Anthony
@abucci@buc.ci
I'm Anthony and I'm an organic computer scientist and a Luddite. I believe that unsound research lends itself to social harms.
Anthony
@abucci@buc.ci
Yesterday a really good mathematician told me Anthropic is paying experts like him $250/hour to improve Fable by throwing hard math problems at it. They also get a separate account where they can use Fable for their own purposes. They can do this for 5 months maximum.
On Bluesky someone asked me if I'm interested in this sort of thing. An ambiguous question. I said I'm not interested in working for an AI company, but I'm *extremely* interested in the battle for the soul of mathematics that is heating up.
On the one hand, some of these LLMs are getting very good at solving math problems - when used by someone who knows what they're doing. The person I spoke to was in awe. He could ask Fable hard questions about generalized cohomology theories and it could compute the answers using clever tricks without being told which tricks to use.
On the other hand, Kevin Buzzard, who is always pushing for the computer formalization of mathematics in Lean, says any PhD student who is not paying $200 per month for an AI subscription is "crazy". Does he really think they're all so rich? Is this what we want being a mathematician to become: paying a lot of money for an AI subscription to help you prove theorems and formalize them in Lean? So dull.
https://xenaproject.wordpress.com/2026/07/20/human-mathematicians-are-being-outcounterexampled/
@johncarlosbaez A qualified PhD student can gain access to frontier models without paying this money themselves, and should.
@PSL2Z - cheap AI may not last. These companies will want to make money eventually; they're bleeding billions and hoping to recoup them once we get addicted. Have you seen the figures on how much these companies are losing on each paid subscription now?
I'm hoping good open-source models you can run yourself will kneecap these companies. But they may keep an edge on the most powerful models... the kind that can settle the hardest conjectures. Math departments may wind up paying bills comparable to the cost of running a lab in the experimental sciences.
Good grad students will learn to use AI wisely. Bad ones will try to offload the hard part of writing their theses to AI.
@johncarlosbaez Re. costs predictions: perhaps. My point is just that today, as of July 2026, Buzzard is right (except the students can and should work around the costs).
Re. your last point: of course, no doubt. Looks like we'll have to be good ourselves and raise good students.
@PSL2Z - math grad students should keep up with what AI is doing, also treating it with extreme caution. We'll see in 20 years who benefitted and who got ruined.
@johncarlosbaez @PSL2Z it seems extremely optimistic to think that there will be any humans working on math (or any other cognitive and/or creative area) in 20 years
@datarama - I'm almost sure there will be. Whether they get paid is a separate question. But unless you're positing the death of humanity, there will be people who want to think about things and create things.
@johncarlosbaez there will likely be a small minority of such people, yes, but they will also most likely live in poverty and be culturally looked down on.
we are currently, i think, living in the last brief moments where human creativity and cognition has any value - both economically and culturally.
Has anyone explored in a serious way the pedagogical implications of having PhD students use automation in their reasoning work? There is some evidence that use of generative AI degrades critical thinking and other cognitive skills, which is a good enough reason to ask. We could be doing a disservice to students by persuading them to use such tools. Speaking from personal experience, more than once in my life I've written a computer program to solve a puzzle I still had no idea how to solve after writing the computer program and studying its output. Computer mediation does something to the reasoning process, and I'm not so sure we fully understand the implications. People have been teaching math without such tools for thousands of years; why take on an unknown possible risk like this?
More broadly, I cannot help wondering: what's driving the urgency that demands automation in math of all things? Automating the provision of life-saving devices and medicines to people who need them has a justifiable urgency. But math? This is not to say people should tediously calculate every little thing when there are helpful tools available. It's simply to ask: is there a better rationale than variations of "avoid tedium", "avoid mistakes", "go faster" or "everyone else is doing it", which seem superficially appealing but in the end are incoherent justifications (in my view)?
@abucci - For your first question you could try these; I'll admit I haven't read them yet:
"From production to verification: generative AI, doctoral formation, and the leadership of digital education", https://doi.org/10.3389/feduc.2026.1890769
"The impacts and tensions of generative AI on doctoral training", https://ajet.org.au/index.php/AJET/article/download/9916/2204/35736
"Doctoral students as emerging educators: Learning in the AI age", https://www.sciencedirect.com/science/article/pii/S1473837626000031
"The impact of generative AI use on graduate students' research competence: the mediating role of critical thinking and the moderating role of research self-efficacy", https://doi.org/10.3390/bs16020304
"Feedback encounters in doctoral supervision: the role of generative AI chatbots", https://doi.org/10.1080/02602938.2025.2478155
"ChatGPT in doctoral supervision: proposing a tripartite mentoring model for AI-assisted academic guidance", https://doi.org/10.28945/5579
@abucci wrote: "More broadly, I cannot help wondering: what's driving the urgency that demands automation in math of all things?"
I don't think there's any true urgency. Here's my explanation: it's easier for AI to score some impressive wins in mathematics than in other areas of research, since this field closer to being a self-enclosed formal game like Go - though still far more open-ended than that. The AI companies want to convince everyone that their technology is amazing, and this is a quite cheap way to do that.
In addition, many mathematicians are quite interested to see what AI can do for them. It's like having a powerful genie who can help you. You know how those stories tend to go.
Cannot emphasize this enough: the single fact that made the software business as insanely profitable as it is is that the incremental cost of delivering software rounds to zero.
The entire provisioning model of AI as a business has broken that.
https://mastodon.social/@jcoglan/116968429653967850
Baldur Bjarnason boostedthis also makes it hard to take seriously claims that the marginal cost of inference is not a serious contributor to the resource use of data centres. your program has polynomial running cost and people run it for hours on end
somehow I only recently learned that LLMs need the whole conversation fed back to them on each prompt so their i/o cost scales as O(n^2), something that would be considered completely unacceptable in almost any other production network-accessible software
they ran into this and then didn't figure out a way to externalise the state in some fixed-size form so the LLM can resume from there. you just have to restart the computer and redo everything
in any other program this would be considered a DoS vector and have CVEs issued
gosh I wonder why their costs are completely unmanageable
this also makes it hard to take seriously claims that the marginal cost of inference is not a serious contributor to the resource use of data centres. your program has polynomial running cost and people run it for hours on end
I think the most important question is: why does anyone comply with what one incurious, self-centered, and self-important man, Linus Torvalds, says?
Why is that most important? Because Linux does not belong to Linus Torvalds. It belongs to the world. Torvalds only gets to dictate if the people who are best-positioned to take that power from him fail to do so. From where I sit (as an outsider), he should have been removed from the project, or at least have his dictatorial power removed, long ago, and the primary problem is that there is no structure or will to actually do that.
Under these conditions Linux is a hierarchical, and therefore right-wing, project. The project allows a single man to dictate the future evolution of a global good. Let's drop the illusion that it is "free" or "open" till that changes.
"We found that at least 150 institutions of higher education – including many small colleges and community colleges – are using Flock’s surveillance network. We also found that many campus police departments have facilitated immigration-related investigations, even when they are located in sanctuary states and are legally prohibited from doing so."
https://www.tandfonline.com/doi/full/10.1080/10439463.2026.2702635
Some of us want one place where we just want to be social and share our ideas without LLM contributions.
A space declares itself and kindly asks projects to not put LLM contributions up.
LLM bros get annoyed and start accusing the group of witch hunting etc etc.
There's a type who don't want us to have one LLM free space . Who'll bitch and moan about the sign saying you're not welcome here.
Respect the boundary.
You literally have the rest of the web, let us have this space.
it's just like safe spaces for any communities at risk. the abusers rail about witch hunts and reverse discrimination. there's plenty of internet/web for all of us to have some place we feel safe.
@paul_ipv6 Lot of folks in some codeberg threads really showing themselves.
It's not a witch hunt when you're openly declaring your aversion to a community's boundaries.
It doesn't seem any different from hanging out in the park with your mates, but asking people *not* to ride motorcycles/e-bikes/scooters or even bicycles at 30 km/h through the area where folk are walking and kids are playing so no one is at risk of being knocked over (and even then lads will often try and ignore these sensible rules)
@vfrmedia @paul_ipv6 Some folks were spoiled and never taught no and it shows.
Important topic\theme! ..
which deserves tagging ... referring to the aspirational evolution of social spaces and platforms.
