Another treasure from my bookshelve. Simply Scheme. I love the fact that the authors created a library of Logo like procedures to go along with it. And it still works with Guile 3.x! So cool! 😎🤩☺️
Apparently Lufthansa has a new "cabin concept" called Allegris. (source seatmaps . com)
Let's break down the basic economy class feature: seating space.
The regular one on B789 (787-9):
32" pitch (that's the distance between two seat) and 3" recline (mean 3" if personal space invasion).
Now the new Allegris on B789 (787-9):
31" pitch (one inch less,, 2.54cm) and 5" recline (2 more inches of personal space invasion). In short 3 more inches (7.6cm) of incomfort for tall people like me.
And people wonder why I hate flying (in Economy). But simply put I'm tall and when I seat my knee already touch the front seat. And it gets worse over. Now when the person in front recline the seat, it hurts. I gave seen: 1. people doing violently (slaming the seat back) 2. people insisting they must when there is no room, and waiting I get to the washroom to do it so I can't seat back. Got scolded my Swiss crew for that.
And then airline charge a premium for "front" seat (bulkhead or emergency exit). They are less convenient like less storage, no bag on the floor, or screen that must be stowed away. But I guess they don't discount that inconvenience, but they do charge more money for the leg room. And even if the seats are empty, you can GFY, at least on Air Canada.
People in front should be allowed to recline their seats all the way. If you don't fit, you should take it up with the airline and the government to make laws guaranteeing minimums! (Please do!)
Marc Garneau (who just died) helped gut air passenger rights similar to this, and we've got a long way ahead to get some of them back and then even more!
@purpleidea you just did it. Pitting passenger against each other. Bit yet continue the selfishness. It’s not the tall passenger that rob the fro passenger that don’t care about who’s behind them.
Experts who reviewed the code for ProPublica found numerous and troubling flaws in the system, providing a disturbing glimpse into how the Trump administration is allowing artificial intelligence to guide critical cuts in services.
The inability to understand WHAT generative models are genuinely good at should be studied as its own cognitive bias network I stg. It's like people are looking for the places where we most obviously have existing solutions.
Reading the designs of these attempts reveals so many foolish mental models from the HUMAN that read just pure software bro to me: e.g., arbitrary limits that feel efficient or something (limiting # of characters considered by the model in the contract)
Very clear reading this, that the person creating this thinks AI can think. Chilling tbh to watch this deployed in this way over resources that impact lives.
I have been coming to believe that generative models (where in order to use them we also need to solve climate demands and ethics demands) in *pre-determined dialogue with strict rules and tests and other conventional approaches* might be the only way forward, much more to say on that, but like, this is the exact opposite essentially
It is instructive to realize so many people in this world -- AND I MEAN SOFTWARE BROS HERE -- didn't know what made computational work work in the first place, or what makes data analysis work, and therefore implicitly expect a bunch of the same rules to be followed with AI, it is wild.
And we still think "software engineer" is the title that gives you the authority here???????? Baffling. Sorry guys. Baffling. Our stereotypes about this field have truly created an unbreakable fail up model.
Another thread here, and then I will be treating my jetlag and techlag with tea and baking scones -- is how little we value the work of our own minds.
"Decide if the size of a contract is reasonable." An ENORMOUS sociocognitive task that human employees at the VA are probably incredibly good at. All that wisdom, tacit knowledge, years of experience, complex work of our human minds, they are just thinking we replace that instead of enable that. Shameful. So little curiosity about our own minds.
Reading this continues to explain to me (finally, and at devastating scale) why I was SO frustrated trying to be a scientist working with software engineers building edtech who would NEVER listen to behavioral science about any aspect of human behavior and instrumented the worst systems for ever actually learning about human behavior. Just constantly thinking we could magically derive answers to our questions. And this was well before AI was used in this way! The mental fragility was there.
I honestly find the technical details of LLMs quite interesting -- like I find *many* models quite interesting. But the mad interaction effects, the amplification of existing exploitable fragilities in our mental models, that is boggling too. I feel it is necessary to understand just how cruel our systems already were (how we deal with health, and concepts of efficiency in a society that won't take care of its children), as a core piece of this and who gets to use this and in what way.
With LibreOffice 24.8 close to end of life, all users are invited to update their free office suite to the latest release Berlin, 6 June 2025 – The Document Foundation is pleased to announce the release of LibreOffice 25.2.
Does it finally stop throwing banners at you when you first launch it? You know the ones "Get involved with" or "Donate to". I know how to do that, thank you. I've been using your product for a decade now. Stop it.
@soc @RandamuMaki We like feedback when it's constructive and gives arguments. Telling volunteers to just "stop it" when they're trying to sustain the future the project is not especially constructive – hope that makes sense.
