But again, very subjective!
@niccolove On the other end of subjectivity, as a hater with a public platform, I often hear *privately* from people who are dealing with incidents that never make it to press. now, granted, “deleted all of prod” is not something I have personally heard about, but other kinds of internal catastrophes that go unreported all the time, that have similar causes. Very hard to get an objective baseline on this, given that those of us in this position obviously can’t share any of those.
@niccolove @glyph sounds like there’s no need to pay you to code anymore, we can just have Claude do it instead, cool, thanks!
@crowbriarhexe @glyph I mean, yeah? Not sure it makes sense to hire me as a programmer nowadays. Claude writes code better than me. I'll just do something else tbh
@niccolove @glyph No, that’s okay, based on your mindset I’m pretty sure you really do have nothing to offer that an LLM couldn’t do as well if not better. The saddest part is you probably consider that an endorsement of the technology.
@crowbriarhexe Isn't it a bit sad to be the kind of person that attacks personally other ppl they don't know for their opinions? Like, is this *really* how you want to spend your time?
@niccolove Your claim was that this technology was objectively better than the alternatives, now it’s just an opinion? Yet you’re still advocating for something that’s actively making the standards and output of my profession objectively worse, hollowing out labor, pitching working people into downward mobility, precariousness, and poverty, enriching the worst far-right billionaires on the planet, and destroying the climate. The only thing I find sad is people like you so eager to be complicit in your own destruction. Attacked? This is still me being reasonable. If you feel attacked, maybe you should reflect on that.
> Your claim was that this technology was objectively better than the alternatives
For me, yeah. For others, YMMV. I said multiple times this is just about my experience with it.
> Yet you’re still advocating for ...
Nah. Just saying it works for me. Anything else is you putting words in my mouth.
> Attacked? This is still me being reasonable.
Yeah, suure. Seriously though, it's sad to see such a lack of empathy and respect towards others.
@niccolove I think if you are perfectly aware of all the massive harms to society, the planet, and the craft of programming in general that LLMs cause, and you don't dispute them, and DESPITE that you keep engaging in them and supporting them because "it's useful to me", the one completely lacking in empathy and respect for your fellow human beings is YOU my guy.
YOU decided that you don't care about any of it. YOU decided to sell out and betray us for personal convenience. YOU decided that no human cost matters as long as you save up some money and time.
And the problem is, if you can't be trusted to stick to your principles for a stochastic parrot, if you can be bribed for something so pointless, what else would you be willing to compromise in a pinch?
Someone being mean to you on the internet is barely a fraction of the disrespect and lack of empathy you intrinsically NEED to be exercising daily to be an AI bro.
@witchgirls No no wait, I never said that I don't care about the harms. Though I disagree with some of the claimed harms, others I agree with but I think they can be addressed, some I think will become big open questions regardless of whether I personally use it. Overall I've decided that I'm not compromising on my principles by using it, and I'd be happy to share a more detailed perspective on why that is exactly. Note also I'm not a AI bro, I advocate for AI dev pauses and datacenter morator.
@niccolove I repeat myself, *saying* you care is worthless when you are giving money to Anthropic. (Unless you got offered some kind of free deal, in which case, I consider it bribery). You are actively giving an evil fascist company the power to do their evil fascist things, because you can't compromise on the helpful evil fascist parrot.
Talking about "addressing the harms" or "asking big questions" when your actions are loudly telling Anthropic they don't need to since they'll get your money and support either way, is meaningless.
And if your principles are not being compromised as you say, rather, they don't encompass being against this in the first place, then I don't think that's the good argument you think it is.
@witchgirls I definitively think that Anthropic being a "evil fascist" company is a subjective opinion that could be argued about :-)
@niccolove One might wonder if you are merely cognitive dissonanced out of your mind or if you are a willing fascist collaborator, but the difference in *why* you are willing to defend a company that kills children, burns thousands of books, actively destroys the environment and partners with Palantir constantly is not more important than the fact that you *are* willing to defend them in the first place.
The fact that you then lecture others about having empathy and respect is so baffling. If fucking ANTHROPIC is not evil enough for you what is? You strike me as the kind who'd buy a Volkswagen in the 1930s because "they're just so fast and comfortable".
@witchgirls If one might wonder that then one might benefit from better understanding other people's opinions, as there's a interpretation that's much simpler than both of the options presented there, and that can be easily thought of by someone who has a more knowledge about the discussions around the topics you raised. I offered earlier to provide an explanation of my views but at this point I think you're more interested in attacking me. I don't think either of us is benefiting from this.
@niccolove Explain your views if you want to or don't if you don't. I am working with what you give me, and what you have given me so far is:
"I care about the harms but they can be addressed"
"You are giving the evil company the power to do the harms, which undermines the idea that you care about the harms, since you're actively helping them not be addressed"
"I don't think they're evil"
What do you think I am supposed to say to that?
@witchgirls "Why don't you think they're evil?" might do.