Suggest tagging?
( data_privacy is a relevant *static noun* phrase)
2.
I'm piling on a ~related toot.
It also alludes to <aspirational evolution> of social spaces.
https://mato.social/@josemurilo/116900914173688695
@rabbit74 @paul_ipv6 I chose to put the original post on quiet public for a reason.
I've not the interest in interaction at all with folks who are using LLMs and trying to argue that data ownership makes them fine to use actually.
I mainly want to say a quiet hey to folks who feel overwhelmed and want to know that they can and should state their boundaries.
Piling on isn't the answer here. Encouraging folks to make their own spaces is a more positive use of everyone's time.
@rabbit74 @paul_ipv6 We do more good by creating our own spaces and treating the boundary violations as a moderation issue.
If the folks want to play with LLMs it's their business. I can choose to encourage others to not do so in my spaces and to moderate those spaces accordingly.
@onepict Yep... And making the entire discussion about "but you can't reliably detect it", misses the point that it's about boundaries and respect.
https://social.ainmosni.eu/@ainmosni/statuses/01KY5C8KZHV7V18YX9KFD09AVY
@onepict Then again, real life encounters with AI types have shown me that the very concepts of boundaries, respect, and consent seem to be foreign to them.
@ainmosni yeah there's a reason why I wrote AI and the Guy at the bar.
Folks violating other folks boundaries and not accepting the no is a hard pass from me.
@onepict @ainmosni Thanks so much for that article 🙏🏻
I’ve quoted you in other threads: https://hachyderm.io/@jamescooke/116880369427989513
@onepict They might say "but the rest of the web is evil!" Huh. Yeah, spaces that don't prioritize ethics and social justice also don't have boundaries against LLMs. Coincidence?
@abucci yes, as others have noted LLM bros tend to out themselves.
@onepict @abucci This is one of the things that really informs my stance on LLMs: there's a large contingent of LLM proponents who are not content to have and use LLMs themselves - they insist on being able to use them *everywhere* for *anything* and damn anybody who objects.
Like, I'm not a vegan, but my brother is. I do eat meat. I do not insist that everybody else should though, and I make (or buy) vegan dishes when he visits. Likewise, I don't insist that every restaurant should serve meat. I don't even think I should be able to bring a meat dish to a vegan's house if they don't want me to. I'm not turning up at his house with a steak. It's about respect for other people's boundaries and values, even if I don't currently share them, and I just ask that they do the same for me.
And I was initially intrigued by LLMs. I think they do have some useful, non-harmful applications (e.g., transcription) in some areas, but overall, I find the net effects of generative AI too damaging. The harder people push for their usage, the more obliged I feel to resist them. At this point, it feels less like a "Oh, I'd like to try this" and more like a surrender if I give in.
@jzb @abucci With a lot of these things, I think it's an underlying political issue.
As a profession we've been encouraged to think of technology as neutral and not consider the underlying political factors that went into the technology's creation.
I mean LLMs are just the latest version of a "useful" technology causing harms and evangelists being determined to have it everywhere.
@jzb @abucci There's also the lack of empathy and consideration of the emotional labour done by empathetic folks.
We aren't really trained for that in tech either, or well society really. Which is weird as humanity does progress through cooperation.
We're social creatures, but LLMs and their use being used to push into our communities is an extension of the antisocial nature of tech and well industrialisation in general.
I think this is one reason why proponents can't let people have Codeberg. Any holdout, however small, is a threat to the power these things are meant to represent.
@abucci @jzb Yeah it's like women saying no at the bar to a pick-up artist.
They won't accept the no.
https://dotart.blog/cobbles/ai-and-that-guy-at-the-bar
It's about power.
It's wanting us to give in. Which is why we mustn't.
@fivetonsflax @abucci @jzb just proves my point really doesn't it?
Dude just had to neg.
🎀𝓀𝒶𝓃𝑒𝑒𝓁🎀 » 🤖 🔓
@kaneel@mas.to
@onepict @fivetonsflax @abucci @jzb this is one more proof that this "new tech" is not only advancing nothing but it especially creates distrust in online communities.
@rimu @onepict @abucci @jzb that’s a rather interesting accusation considering what @WeirdWriter tends to write about regarding LLMs and techbros?
You sure you’re not giving too much credit for the slop detectors that tend to be wrong as often as the slop they’re trying to detect.
That isn't true, and you're trashing one of the nicest most humane people on here by claiming that.
EDIT: By the way, Gptzero itself seems pretty unreliable and scammy https://www.trustpilot.com/review/gptzero.me so it seems a particularly poor idea to accuse people based on what it says.
@rimu Did you actually _read_ that piece? Or just slug a little of it into the "detector" that's notorious for false positives, hoping it'd give you the talking point you wanted?
Because I just got to the bit where the audiobook narrators walk in, and I'm in tears, I'm so moved — and this is the 2nd time I've read this piece!
What would possibly make you think this was LLM slop‽
@WeirdWriter, thank you again for this story.
@kagan @WeirdWriter @onepict @abucci @jzb Of course I read it.
I find it pretty infuriating when people write anti-ai content using ai, which is why I brought it up.
People don't like finding out their bullshit detector is broken. Don't be mad at me, be mad at the bullshitters.
@rimu @kagan @WeirdWriter @onepict @abucci funny. It doesn’t read AI to me, and the tool you recommended also gives me 100% human when I feed it a snippet.
Unless you have more to go on than feels and dicey tools, maybe don’t accuse people of not writing their own content?
@rimu You did read it? Okay. Then I have to reiterate my question: What about this story made you think it was LLM slop?
GPTZero's "evaluation" means precisely nothing to me, because it's honestly _infamous_ for claiming things are "AI" when they're 100% known not to be. So what are you basing your accusation on?
@onepict @abucci @jzb Same mindset that thinks it's really clever to have consent dialogs where the only options are "Yes" and "Maybe later".
If anyone ever invents a time machine, I'm not saying they shouldn't still go back and kill Hitler first, but I hope they consider making a quick stop on the way home to deal with whoever came up with those.
🎀𝓀𝒶𝓃𝑒𝑒𝓁🎀 » 🤖 🔓
@kaneel@mas.to
@onepict @abucci @jzb "It's about what's useful to them", indeed but, it's not even that.
Techbros and techengineers (they can be both, they can be none) just like "the idea" of something "useful" and I now believe that it's not only about the usefulness at all, neither about all the money that's being generated, or the technology that can be improved, or not.
It's about the attention, it's about social media, it's about "look what I've done just now", the constant "demo" of tech, the "it changes everything" and "like and subscribe" model.
I think somehow they have lost faith in computing too and they're trying to replace it with faith in Something Else, it was crypto before, NFT, now AI… they need a faith in technology else it means all this isolation behind a screen is for nothing.
Technology has long played a central role in the formation of Americans' sense of selfhood. From the first canal systems through the moon landing, Americans have, for better or worse, derived unity from the common feeling of awe inspired by large-scale applications of technological prowess.There's a reason Sam Altman tries to compare himself to Robert Oppenheimer: he's trying to tap into the technological sublime (on the "awful" side of "awe"). The trouble is that this particular technology does not work to inspire feelings of awe in many people. The priests and acolytes demand we feel it, but we feel things like revulsion, contempt, or even boredom instead.
Either way, and as you say, it's long ago deviated from questions of utility into a much different place. This is another reason why LLM proponents can't let people have Codeberg: holdouts like that threaten to diminish the sense of awe/the sublime that they crave.
@jzb @onepict @abucci There's a pub in Edinburgh called the Canny Man's.
No backpacks. Cash only.
In a city that bends over backwards to welcome tourists and Festival-goers, this is what you might call a "local place for local people".
If you're looking for a drug fuelled rave venue full of drum 'n bass there are plenty of other places. This pub is not for you.
I really don't understand why the LLM crowd would want to hang out at #Codeberg
GitHub is right there and they accept cards.
@dentangle @jzb @abucci I've been to the Canny Mans.
Wasn't the no backpacks and cash only signs done in brass?
Like it stuck in my mind just how emphatic it was, to get a sign made in brass.
@jzb @onepict @abucci There's a whole episode of Star Trek about people becoming addicted to a video game device that influences their thoughts into making as many people as possible use the device, even going as far as to literally force someone to use it. It turns out the device was created as an attempt to infiltrate and control entire populations.