@soc In an environment where you just want to open a file, work on it, and then go on with your life, having banners which need to be clicked away adds another process to the work. This is counter-intuitive, bothersome, and those who have the capability to help out will have already moved to help out. All you're doing is nagging the end user. Just like how xscreensaver's "your distro is doing you a disservice" message got removed, it's a matter of time before these banners are.
@RandamuMaki The banner appears once every six months. We think that's not a lot for something that people typically use every day. One click, and you won't see it again for half a year.
You say "those who have the capability to help out will have already moved to help out" – yet we've had an increase in donations since we added the once-every-half-year banner, so your claim is simply not true.
Donations are vital to keep LibreOffice strong and provide the world with a FOSS office suite.
I might just have the misfortune of seeing it more frequently as I have to reinstall and use libreoffice on multiple computers on the regular. If it saves that timestamp somewhere this might get wiped in the process. It just seems to appear every other time I open it, which grates me.
@RandamuMaki Understood – that could be annoying. We'll check to see if the timestamp is saved in the profile (so that it persists over upgrades). Thanks for the feedback!
LibreOffice is my absolute go-to office suite! Switched from MS Office years ago and haven't looked back. Hope to see an even more modern UI soon. Keep up the fantastic work!
Re: last boost (mathstodon.xyz/@j_bertolotti/1…), about the researchers who trained a model trying to use only public-domain and openly licensed data, here's the paper: arxiv.org/abs/2506.05209 and here's the HuggingFace for the comma-v0.1-2t model: huggingface.co/common-pile (For anyone itching to play with it, no, it's not on Ollama, at least not yet; it's brand new.)
Large language models (LLMs) are typically trained on enormous quantities of unlicensed text, a practice that has led to scrutiny due to possible intellectual property infringement and ethical concerns.
"AI companies claim their tools couldn't exist without training on copyrighted material. It turns out, they could — it's just really hard. To prove it, AI researchers trained a new model that's less powerful but much more ethical. That's because the LLM's dataset uses only public domain and openly licensed material."
tl;dr: If you use public domain data (i.e. you don't steal from authors and creators) you can train a LLM just as good as what was cutting edge a couple of years ago. What makes it difficult is curating the data, but once the data has been curated once, in principle everyone can use it without having to go through the painful part. So the whole "we have to violate copyright and steal intellectual property" is (as everybody already knew) total BS.
I said "trying to use" because, as noted on the model page, license laundering and inaccurate metadata are risks. It says the paper discusses these things.
The model page also says this is a base model that has not undergone any form of alignment. I gather this means that it hasn't been post-trained for instruction following like the well-known LLMs. Someone please correct me if I'm wrong about that.
If someone (@simon?) puts together instructions for trying out this new model, it will be interesting to see how it compares to the well-known models that take the typical unscrupulous approach to training data and have undergone post-training (e.g. OpenAI's RLHF). I suspect that this model won't be nearly as useful (to the extent that LLMs are useful in the first place).
Note: My intent with that last sentence wasn't to defend the practices of the well-known LLM developers, but to speculate that even with this new development, we won't have an LLM that's both useful and free of the ethical issues around the training data and process.
Calling all analog film photographers! 🎞️ I've been working on Filmbook, an open-source app to help you keep track of your film usage. It's built with Rust & GTK4/libadwaita for a smooth & modern experience – and it even runs on Linux phones like the Librem 5 and Pinephone Pro! 📱 The first version is ready for testing, and I'd love your input on what features would make it even better! Join the community & help shape Filmbook: codeberg.org/bjawebos/filmbook ✨ #filmphotography #analogphotography #rustlang #gtk #opensource #community #testing #featureideas #librem5 #pinephone #linuxphone
Your voice, our drive: We had the privilege of interviewing @tolgayenici a mechanical engineer residing in Canada, to find out why he chose Tuta Mail, how he became a privacy and open source enthusiast, and why privacy matters to him. 💬🔐
Tolga Yenici used to be an average tech user, but now he is on a journey to choose private, open source tools like Tuta Mail. We sat down with him to find out why.
"AI companies claim their tools couldn't exist without training on copyrighted material. It turns out, they could — it's just really hard. To prove it, AI researchers trained a new model that's less powerful but much more ethical. That's because the LLM's dataset uses only public domain and openly licensed material."
tl;dr: If you use public domain data (i.e. you don't steal from authors and creators) you can train a LLM just as good as what was cutting edge a couple of years ago. What makes it difficult is curating the data, but once the data has been curated once, in principle everyone can use it without having to go through the painful part. So the whole "we have to violate copyright and steal intellectual property" is (as everybody already knew) total BS.