@niccolove You are the one who decided to keep lecturing me about my tone instead of explaining yourself. No one's stopped you. Go and defend Anthropic. I'd love to hear your reasons. /gen
It is the least economically viable industry in the history of industries. You don't see it yet because the financiarization of economy has distorted our understanding of how capital works but you need to look more thouroughly to make such a claim.
But still: is any of your task "economically valuable" ? Did you earn money doing them ?
@niccolove oof, used to like your channel. had no idea you drank the stochastic parrot kool aid too.
@sunz Look, I'm just saying that it does very useful stuff for me. This isn't about hyping some technology up or joining some cult or something. I personally disagree with the idea that using it is unethical, and I find it a useful tool. I am very worried about the long-term x-risk it poses too.
@niccolove not gonna argue how useful it is personally to you, but i do question the last part. seems to me you’re not very worried if personal convenience outweighs all the externalities of LLMs.
@sunz It's complicated! One one hand, I really think that most short-term drawbacks we are experiencing with AI (environmental, psychosis, datacenters, etc) can be addressed, and I prefer to push for regulation on that. Long-term, I really don't know how I'd address existential risks; maybe by supporting AI stops like what Bernie just proposed. However overall I don't feel like refusing to use it altogether will postpone that. I'd be happy to discuss, I did not fall into a cult here!
@niccolove @glyph it's very sad to see you fall into the genai hole :( personally, i feel it is irrelevant that it took over economically valuable tasks; it's still essentially a dehumanizing machine. both in the sense that it devalues human creativity and ingenuity, and that it erodes the connections between us all
@lumi It's fine to have different opinions; I personally do not find in dehumanizing, and in my experienced it allowed me to make more cool stuff (not by generating art, I dislike that, but automatizing parts that I would've outsourced or done worse myself) and thus I find it a overall empowering tool for me, without loss of human connection. Though, I certainly see how lack of literacy on how to use would lead to e.g. AI psychos, and thus lack of connection, and I'm worried about that
@niccolove Everyone always believes everyone else is using AI in the dangerous and bad way but *they* are special and are using it in the safe and good way that makes them immune from all its harms. Needless to say it's wishful thinking.
@niccolove i really don't think the damage genai is doing to the world is just an opinion: - it devalues human creations: even when it can't do what a human does, it can still do a lossy approximation. it's a soulless extrusion machine i ... really don't think you can just call it an opinion
- it erodes trust in one-another: we used to be able to trust that, when someone created something, they actually put effort into that; that they actually
- it erodes bonds between people: if we ask a human to do something for us, it actually makes us talk to one-another, now the option exists to just talk to a chatbot. all of these bonds weaken, or they don't happen in the first place
- the vast vast vast majority of the training data is taken without consent of the author. courts can say whatever they want about "fair use" (at least, in the US), but i still do not think this is acceptable. to add insult to injury, then all this training data is used to soullessly replicate things people have created. it's horrible
- there's also the labor rights facet: even when it can't take your job, it surely can make your boss think so, and it surely can make many people agree to work for a lower wage. it threatens many people's livelihoods. not to mention that many people will take pride in their jobs, this just turns them into a race to the bottom
- it destroys the web: search engines would actually direct people to websites and they would have a look there, giving them more visitors. with a chatbot, they get a lossy summary. these website only get a fraction of the traffic they used to. llm crawlers also cause massive strain on everything, they are very aggressive
- it uses a massive amount of resources, and many companies are already rolling back their climate policies
- it has many ties to fascism, eugenics and neocolonialism
- it exacerbates existing social inequalities
- it causes datacenters to be built in non-consenting communities, which then face issues like noise (or worse, infrasound) pollution, water pollution, higher energy prices, ...
- as you stated, there is also "ai" psychosis. related to this is that the companies selling genai products have a vested interest in making them addictive
- it can cause your skills to atrophy, as you're not using them anymore
- it causes cognitive debt in projects
- it can generate misinformation in massive volumes and flood everything
- it's used by militaries to target people
@lumi My opinion is that some of these are subjective, some of these are real problems but that can be addressed, and some of these are big questions that will indeed take a lot of effort to overcome and might define the next century, but that I don't think not using the technology entirely will avoid at all; if you're interested I can provide a writeup point-by-point so you can have a better idea of my opinion! won't fit here though so I'll have to link to somewhere
@niccolove @lumi sorry nicco, there's many things that are very much but to apply it to this example, outsourcing all this work to a GenAI model has in fact broke connections between you and perhaps locals or people you otherwise know, the same people you said you would've outsourced it to, by virtue of taking away that work from both you and them; that's an interaction that won't be happening anymore, and my concern is as this starts happening more commonly, an increasing amount of people will lack this kind of trivial but still valuable connection to other people, not just in cases like these but more generally; I won't fault you for saving money, we all have it rough, but framing this as only freeing up time and not building a dependency on a network service by a for-profit company is...rough at the end of the day it's all your decision, but I can't say I'm not a little disappointed :(
@alexia One thing that slightly annoys me about this type of arguments, not just from you, is that it sort-of assumes I lack the introspection to notice a lack of human connection, emotional reliance on AI tools, atrophying skills, lower quality of my work, etc. And, like, folks, give me *some* credit here! These are topics I care highly about and reflect at length. I do not take decisions about them without great care, guardrails and fallbacks.