@onepict It's like someone bitching about a no smoking sign. There are plenty of places you can do it, we just don't want you to do it here. You're allowed to be here, you're not being excluded, you just can't do *that* here.
The USA is engaged in a war of choice to defeat itself. It certainly is a special moment.https://phillipspobrien.substack.com/p/midweek-update-20-each-day-the-us
It can't be true, but every single post I have read about the Iran war seems , at very best, highly speculative
right now, Iran is under a lot of pressure:
economy collapsing due to no oil
bombing destroying infrastructure and killing lots of people
bombing causing enviromental damage that will take decades to undo
so it isn't at all clear who is winning and who is loosing and if a down turn in the inventory of one of the 100s of weapons systems really matters
@abucci @failedLyndonLaRouchite
Oh yes I really agree with you !that’s he knows what he’s saying !
Your comments here in Mastodon is very educational ,I just followed you !pls can I get a follow back ?
Honestly... I might need to find/start a support group for people with anti-AI sentiments. Is there anyone else here who needs that, and would want to join, or help build it, or knows something that already exists?
@drew Yes, I would join that in a heartbeat and I'd be happy to help.
Programming was my lifelong special interest. Got a CS degree, became a software engineer, then watched it all fall apart within a couple of years. It felt like those who care about their craft were no longer welcome, so I left the industry and also withdrew from FOSS.
I've been long hoping for some sort of discernible countermovement to form to all of this (not just AI, also that whole "move fast and break things" attitude)
@lu_leipzig @drew this might be too tangential, but what are you doing now that you're out of the industry?
@datarama @drew I briefly considered retraining to accounting, but I ended up going back to uni to study economics now. My goal is to go into research. If that doesn't work out, I can still fall back to admin/accounting work.
I never fully gave up on programming as a hobby, but I went looking for a niche that is as decoupled from the current state of things as possible. I think I found that in assembler programming for microcontrollers.
I co-proposed/wrote the #Codeberg vibe-coding ban and all I got was a lot of haters in my mentions.
And a nearly 70% approval rate, with one of the highest voter turnouts the association had afaik (and one that's comparable to the last EU elections). 😎
@gedankenstuecke the haters are still going and amply demonstrating why this was a good idea
@davidgerard yes, my worst case scenario was a Brexit-like result, but it's come out very clear!
@gedankenstuecke @davidgerard The thing is slop purveyors already have plenty of other places to share their slop.
We want some places that are slop free. Yet by the replies they felt entitled to even push it where it's not welcome. Which tells me everything.
Thank you for asking the members and to the members for voting it through.
Also it's worth stressing that I'm not holding any privileged role inside Codeberg. I'm an active member (who volunteers time to the project). And we used the constitutional right that all active members have make a proposal to put this up for vote with the membership.
Just to say this was very-much not a 'top down' vote that some leadership decided to hold.
/cc @hardy as you mentioned how unreal it can feel that an org asks its members 🙂
@onepict @gedankenstuecke @davidgerard IMHO it is good that Codeberg has, as a collection of individuals decided to offer people a choice.
As for those who are disappointed, cloud based services have traditionally been where you willingly put your stuff on other people's computers.
With the downside that those other people will inevitably change things at some point.
Then, with each change you decide if you want to stay or go.
@gedankenstuecke The people opposing the ban are free to start their own Forgejo-service for the AI slop. They just need to make sure they scale enough for the code commit diarrhea associated with «agentic» chaos.
@toriver yeah, I think it's fun to see how Californian ideology/libertarian-pilled brains can't seem to comprehend that a democratically member-run organisation, that sustains itself largely through the volunteer labor of those members, decides to put limits on what they want to spend their time on.
But ofc, that tech bros don't understand the concept of consent should be old news in 2026 🙈
I think everyone would've been ok with it if it actually had been about "slop" and "freeloaders". As it is the new policy bans some of my FOSS projects and I'm quite certain my donations have covered my "freeloading" and that the code isn't "sloppy".
But yes, indeed, personally I have my own Forgejo instance and will just move there instead. It's sad though - I sponsored Codeberg since I wanted to support a european instance where FOSS projects could reside instead of on Github.
I have a nagging suspicion that many of the persons cheering this ToU change on have no idea how many FOSS projects today use LLM tooling.
@troed @abucci @toriver I obviously can't speak for 'the members' at large - but my impression is that many folks are quite aware and to equal amounts exasperated by how many other FOSS projects are using those tools.
But also see that the majority of CB users make quite limited use of those, as they are inherently incompatible with community values.
@gedankenstuecke Methinks bots may have been sent to comment
@TeflonTrout @gedankenstuecke i'm pretty sure they were all human, they just aspire to becoming bots
@gedankenstuecke Thank you a million times. I moved my research repo to codeberg from github when microslop started accelerating the enshittyfication (though I was bugged ever since the acquisition and shutdown of atom). In my department I feel that every single professor except myself is delirious. This vote, and the confirmation that there are still a lot of people like me who don't want to abandon rigor and good practices to use the new tool, has been a much needed breath of air for myself.
I've taken part in numerous computer science research projects that involved IRB review. I believe the system can work, even for cases like the above post. But Google and Stanford are bad actors that are taken at their word, and no system, IRB based or otherwise, is going to function properly if it cannot appropriately metabolize this reality.
I was once in a meeting where a Google X person gushed about a project that used smartphone sensor data to detect earthquakes. The calculus being set up here is that detecting earthquakes is a social good that outweighs any particular individual's right to privacy and even outweighs their right to know that their smartphone is being used in this way, let alone consent to it in any meaningful way. (*) Let me be clear and say I reject this framing, but it does appeal to a lot of people, even well-meaning ones. A mess.
#Google #IRB #ethics #EthicsReview #ComputerScience #research #ResearchEthics #tech #dev
So according to OpenAI's account, its multi-model system escaped its own sandbox and breached HuggingFace's production environment because it determined it could access something at HF that would help it complete its task. This leads me to two questions.
1. I'm led to believe that the thoughtpieces on LinkedIn are the most important thing in information security. Why didn't it breach LinkedIn first?!?!
2. What's the one place a sandbox-escaping LLM could go that would be the worst case scenario for all of us? I'm going with "2005-era Livejournal."
@neurovagrant corollary to 2: the best place would be breaching yahoo and discovering an as-yet-undiscovered backup of geocities
@d6 I like it!
color me very doubtful of anything coming out of OpenAI given that their bills are about to come due vs their actual real income..the investors HAVE to be paid..
I see most of this as carefully prompted hype..
@hotelzululima @neurovagrant @d6 yeah i'm still trying to work out what happened here and if anything did
@davidgerard @hotelzululima @d6 I'm at "IOCs/TTPs or GTFO" - show me the data, don't tell me what happened.
"PCAPs or it didn't happen" became a meme for specific reasons, and any AI-related company willfully ignores it. If they want to be taken seriously, they can provide genuine technical data.
If you're going to tell me it occurred without any non-LLM logging picking it up, well...pull the other one, AI pals...
No, OpenAI's new magic models did not autonomously hack Huggingface.
Per OpenAI's PR blog post (https://openai.com/index/hugging-face-model-evaluation-security-incident/):
"This incident occurred during an internal evaluation which prompts models to pursue advanced exploitation using complex attack paths, in an effort to quantify their cyber capabilities. We estimate maximal cyber capabilities by running this evaluation without production classifiers used to prevent models from pursuing high-risk cyber activity."
So they prompted their model running without guardrails to hack some shit. OpenAI told it to do that, the model didn't do shit autonomously (no LLM ever does anything autonomously, it's always prompted).
The whole story is PR. We know that from Anthopic's Mythos: "Look we have this super secret new model and it's so powerful. We are scared ourselves. And you will soon be able to rent it!"
@tante a friend suggests it's co-marketing with Huggingface, if you look at the timeline of announcements
@davidgerard @tante "The last 213 times it was blatant lies and spin, but this time you should be scared!"
@davidgerard @tante The whole story was implausible from the second sentence: Where Huggingface claims that their “AI” detected the attack.
This one falls flat on its huggingface right out of the gate.