It'd be interesting if AI companies lobbied to increase the works in the public domain by decreasing copyright duration. That's something I'd actually support. Copyright is too long. And it would then be a legal, more ethical, industry instead of pack of VC-funded thieves. Strange bedfellows.
@shanecelis IMO, the right approach here is to essentially copy what radio does. Give AI companies a blanket right to train on anything as long as they pay royalties to some central organization, which would then distribute them to content creators.
@miki @shanecelis Not sure if it is that good. Current central organizations existing for, say, music or films, are assholes, and I don't see why this one would be any different. Moreover, I wonder how does one pay royalties to social network users (whose posts are being crawled as well). That's millions of people
Just to iterate on your idea (I still believe we can do better), we can force social networks to act as intermediaries between end users and AI bros. If the AI bros love subscription capitalism so much, let them subscribe to, say, Meta to scrape Facebook, with Meta being obligated to distribute the fee obtained between users on the platform.
...with an option to opt-out from scraping. The world where I can sell my kidney but can't sell my personal data is kinda weird.
For day 6 of #AudioMo, here's an AI attempt at audio-describing a long point of tenis. Unfortunately I've had to butcher the description for time, but the words are all its own.
So, whatever you are saying is so quiet that it's almost inaudible, and you just put us in a player that doesn't show any tags. In saying that, I bet we could get some really good info about tactics from AI that sighted people wouldn't even think to mention, even radio commentators
@jakobrosin I imagine that it won't be long before we can ask an LLM to fit its description around unspoken parts of a video. Especially if it has the script, as well, there's real scope for it to do a good job at this sort of thing.
@jakobrosin Gemini. I gave it the entire video and asked for a description in very broad terms. tempted to give it a scene from a movie or something and ask it for a subtitle file around the nondialogue bits, see how it handles it.
It looked like it used both the image and audio as its imput. Thanks for the tip, I didn't know gemini can take video as an imput now. Maybe I need to broaden my llm subscriptions
It's like the wettest dreams of the TDD hardliners of the oughties are coming true, but in an infinitely more stupid way. old.mermaid.town/users/Kymberl…
By the way, I can't be the only one who's noticed the similarities between today's LLMs-in-programming "discourse" and the Agile fads of the previous decade or two. If you don't do TDD you'll be "left behind," if you don't do Scrum you'll be "left behind," if you don't do pair programming you'll be "left behind" etc. I'm just waiting for some vibe coding enthusiast to rediscover the derogatory term "cowboy programmer."
Jeff Geerling got the brunt of YouTube wrath when he showed people how to host their own media {media that they legally own, which means that they either have the originals, or have paid for whatever digital version they have} with the power of Open Source tools Jeff explicitly made sure that he never ever told people how to circumvent subscriptions or worse. Nothing that could harm YouTube bottomline was ever discussed in this video.
Yet for reasons obvious to Open Source people like me YouTube gave him his second strike.
3rd Strike and you're gone. This is how Google / this is how Alphabet is treating their Golden Geeze.
Creators like Jeff are very valuable both to the people who follow them and to YouTube. However Google seems to be at Super odds with Open Source, needing it to run their data centers but hating it because they have to share the code again that they've worked upon.
Google is a paradoxical Company which is being controlled by Alphabet, a schizophrenic Entity drunk on power Ads and control
Deutsche Bahn mal wieder zu spät, Zug wird mal wieder nicht in die Schweiz gelassen, Halt fällt deswegen mal wieder aus, und damit meine Fahrt nach Essen. Die Alternativen für die nächsten 2 Stunden sind alle komplett ausgebucht. Immerhin hab ich Trost dabei
@Bubu ich rede von DE, die Alternativen ab Basel, resp. Basel Bad. Bis dahin komme ich problemlos und pünktlich, und das obwohl ein Zug auf einem Gleis verendet ist. In der Schweiz gibt es keine Reservationspflicht, ganz im Gegenteil, Reservationen sind sehr unüblich
I don't quite understand why anyone would do this, then, but I also don't know much about the whole software stack running on the Switch, to be honest.
@thedoctor @claudex there's also a possibility that they have the copyright lines intact from an early import even if they updated the package later...
I'm re-launching my original version of ReaKontrol, a #REAPER extension to integrate with #Komplete Kontrol keyboards, now with support for S-series Mk3 keyboards. It doesn't have some of the visual feedback or other features of the Brummbrum fork, so if you need those things, this is not for you at this stage. Website and downloads: reakontrol.jantrid.net/
May I ask what other features the other fork has besides visual feedback? I am totally blind myself and have always been a bit confused on which version to use.