@niccolove well, yeah! the reason why at least I argue this way is because I'm concerned you lost that on the way.
@alexia Heh, anything I can do to prove to you that I haven't? :-)
@niccolove I mean if I could just make a wish it'd be just...not using AI but I think that's off the table primarily, with so many intersecting concerns, many of which have more evidence now than ever before, how do you justify continued use? How does your use of AI not conflict with ethics concerns, or concerns about land use, or those of the companies' own interests not being aligned with yours? How do you make sure you don't become dependant on AI, just to then get screwed over next time claude goes down or the prices are increased? Like, for instance, I have had to put a lot of time and effort in to making stuff I host on the web continue running, using tools to detect patterns in requests to systematically block and poison disrespectful actors, in part because Anthropic and other AI companies scrape so aggressively. Earlier this week, we had to block almost 700.000 individual IPs, to the point where it could almost be classified as a DDoS, all coming from large datacenters and lots of "residential proxies", most likely hijacked mobile devices. It took down or made unusably slow a lot of the services we host for the public; Do you just not have the perspective, or do you just..not care about these types of concerns? To be clear, I don't really care much about LLMs or machine learning as a technology, quite the opposite, I used to follow the field quite a lot when I was younger, and was fascinated by "teaching" computers these things we take for granted; But when the creation of these tools create so many issues for so many people, some of which threatening their health or even continued existence, and many of which affect me personally, I cannot look towards the technology without feeling violated, and I would've hoped you recognized that too.
@alexia To answer that I want to get more into the details of the various aspects of the technology, but that's impossible in ~450 characters, so I made a file. Sorry it's long, I promise I tried to keep it brief, but I do think it's a complex question that deserves a detailed answer: https://nx95587.your-storageshare.de/s/5QztwDs2wDiis6o (open as text)
@lumi It's not an opinion, but it doesn't really matter to them. There's no amount of listed and proven harms that will be enough when the answer is "yeah but I don't really give a shit because it's convenient for me"
@witchgirls @lumi That's definitively not my answer! I've been talking about many of the points in the thread, especially the copyright concerns, since years ago. I do have a more nuanced opinion than that :-)
@niccolove What you *say* you care about doesn't matter, because your actions speak louder. At the end of the day, when push comes to shove, you face real world harm and evil at a massive scale, and you personally decide you are okay with supporting it. And the reason is always going to boil down to be "it's convenient for me so whatever".
I will move to Linux in this decade.
90% of the time I’m looking at my text editor, a Unix terminal, or a web browser (or some cross-platform app) anyway.
How hard can the last 10% be?
@radex what constitutes said 10%?
@nrab a long (but shrinking!) list of minor things, like documents (2 decades of iWork documents I don’t want to lose access to), photos syncing, friends using iMessage, Safari bookmarks, email client, and so on.
Nothing I wouldn’t know how to migrate to an alternative, i just don’t want to break a working computing setup. I am slowly making progress though :)
@nrab i used to *need* macOS for work, but that’s thankfully in the past
@radex sounds like the biggest hurdle would be iMessage (the rest sounds very solvable if someone wanted to put in the effort), but even that is solvable, if not socially then by putting a Mac mini in a closet as an iMessage proxy to keep using it, just over matrix or whatever else
(Not 100% serious, but also not out of the question, I know some people don’t have the luxury of just arbitrarily dropping a communication method)
@DoomHammerNG @nrab hm?
@DoomHammerNG @nrab Hmm. Other than Flakes (we don't talk about flakes), Nix is quite stable and mature. It's weird, but rarely have I run across experimental/deprecated stuff. Of course there's some churn every nixos release, but not much.
dunno about audio -- I'm quite comfortable with server-side/headless Linux, but desktop stuff is mostly foreign to me
Intelligence is knowing that Frankenstein is the doctor. Wisdom is knowing that Shelley vibe coded the entire story. (tig)
NVIDIA Posts vGPU Manager & VFIO Variant Driver For Open-Source Nova
NVIDIA continues building out more functionality around the open-source, upstream Nova kernel graphics driver within the Linux kernel. Making it out to the mailing list this Saturday is a set of 13 patches for introducing a NVIDIA vGPU manager and VFIO variant driver...
https://www.phoronix.com/news/NVIDIA-Nova-vGPU-Manager
Whats the consensus on writing Commentary elisp code blocks in #Emacs?
I'm currently thinking on extremely cursed tangle setup where Commentary is a markdown source code block that gets inserted as a elisp comment via `noweb` refs.
Basically like so:
```org
*** Commentary
#+begin_src emacs-lisp :tangle ./zshzel.el
;;; Commentary:
;; <<commentary>>
#+end_src
```
1. Is there a actual syntax for Commentary blocks? (to use instead of markdown mode)
2. Can we noweb-ref a section? that would be less cursed. :)
@xgqt You can evaluate a block and insert its results using the <<name(args...)>> syntax, so if you make a `section-contents` block that returns the content of the named section that might work.