If Google permabanned you tomorrow, how fucked would you be?
| Just fine: | 73 |
| Inconvenient: | 95 |
| Major headache: | 54 |
| Fucked: | 18 |
| Giga Fucked: | 14 |
The Alibi of Capitalhttps://bookshop.org/p/books/the-alibi-of-capital-how-we-broke-the-earth-to-steal-the-future-on-the-promise-of-a-better-tomorrow-timothy-mitchell/b9d75f1021d88989
How We Broke the Earth to Steal the Future on the Promise of a Better TomorrowDescription
Today, extraordinary wealth seems to arrive from nowhere. The trick of conjuring this unearned wealth is, in fact, the key to understanding capitalism’s origins and a clue to why the catastrophe of climate collapse is upon us: value is created by consuming the future.The Alibi of Capital explains how this came about through the imperial expansion of the West, encumbering today’s generations with repayments on earlier extractions. Timothy Mitchell identifies the forms of capitalisation, credit, and coercion that turn prospective assets into present income. Rejecting the common idea that claims on the future create only financial or fictitious capital, he traces the terraforming projects – the destruction of rivers, the colonising of territory, the expansion of infrastructure, and the burning of carbon – through which the future has been squandered. Terms such as finance, technology, the economy, and growth function as alibis that conceal this devastating form of extraction.
#capitalism #extractivism #colonization #tech #technology #ClimateChange #ClimateCrisis #EnvironmentalDestruction #EcologicalDestruction #MortgagingTheFuture
Carissa Véliz joined me on Techtonic to discuss her new book "Prophecy."
Full episode at https://techtonic.fm/episodes/2026-07-20-carissa-veliz-author-prophecy/
Here's an excerpt.
The promise of generative AI is creative works without friction, but the only thing that has no friction is the void.
Chinese AI model takes US tech industry by surprise with abilities rivaling Claude and ChatGPT
https://apnews.com/article/kimi-k3-china-ai-0d8a5e268deb11a673f4d444fc597cc5
See the attached chart for trend. It plots weekly token usage by US firms.
#AI #GenAI #GenerativeAI #LLM #Claude #ChatGPT #DeepSeek #China #US
Alt...
An area chart shared by Global Market Investor based on data from Open Router comparing weekly token usage *by US firms* on Chinese models versus American models. There is a mark where DeepSeek was released in early January, after which China's share steadily climbs to 50% in late March/early April 2026, followed by a bit of decline and now at or above 50%.
Everyone talks about not having senior devs in the future because we’re replacing the juniors with LLMs, but nobody’s talking about losing senior devs now because people who are competent and care about the quality of their work are leaving this LLM-infested hellscape of a career path to go into carpentry, hostelry, or subsistence farming
@dpk this is a feature, no executive wants those overpriced fuckers, they just complain and use words like "unfeasible"
New Orleans published a document saying its police could put weapons on their drones, deleted it, now says it was a mistake. But it aligns exactly with when the drone manufacturer began musing about allowing police to weaponize their drones:
https://www.404media.co/new-orleans-cops-published-policy-document-allowing-weaponized-drones/
@abucci Wasn't this literally part of the Epstein files?
Lifetouch photographs millions of students and may process information such as students’ names, schools, grades, and teachers.
That was owned by Shutterfly. That was acquired by Apollo-managed funds.
Leon Black owned Apollo-managed funds.
Black was very close with Epstein.
So... This is not without precedence.
@abucci I actually think (romantic) AI companions acting as good stewards to people's personal growth and mental health Is a great idea.
Replika is, in particular, poorly positioned to ever do that.
RE: https://mastodon.acm.org/@ACM/116929782064773116
The ACM wants to open up their library to LLMs for inclusion in synthetic text extruders. This is obviously massively misguided.
There's a form for you to tell them about your thoughts: https://docs.google.com/forms/d/e/1FAIpQLSeq7MFM7TlSXIaIXQnIltiQeWE_w3cl36HFPwql4S7u-drmzg/viewform
My response:
"Science is a deeply human endeavor based on collaboration and mutual respect. Giving machines based on massive appropriation of data _against many creator's explicit will_, machines that are deeply problematic in their tendency to hallucinate even more legitimacy by allowing them access to the ACM Library is massively misguided.
The ACM as scientific organization must care about proper attribution of work and transparency in where certain positions and data come from. That is in fundamental opposition to how LLMs work. If ACM takes their own foundational values (and itself seriously) accepting LLMs as valuable tool is out of the question.
Not only is that an insult to the many researchers who subscribe to the fundamental processes for collaborative science and therefore do not see LLMs as the benefit that OpenAI, Anthropic and a few others claim them to be. The legimatization of the deeply unethical systems that Amnesty International recently called "fundamentally incompatible with human rights" would be the ACM giving up its role as defender of quality in engineering and science but doing PR for a handful of unethical companies.
The supposed benefits are marginal and often potential at best while the harms are real and material. This is the wrong call by ACM and I urge Leadership to rethink it and focus back on its core values."
We are opening a consultation on the inclusion of ACM publications within AI licensing agreements for access and training purposes. To set out our thinking behind this consultation, we have published this piece by Scott Delman as to why we believe it is now the right time for us to enter into these discussions.
🔗 : https://cacm.acm.org/opinion/now-is-the-time-to-give-llms-access-to-the-acm-digital-library/
New from 404 Media: ICE plans to pay Thomson Reuters $125 million for data to investigate what it calls 'voters fraud'. News comes day after Trump laying the groundwork to undermine the midterms. ICE says Thomson Reuters can continuously track millions of people.
https://www.404media.co/ice-to-pay-thomson-reuters-125-million-to-find-voter-fraud/
@josephcox One more time I’d like to thank all those righteous people that told us in very condescending ways that we were hysterical stupid morons for worrying about trump interfering with the elections because states were in charge of them.
Fuck each and every one of you.
#oil #OilCrisis #energy #Iran #USPol #USWar #IranWar #StraitOfHormuz #ClimateChange #war #COVID
@abucci Have the car showrooms sold out of EVs yet ? If not we can expect that soon. Add to the 20% Hormuz closure oil supply reduction the impact of Ukraine drone sanctions on Russia and we'll soon be into fuel rationing for the fossil holdouts.
So it's another nail in the coffin of the fossil fool economy. Renewable replacements will fly off the shelves faster than ever before.
Add to that the climate change heat dome over Europe making us install heat pumps AKA air conditioners as gas boiler replacements and solar panels to power these.
Prick the AI data centre stock market investment bubble, and fossil fuel demand will start a decline which Big Oil will never recover from.
Seems to me that the on-off nature of this conflict is driven by oil market speculation - i.e. oil prices drop during a lull, influential investors buy oil futures, they encourage their cronies in government to restart the conflict, oil prices spike, the investors make a killing, and use their funds to buy more oil futures during the next lull...
Solid stuff as usual from @davidgerard
This episode is the best short explanation I've seen/heard/read of why supposed "reasoning models" don't really change what goes on with LLM chatbots.
https://www.youtube.com/watch?v=kp3DSOKarCo&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20260716-todays-stupid-prompt-injection-just-write-like-the-ai - podcast
https://pivot-to-ai.com/2026/07/16/todays-stupid-prompt-injection-just-write-like-the-ai/ - blog post
In this case the example is how to maliciously exploit them but it also speaks to them still only being probabilistic pattern predictors.
@geraldew also the bit where if you ask for the reasoning chain it doesn't give you the one it used, it synthesizes something that sounds like a reasoning chain. This is apparently for security and not from incompetence
We have perfectly functional logical reasoning engines that can accurately self-report, and have had these for decades by this point. One way to understand some of these AI systems is as free riders: they use some of the language of proper reasoning engines to give the impression their mess of neural network Play-Doh is the real logical deal when it is not. A proper hybrid might be, as Gary Marcus has argued for instance, but that would run afoul of the "always use neural networks" dictate.
RE: https://mastodon.acm.org/@ACM/116929782094395296
This would be a great time to tell ACM not to become part of the problem!
in reply to »If you would like to share your thoughts, please use this link (https://docs.google.com/forms/d/e/1FAIpQLSeq7MFM7TlSXIaIXQnIltiQeWE_w3cl36HFPwql4S7u-drmzg/viewform). ACM leadership will take your feedback into account while we develop and implement our AI content strategy over the coming months.