WANTED: Intel Architecture Labs 1990’s CD-ROM’s. They appear to have maybe been monthly. They contained a mirror of Intel’s ‘download.intel.com’ ftp server, specifically the /ial/ subdirectory which is not in the 2014 backup of the site on archive.org.
Lots and lots of white papers and design guideline documents in there. Especially looking for ones from the late 1990’s (1998-ish onward) if they exist. I’ve seen references in mailing lists to them that lead me to believe they do. Example gem: intel trying to cover its ass after the FDIV bug, and have some more FDIV
”The opaque and unaccountable Silicon Valley companies that facilitated both Brexit and Trump are now key players in an accelerating global axis of autocracy.” —Carole Cadwalladr, The dark lord of Silicon Valley broligarchy.substack.com/p/the… #broligarchy
”…the real point of DOGE is not to save American taxpayers money. … Musk’s operatives have been systematically gaining access to databases across the US government and are now merging them. …these datasets are now being merged into a giant surveillance machine. A giant surveillance machine presided over by the dark lord of Silicon Valley, Peter Thiel.” —Carole Cadwalladr, The dark lord of Silicon Valley #doge #musk #thiel
”JD Vance, the vice president, is a wholly owned Peter Thiel project that has been 15 years in the making. He first met Thiel when a student and he owes his entire career, first in finance and then politics, to him.” —Carole Cadwalladr, The dark lord of Silicon Valley #jdvance #vance #peterthiel #thiel
🚀 Big thanks to MacStadium for sponsoring macOS runners to power our CI pipeline! This support helps us manage macOS builds more efficiently in GNOME. 🙌
Introduction: GNOME GitLab now uses macOS runners from MacStadium for managing our macOS CI pipeline. This offers significant improvements in supporting the GNOME platform on macOS. In this blog,...
Okay, now I'm frustrated. I was trying to get the ancient A2 Apple II+ emulator, a completely text-based Apple II+ emulator, going on modern Windows, and I can't for various reasons. So I was gonna try to contact Rich Skrenta, the author, to see if maybe he'd even thought of it since twenty-five years ago when I wrote to him and compiled it for DOS under DJGPP. There appears to have once been a page about the emulator at skrenta.com/a2/ but that entire site now seems to immediately redirect to Linkedin, and you have to join Linkedin in order to be able to get contact info. I mean that's fair enough but as far as I know I would have no other use for that service since I'm not in the professional world, etc.
his contact details are private on linkedin anyway. You can message him through the platform, but his email, phone etc are hidden. he's been active a week or so ago judging by his activity but no idea how often he checks his messages on there
@cachondo And it gets worse and worse. So I created an account on Linkedin…only to find out that to actually message him, I apparently need to have a Premium subscription. Not to worry, as I can get a thirty-day free trial. But oh yeah, they do need a payment method, and give a vague reason for needing it, but I'm sure it's really that they're hoping I'll forget about it and not cancel in time so they can bill me for something I have no use for…all to contact one guy about code he probably hasn't even thought of in twenty-five years! Sheesh! I'm not going *that* far, at least not now.
In a groundbreaking move that has caught the world’s attention, the White House recently announced its plan to establish a lunar time zone. This decision marks ... Continue Reading →
If you’ve ever left home for a while and returned to a cat that looked at you like a stranger, you might have wondered—do cats actually ... Continue Reading →
Talked to a DV org today that may be able to help with some expenses. There are still a lot of unknowns and for now, my big fundraising focus is trying to get a car by the end of June. Another 2-3k should get us there. I am going to look into car loans as well since I was able to get my credit cards down slightly, but it would be ideal to be able to buy outright.
Alright y'all... it's halfway thru the month, so what I have on hand = July rent. And while I'm figuring out how to get by without a car, it will be a MASSIVE difference to quality of life as a newly solo parent. Getting out with the baby is so important, not to mention work, errands, and my own social needs.
Y'all have been incredible these past few weeks—I hope another fundraising push can help meet this urgent goal 🚗🚙
Between court things, I'm working on job applications. I know it's a tough market so the more cushion I have, the better! The fundraiser has stalled this week, so please share & consider pitching in if you're able to. This is a massive shift & I'm so lucky to have this village to help me get to the other side ♥️
Most individuals understand tactile graphics as raised-surface representations of non-textual information, such as shapes, maps, and diagrams. Tactile graphics are...
I hesitate to call this a review since I did not use this Motorola Razr as my primary phone for any substantial amount of time. However, I did acquire it
Samsung has begun rolling out the first beta of One UI 8.0, based on Android 16, to its flagship Galaxy S25 series. For visually impaired users relying on
Cleverson has moved
in reply to Lukianos (Fabian) • • •