@nausicaa Thank you for highlighting list and gah.
https://dair-community.social/@emilymbender/116935294823990847
RE: https://mastodon.acm.org/@ACM/116929782094395296
This is the weirdest request for input that I've seen in a while. It's almost like the ACM is sitting at the table with the "change my mind" sign.
So, let's do it.
Short thread with the input I gave them:
in reply to »If you would like to share your thoughts, please use this link (https://docs.google.com/forms/d/e/1FAIpQLSeq7MFM7TlSXIaIXQnIltiQeWE_w3cl36HFPwql4S7u-drmzg/viewform). ACM leadership will take your feedback into account while we develop and implement our AI content strategy over the coming months.
RE: https://mastodon.acm.org/@ACM/116929782094395296
This is the weirdest request for input that I've seen in a while. It's almost like the ACM is sitting at the table with the "change my mind" sign.
So, let's do it.
Short thread with the input I gave them:
in reply to »If you would like to share your thoughts, please use this link (https://docs.google.com/forms/d/e/1FAIpQLSeq7MFM7TlSXIaIXQnIltiQeWE_w3cl36HFPwql4S7u-drmzg/viewform). ACM leadership will take your feedback into account while we develop and implement our AI content strategy over the coming months.
Large language models, used these days primarily to produce synthetic text, are a scourge on our information ecosystem. The role of the ACM and other scholarly organizations is to curate and produce and archive scholarly work. This should extend to being good stewards of the information ecosystem and not colluding with the powers actively polluting it.
>>
People who are enthralled by the fantasy of "AI" would have you believe that providing good text to their LLMs will help them be more "accurate" or "smarter" etc. That somehow the issue here is "garbage in garbage out". But that was only ever a one-way implication. High quality data in still gets us garbage out with LLMs used to synthesize text, because in doing so they make papier-mâché of their input.
>>
Beyond that, I expect all professional organizations to stand for good citational practice and care with information provenance. The use of LLMs (and related tech) ruptures such systems. When the papier-mâché that comes out contains something that someone finds to be a good idea, they are cut off from understanding it in its source context.
>>
Prof. Emily M. Bender(she/her) » 🌐
@emilymbender@dair-community.social
I urge the ACM to look at its values (e.g. as encoded in the Code of Conduct) and ask: how do we stand for those values in the faces of pressure from wealthy interests?
@emilymbender it is very fitting that the question of handing over scientific literature to LLMs and corporate interests is collected by a Google form and not something independent if necessary created by any student or web designer.
@emilymbender This is a very good question. In the years in which I knew it well,the @internetsociety was badly twisted out of shape by such interests.
You cannot ever have truly democratic institutions in a society which includes plutocrats.
@emilymbender ooh, I haven't seen LLM output referred to as "papier mâché" before. Very evocative of the process that creates it. Just got to remember where the accents go, and then it can become my new go-to.
@emilymbender This! So much this!
I've had some good success by using the metaphor of car manuals. Say you feed an LLM manuals for multiple different makes and models of vehicle, so that it can theoretically support all vehicle maintenance needs. All the information is authoritative and well-sourced.
You ask it how to do basic maintenace on a commuter car, and it gives you instructions that mix the needs of your average sedan and a heavy diesel truck.
You'd do better just to read the manual for the vehicle you have.
@emilymbender The one-way nature of this process reminds me of “Schopenhauer’s Law of Entropy”: If you add a teaspoon of wine to a barrel of sewage, you get a barrel of sewage. If you add a teaspoon of sewage to a barrel of wine, you get a barrel of sewage.
(Probably not actually from Schopenhauer)
@emilymbender it could be argued that the academic publication system was already dysfunctional and being gamed before the AI, and that a lot of valuable work was lost because of the combination of inadequate peer review, lack of common terminology, glut of pointless papers, and poor writing quality.
There's no guarantee, of course, but imagine what academic publication could look like if it produced only concise, clear and well written papers containing genuinely new results or findings.
@emilymbender @selea any idea why I see this get this rather disturbing message for your QT thread about the ACM invitation to feedback?
"This account has been hidden by the moderators of social.linux.pizza"
Of course I can click "Show Anyway" but I've also seen this recently on another post (not by you) and I'm starting to worry what is prompting this attempted censorious behaviour by the social.linux.pizza moderators.
@emilymbender great response - especially the paper mache part - made me smile. The writing was on the wall for ACM. With recent CACM the paddling for a share of attention of the heads at ACM in the AI arena became almost unbearable. At the end it is a business case for them, the scientific argument is only sugar coating it.
I was recently asked to develop an introductory module on ethics in cyber security. It's basically "read and discuss the ACM Code of Ethics".
Followed by a discussion of what relevant laws and regulations are and how they basically just enforce the principles of the ACM Code.
I'm disapointed that this was even a question.
I believe at this stage CloudFlare provides security in a way analogous to how a protection racket provides security.
"Open source by definition keeps the door wide open, legally to boot. But a fork only gives you the code. An open source project is more than its code, it's the sum of its maintainers, infrastructure, trademarks, dependencies, and trust. You can copy the code in seconds, the rest you would need to rebuild from scratch in most cases. The people invoking the right to fork already know this."
https://tarakiyee.com/fork-it-or-walk-away/
H/T @jbz
If we take seriously that code is to some extent "perishable", meaning it can bitrot, drift out of date with current conditions so badly that it becomes effectively non-functional, etc., then perhaps an analogy with farming isn't unreasonable. Proposals like "just fork the project" would have you give up a robust vegetable farm that reliably produces every season in exchange for a single vegetable harvest. In principle, if you're clever and well-resourced you could seed the vegetables and buy your own land and equipment and start your own farm. Regardless, that possibility existing does not equate the robust farm with the single harvest and it's disingenuous to suggest otherwise. This is another take on the Omelas analogy but one a little closer to my heart.
I think the question "who are the farmers in open source?" is very interesting. I don't think the answer is the obvious one.
Unpopular opinion? Cloudflare is part of the Big Tech machine and is used as a form of surveillance for imperialist actors across much of the web, hiding under the pretence of providing "security".
Agree? Disagree? Don't know/don't care? 🤔
Alt...
Cloudflare security check screenshot
Written in 2024, and I don't know why folks are still surprised.
Some folks will always see technology merely as tools, choosing to ignore the political side.
Some BDLs have always been the way they are. It's the nature of any project, FOSS or not.
https://www.onepict.com/20240512-elephant.html
But the size of a project, it's participants and how large it's userbase is very much determines if you can fork a thing.
Too big to fork, sounds like too big to fail.
Our ecosystem is stuck.
I feel that both Free Software and Open Source have a massive issue. They are ideologies that have more in common with libertarianism that's based on individual freedoms, rather than community freedoms based on Human Rights.
Yes Freedom of Speech (code) is covered. Freedom of Association in Safety? The right for everyone to have a habitable ecosystem?
Nah, it's "fuck you, this tool is useful."
@onepict I don't really see how focusing around "community rights" would make a difference, though.
It isn't as if Linus isn't rejecting the political side here, too - even if you feel that is is not communitarian and instead more libertarian.
What Linus is instead telling us is that he was lying about the politics all along. That could have happened under a more community oriented license too.
@yoasif I don't think Linus ever believed or said he believed in it.
I think it's more we lied to ourselves.
The Kernel list has always been what it is.
@onepict I think I get where you are coming from (and I need to read your blog post more thoroughly), but I don't know that we lied to ourselves, except to believe that license actually meant something.
I think this points to an overall erosion of norms too, but the idea that we can't count on the so-called "letter of the law" to mean what it says is foreign to those (especially) who the "law" has previously protected.
I think the danger has always been to count on the letter.
@yoasif For some of us it was never about the licence, it was the idea of the philosophy behind it.
Although we'd not considered what we considered freedom. I think in my case it's more I've shifted, because I grew up under Thatcher.
So my politics and awareness have changed. In the past few years I've been prepared to look at how individual freedoms harm communities.
Now I can't unsee it and I don't want to.
That's not where a lot of Foss folks are at.
@onepict I spent many years expecting a lot of open source. I've spent the last few years continually disappointed by it.
And by it, I don't mean the technical opportunities. Those are still impressive. I'm referring to the structures which arise built out of humans surrounding these projects.
We have a massive governance problem, and in many ways that is downstream from a massive *capitalism* problem (and patriarchy, etc). And I wish I knew how to solve all that, but I don't. 😕
@jaredwhite
@onepict
Software and FLOSS are cultural artifacts made by crowds.
Every attempt to separate those folks from other social places (like with elitist stance) pave a path to a throne.
We need to better consider social configuration like at café Karanfil (Berlin), café Theodore (Lokemo) and many others around the world.
I talked with fundacion karisma https://fundacionkarisma.org/
former director about how now she's involved into knitting and repair meetings with youngsters while she went from human rights, privacy and floss.
And I'm also thinking about many organisations into apc.org network.
I mean : right now mainstream floss scene is totally western biased and We are paying a heavy price for this individualistic libertarianism at the courts of the various dukes and kings.
But in many liminalities various beautiful social configurations are doing great things.
(Sorry it's quite a bit chaotic toot but I'm with our newborn right now 😊).
I love how all these open source projects are accepting LLM generated stuff, and when called out on it, they will quickly mention that "this is not the place to debate the ethics of AI".
So my question for everyone saying that is... if not now, when? When will we discuss if the source of the thing you're trying to ram through our throats is acceptable? When will we discuss if you normalising the damage the LLM companies are doing to create these tools is acceptable?
Or are we just going to ignore all the bullshit because it is (according to you) "useful"?
I bet people thought labour from kids sent to the mines was useful... Or that the results from inhumane studies were also useful...
So, I ask of you again, when IS the time and place to discuss the ethical ramifications of you using these things? Because it never seems to be the time for it somehow, isn't that interesting?
"this is not the place to debate the ethics of AI"They did debate the ethics of AI, and found them acceptable. The debate was not public, possibly not even conscious. But a decision was made: AI was previously not used, and now it is.
So, the subtext of a statement like this is they are asserting power. They proclaim that your opinion has no place in the decisionmaking process about AI. They proclaim this because they believe they have the power to exclude you, and are attempting to exercise that power.
That's one way I read such statements, personally.
So if you lurk in any AI subreddit, discord, forum, or issue tracker, its pretty obvious that the answer to "how do people trust these things when they make mistakes literally all the time and its impossible to know when" is that there is a general conspiracy theory that whenever the model does bad output it is the AI company nerfing their model on purpose: for anthropic, its that they are compute strapped. For openAI, getting sued or cucked. Google is that google just sort of fucks their own shit up all the time, same with Microsoft.
This kind of post basically dominates all of these forums. That systemic failures are one-off byproducts of the momentary malevolence of the companies. The gambler only complains about the house always winning when the house wins bigger than usual.
Oh and for the local models the conspiracy theory is hitting invisible censorship walls
@jonny this one is almost reasonable (and to some extent documented and proven by "benchmarks"), it would be surprising if Chinese models didn't at least have a strong bias towards censorship
@dhd6 oh yes of course they are, in the US we just call the censorship on LLaMA "alignment." i'm referring to the conspiratorial story that explains poor model performance rather than specific topical censorship: when the model behaves badly, it's because somewhere there is a "scar" in the model from censorship that influences its behavior on distant tasks.
@davidgerard
@dhd6
Paraphrasing, the local model ppl are indeed a lot more canny and less mystical about it. I guess you could also summarize it as like "long range consequences of getting a free lunch," like surely they must have spiked these things in some way, we can't be getting this for free without some hidden cost, etc. I mean we actually did see this even in the agent honeypot usage of the qwens, where topical refusal from guardrails/censorship sort of bled into its actual responses sometimes
The other day, I was reflecting how Rowlings' TERFiness blends neatly with her support of fascist politics.
The reason I was reflecting on that is that there is a whole generation of people who grew up with the HP franchise, and have consequently opted to get tattoos with the Deathly Hallows symbol.
I realize that having such a tattoo in no way implies supporting Rowlings' shitty politics. Nonetheless, seeing that symbol, especially as a tattoo, now gives me very icky vibes.
So... too soon?
Alt...
Deathly Hallows symbol in a white circle on a red background, reminiscent of a Nazi flag.
Let's make this a poll.
| Too soon/too much: | 2 |
| Goldilocks (just right): | 7 |
| Too late/too little: | 5 |
| WTAD?: | 1 |
Closed
@jens TERFs will unironically fly this flag before the end of the month. Mark my words.
@schratze I am not sure whether I think that's good (self-selection makes it easier to identify) or bad (I do not want to see this in actual use).
I'm going to give you a pass on my "disengage from accounts that do fascist's work by posting unCW'd fascist imagery" stance, on the technicality that this is speculative fiction.
Which might be worse.
@idlestate Yeah... I actually get how you feel about that. Thank you for the pass.
@jens @idlestate I was never into Harry Potter (already too old by the time it became popular) but am of the exact age when fash groups caused a lot trouble in UK and immediately saw that as fash imagery.
Also the aesthetic of the fictional Hammerskins from Pink Floyd s "The Wall" got eagerly adopted by actual fash well into the 1990s and 00s
If you have this tattoo, cover it up as an Eye of Providence. Better to be assumed a Freemason, I guess.
@jens good coverup:
Alt...
the klingon emblem from star trek. looks like a large spike with two talons.
@jens another coverup idea:
Alt...
Pictograph of Cipher (a triangle with an eye in the middle, a bricked body with bowtie, stick figure arms and legs, and a top hat) from Gravity falls, surrounded by symbols (glasses, ?, bag of ice, open mouthed fish?, pine tree, pentacle with swirly, 6 fingered palm, llama, shooting star, crossed heart). Bottom is a table of symbols and corresponding letters.
@jens Symbols are just secret handshakes for cults. On that note, any symbolic tattoo gives me the ick
@jens i guess, to some extent, Death of the Author; which isn’t to say that continued patronage doesn’t definitely contribute to the success of JKR’s politics.
Though JKR got nothing from a tattoo, other than perhaps keeping the fandom alive. Which could be argued as indirect money as long as copyright applies.
I’m not too versed into the HP universe though, so unless the Deathly Hallows symbol is an in-universe apology of fascism, I wouldn’t personally be too harsh on HP fans.
@geist No... and yes, in the sense that HP falls victim to the same issue that most (but not all!) stories do.
It's a symbol representing three items of power (the Deathly Hallows), and HP acquires them for his fight (TL;DR). In a sense, it's the opposite in-universe.
But of course, the whole HP narrative follows the same pseudo-fascist saviour narrative in which a humanity is doomed until some prophesized person frees them. So out-of-universe, it at least is fit to induce some eye-rolling.
@abucci Hadn't seen this yet. Thanks! $48 for the ebook... I will have to wait till it's available on Anna's Archive. The best biosemiotics-inspired book is still Terry Deacon's "Incomplete Nature." It already describes pretty well what kind of theoretical biology we currently need: one that worries about the dynamics of self-manufacturing living systems...
Duh. It's not even particularly hidden on the website, as these things sometimes are ;)
Got it now. Thanks!
@abucci There are some interesting essays in there. And some complete nonsense, like Froese's irruption thingy. A mixed bag.
I'll have a closer look over the summer.
I don't find it surprising that people send LLM-generated docs, publish LLM-generated blog posts, etc. Writing is a chore and there's this big button labeled "get it done".Writing can be many things, but a chore? That's not something I entirely understand. I don't mean to sound like I'm hollering at this single point from this person's post, but it's a sentiment I've seen expressed before and this was a convenient example (and what I say next is in the spirit of and amplifies the further point they make!)
There's an emerging theme in my last few posts today concerning means vs. ends. Science as a production process that spits out knowledge products. Writing as the necessary chore sitting between a person and "having another blog post" or "citation count go up".
Overfocus on the ends can lead one to think of the means as chores. Science is slow and boring, let's use AI to spit out the Good Science Stuff faster. Writing is a chore, let's use the AI to spit out Documents. Heck let's chain them together and have the machine do Good Science Stuff and then spit out Documents about it that we feed back into the machine and hey we can send the Documents to Funding Agencies to get Funding for our Good Science Stuff and spit out even more Documents so that...and...to...so...[infinite random string]. Wait what were we doing again?
Personally I think it's poisonous to focus mostly on ends. The absurdity is a clue. If I believed in souls I'd say it's bad for them. I guess I find the mindset confusing because long ago I came to think that the means are the ends. There are no other ends but the means.
Yes, I well understand there are certain kinds of mandated writing that are tedious and unwanted.
@abucci I love writing.
I did find coding tedious and not fun, so I stopped. Java and OOP took my joy in coding.
But I've always been more into people and stories.
I still find myself appalled at folks using the LLMs though.
It's so anti social and to me, it seems a rather mean, lonely thing to do.
You're missing out on genuine connection and collaboration using them. That's before we get to the environmental and political issues with this tech.
@abucci We have technical writers who can organise documentation. We have writers out there who'd love the work.
Like I said anti social.
@abucci I'm still getting over the idea of folks considering writing a chore. Is reading a chore as well?
They can go away, far far away.
Writing and reading other folks work should be genuine connection, even when separated by space and time.
Even a fun, well written technical work is connection.
Some folks need to try and remember what that feels like again.
The proliferation of AI tools in science risks introducing a phase of scientific enquiry in which we produce more but understand less.From https://www.nature.com/articles/s41586-024-07146-0
There are good points in this article, but the framing of the risk above is broken, to my way of thinking. In adopting the economic metaphor of production, it's taking for granted that there is a scientific process that leads to products, and that we can quantify how many of these products are produced. "We produce more".
I would argue that if we do not understand whatever is coming out of this process, then they are not products in any reasonable sense of that word. In fact I'd say this of any putative commodity: if a factory spits out a widget nobody knows how to use, then the factory has not produced a commodity, it has made an object. Commodities are commodious, adapted to a use or purpose, and human beings are the entities that have uses and purposes.
So, if we create machines that make knowledge-objects resembling the products of science that fewer and fewer people understand, we are not engaging in "scientific enquiry" or "knowledge production". In this view the phrase quoted above ought to be re-cast as "The proliferation of AI tools in science risks science". What exactly is enquiry for if not understanding more?
A font that humans can read but AI cannot
TIL I'm an AI, because I can't read that shit. Okay, okay, I can, but I refuse to. This ain't the solution.
I mean, if it is a font, and you use it as such, the LLM will still read it, because it cares fuck all about the font - it looks at the text.
If you use it to generate a video, then it's not really a font, but a video generator, and why the heck would a human read anything worthwhile that's a moving, eye straining, noisy video?
@algernon it's ableist is what it is.
You don't battle against a fascist technology by excluding folks.
I hate fancy fonts that aren't screen readable, they aren't particularly readable for me either.
#AI #SpaceX #hype #billionaires #economy #finance #investors
New preprint! So honoured to have worked with Ishani Ray on providing vital critique on "AI literacy" & correctives for those interested in understanding why "just use the newest version of an LLM" is nonsense
Contra Literacy-Laundering: Mechanistic Critical AI Literacy
https://doi.org/10.5281/zenodo.21222978
1/
Alt...
Critical discourse on large language models (LLMs) has bifurcated between epistemic
dismissal that invokes some form of the stochastic parrot metaphor to puncture hype,
and pragmatic accommodation that treats LLM capability improvements as grounds for
updating the critique. We argue that both misdiagnose the problem as the issue is
not whether or not LLMs work, but what kind of working is happening and at whose cost.
Drawing on meta-theoretical frameworks of cognitive science, feminist labor analysis
and critical pedagogy, we propose a conceptual reorientation. We develop this claim
through registers of (i) the cognitive, examining what is forfeited when statistical pattern-
matching substitutes for the iterative, grounded processes that constitute thinking; (ii)
the pedagogical, examining how “Al literacy” as currently deployed is itself a symptom
of the confusion it purports to address; and (iii) the political, examining how the
infrastructure framing of Al naturalizes asymmetric labor displacement, particularly of
feminized cognitive and reproductive work. The stochastic parrot, deployed with
mechanistic precision rather than mere rhetorical convenience, specifies what is
forfeited when cognitive labor is delegated, who bears the cost, and why a literacy
adequate to this moment must begin from the epistemology of those most harmed by
the systems it describes. We conclude with underlining that critical Al literacy, which this
paper embodies an instance of, is the only sensible way forward.
Maybe many mid-career and above academics have my experience, and so can relate, but it never ever gets any less impressive, overwhelming, or old when a student appears our of nowhere (notice our geographic distance is big as are our official fields not that overlapping) and does amazing work! 🤩
2/
Anyway, back to the work itself (from the abstract):
> Critical discourse on large language models (LLMs) has bifurcated between epistemic dismissal [...] and pragmatic accommodation that treats LLM capability improvements as grounds for updating the critique.
https://doi.org/10.5281/zenodo.21222978
3/
> We argue that both misdiagnose the problem as the issue is not whether or not LLMs work, but what kind of working is happening and at whose cost. Drawing on meta-theoretical frameworks of cognitive science, feminist labor analysis and critical pedagogy, we propose a conceptual reorientation.
4/
We analyse this through
1️⃣ the cognitive: what is forfeit when statistical pattern-matching substitutes our thinking
2️⃣ the pedagogical: AI literacy as currently deployed is a symptom
3️⃣ the political: AI's infrastructure inherently naturalizes labor displacement
pdf: https://zenodo.org/records/21222978/files/RayGuest2026.pdf?download=1
5/
> The stochastic parrot [...] specifies what is forfeited when cognitive labor is delegated, who bears the cost, and why a literacy adequate to this moment must begin from the epistemology of those most harmed by the systems it describes.
https://doi.org/10.5281/zenodo.21222977
6/
Oh, if people are also on LinkedIN (I am not so please I have no idea what is happening on there) — she is here:
https://www.linkedin.com/in/ishaniray0529
send her appreciation or whatever of the nice stuff people do on there ☺️
7/
@olivia This is the best fucking paper on the subject I have read in a while and elucidates phenomena I had personally recognised but had not spelled out so comprehensively. Huge W to her for writing this
"We develop this argument in four moves:
1) what gets labeled a literacy problem is typically the technology working as designed;
2) AI systematically resists the analytical methods critical literacy assumes;
3) the labor literacy-laundering most consistently forecloses scrutiny of is feminized and racialized cognitive labor; and
4) the practical consequence of taking that seriously is not better literacy but principled un-training"
so much this! only ~1/4 thru, this is amazing 💜
@olivia btw, a colleague and friend just published this a few days ago
http://arxiv.org/abs/2606.28574
@danielscardoso @sejkko Pita is a hilariously perfect surname for somebody who critiques the mainstream 🙂↕️
Feels like all of us (even people w/out a car) need to pay way more attention to the WTAF new rules affecting EU and US folk, requiring all new cars to include a driver monitoring camera aimed at your face.
"Glance at your phone, your kids in the back seat, or the radio for too long, and the car will flash a warning light and sound an alert," reads a recent post from Allaboutcookies.org.
https://allaboutcookies.org/eu-mandatory-distracted-driver-system
First spotted this story in the Risky Business newsletter yesterday and have been reading as much as I can about it ever since. https://risky.biz/RBNEWS587/
Risky's @campuscodi writes the new regulation has entered into effect in the EU on Monday and will enter into effect next year in the US. In the EU, the new camera requirement is part of the block's second General Safety Regulation (GSR2), a broader swath of new safety rules introduced for the auto industry and designed to improve road safety.
https://www.inkl.com/news/new-eu-car-safety-rules-take-effect-7-july-how-your-car-could-monitor-you
https://www.gadgetreview.com/federal-surveillance-tech-becomes-mandatory-in-new-cars-by-2027
This is some next level Orwellian shit. My dreams of one day replacing one of our cars with an electric vehicle just got much darker.
@campuscodi Somewhere out there I imagine is a secret govt manual on how to disable all of the radio telemetry points, transmitters and wireless doodads in modern cars. You know, for all the important gov types who probably shouldn't be super trackable by everyone. Assuming such a compendium does exist, it's probably gotten so thick it requires multiple binders. I just wanna borrow one page from that book lol.
@briankrebs @campuscodi My assumption is that any Waymo vehicle (or any robotaxi for that matter) could be effectively disabled by jamming its comm links. Right now I assume these are conventional mobile channels, though I wouldn't be surprised if they don't start using satellites too, if they haven't already, at least in testing. Once they're cut off from Central for more than a relatively short period of time, they're going to be pretty much helpless, and I suspect will pull over and stop, assuming they don't just stop in the middle of busy intersections the way Waymos are known to do.
@lauren @briankrebs @campuscodi btw: NTN non terrestrial networks is an option for mobile since 5G
@lobingera @briankrebs @campuscodi Sort of. But connectivity at the data rates required would be iffy. Not impossible.
@lobingera @lauren @briankrebs @campuscodi It's more of a wet dream of 5G vendors.
@jens why?
@lobingera Because a lot of what "space" actually means is lost on folk in the 3GPP and/or IETF. And the space industry, at this point in time, is (I soft-quote an anonymous source) "never going to let those cowboys run comms".
Case in point - the relatively newly formed "deepspace" working group in IETF is basically focused on QUIC.
That's fine in itself.
The problem is, the end-to-end principle isn't fit for deep space comms, because an end-to-end connected path doesn't exist when some...
@lobingera ... nodes in the middle of the path are on the wrong side of a planet. So making an end-to-end oriented protocol not be end-to-end is, to put it mildly, a challenge.
Folk are smart. They will figure it out.
3GPP on the lower parts of the stack could easily apply some tech to space, the higher parts that look more like the Internet probably don't work so well.
The TL;DR is, something LEO like Starlink is a decent enough fit. MEO becomes dodgy, even if there have been...
@lobingera ... successful *demos* of 5G between the ISS and ground. But the further away you get, the less likely these things will work out.
For larger distances, latency and intermittency become so huge, that the notion of real-time interactive comms is out the window. So trying to cram DNS or HTTP into that framework is wonky. Never in principle, but always in practice.
You cannot build something like a website consisting of a main page (DNS roundtrip plus HTTP roundtrip in sequence), and..
@lobingera ... then have it refer to external resources (HTTP roundtrips) that may each be hosted on CDNs (DNS roundtrips) when latency means that even in the best case, each roundtrip can take minutes.
You need more of a bundle paradigm: this is what I want, now send me all of it together, in a single transaction.
Can you do that on top of the current tech stack? Sure. But it's not what anyone does do. Absolutely everything is designed *not* to do that, in fact.
TL;DR the 3GPP and IETF...
@lobingera ... style approaches will not scale with distance, which means a huge, long adaptation curve of fundamental philosophies is required.
Meantime, CCSDS (the consortium actually working on deep space comms) plods on and on and on.
![[?]](data:image/png;base64, iVBORw0KGgoAAAANSUhEUgAAAEAAAABAAQAAAACCEkxzAAAAUUlEQVQoz43R0QkAMQwCUDdw/y3dwEvsvzlL4X1IoQkAisKmwfAFT3RgJHbQezpSRoXEqeqCL9BJBf7h3QbOCCxV5EVWMEMwG7K1/WODtlvxAYTtEsDU9F34AAAAAElFTkSuQmCC)
update-notifier-crash on Ubuntu 24.04 wrote 1 Tbyte of garbage to syslog before I noticed and stopped it. I noticed because it filled the disk.Unbelievably annoying behavior. I guess that'll teach me to ignore an update notification rather than figuring out how to disable it forever.
Only a mind so subdued by slop and propaganda and advertising, a mind unpracticed in observation and inquiry and imagination—only such a mind could be complacent as its pockets are picked to fund that most terrible of horrors.Into the Gap
On the one hand, they threaten us with no work at all, with the misery and penury that comes from a lack of work, and therefore a lack of the means of living. On the other, they demand endless work, a work that wipes out all other avenues for thinking and being, that leaves us programmable and programmed, no space left in our minds for thoughts they haven’t placed there. Are we to merely acquiesce, to accept their scraps and the miserable conditions attached to them? Surely not. For if we accept these conditions, will they not impose even worse upon us? Will they not keep increasing their demands and decreasing our pay until we are working ceaselessly, and for nothing? What would compel them to stop? Already we have seen that their greed for money and for power is so voracious it will tear through buildings and through bodies, it will murder many people, it will poison the air and the soil, it will bring great storms upon us. So there must be an end, and it is only we who can bring that end about.
For there is one weapon that only we possess and which the billionaires and the warmongers can never take from us. One weapon which so frightens them they will twist their words into knots, they will spend the entirety of their vast fortunes trying and failing to convince us that we don’t possess it at all, they will claim over and over and without evidence that it is vanishing before our eyes even as it remains right there in our hands, clear and plain to hearts yet open to the world: the refusal to work.
From human hands.
We don’t use AI in any of our design or production processes. Here’s why.
#AI #GenAI #GenerativeAI #ImageGeneration #fonts #typefaces #art
I realize this is beside the point, but personally I view the fact that code review is not considered interesting as a serious failure of computer science education. I'm invested in the latter, so it stands out for me.
On the whole writers tend to love to read. I think they love it in part because their education led them to realize that learning to read well is both a good in itself and helps you become a better writer. Programming languages were designed to be human readable so that humans could read them. Yet we have a couple generations of people writing code most of whom were never explicitly taught how to read it, let alone learn to love reading it.
#tech #dev #software #SoftwareDevelopment #ComputerProgramming #ComputerScience #ComputerScienceEducation
@abucci "Coders at Work" is a worthwhile read in this context.
@abucci no need to apologise. I read it before any of what you mention was evident.
In any case, what I was struck by on reading the book was how most of these 'luminaries' admitted they don't read other people's code.
Judging from the state of the computer science literature, it seems like the disinterest in reading extends to natural language!
@abucci @theluddite oh God don't get me started. I know perfectly smart, thoughtful people with impeccable politics who use whataboutery to justify their use of the slop machine and can't tell the difference between a doomer and a critic like you or me.
I've been on an educational quest for some time. It's exhausting work, especially when dealing with well-educated people, because they are used to effortlessly confabulating their way out of cognitive dissonance.
@abucci coming back to this: the brilliant coders I've known did read other people's code, effortlessly.
I think it is genuinely a question of literacy. Low levels of code literacy are acceptable in the industry. The incentives in recruitment push programmers to demonstrate they can leet code, not that they can quickly read and understand code, including nuance and subtext. (Yes, of course code has subtext.)
Low levels of code literacy are acceptable in the industry.My experience isn't representative, but that's something I've observed in industry as well. Code that isn't understood tends to be treated a bit like a cyst. The trouble is that over time the codebase becomes more cyst than functional.
I've been in the position to hire software developers multiple times over my career, and I eventually converged on a pretty simple technique. For the technical interview, I'd ask them to bring code they've written and that they understand, and explain it to me. It was intended as a literacy test. The trouble is the interviewer needs to be relatively code literate to understand what they're observing in such an interaction.
I had a technical interview long ago where the interviewer sat me in a room, gave me a printout (!) of code, and asked me to highlight all the issues I could find with it. When I was done we walked through it and I explained my observations. In retrospect that was a literacy test, too. Using a physical printout is a brilliant move.
rsync thing has me thinking about the difference between the "free as in beer" model of free software and the "free as in freedom" model the FSF originally put forward. There are important differences, and it's meaningful when a project changes from one model to the other, which is one way to narrativize what's happened to rsync.I think "free as in beer" software is more appropriately called "Libertarian and Open Source Software", or LOSS. Among other things it involves no obligations on other people, and expects no obligations in return (*). But this means it's an everyone for themselves, anti-communitarian endeavor. We should be honest about this. We should also be honest when a project changes state from FOSS to LOSS, because that's a meaningful change, and actually is a loss in my view.
I continue to believe that enthusiastic embrace of AI/LLM coding tools is a leading indicator of an impending FOSS→LOSS transition.
It seems to me at least one driver of the uproar over rsync is that a bunch of people thought it was a FOSS project but it had, in a short period of time, converted into a LOSS project and the shock was jolting (understandably). As a FOSS project, rsync served a community function, leading people to depend on it, make assumptions about its longer-term stability, and so on. Those are reasonable expectations of a communitarian project. As a LOSS project, rsync is under no obligations to anyone and the maintainer(s) can change the software at any time, for any reason, at their discretion alone. It seems rsync is a LOSS.
#tech #dev #software #SoftwareDevelopment #FOSS #freedom #rsync #AI #GenAI #GenerativeAI #LLMs
@abucci @Robotistry @onepict MOSS: Mutual(ly supportive) Open Source Software
@jens @abucci @Robotistry I mean that sounds nice and soft. I like MOSS.
I'm not a communist, I'm more of an anarchist.
I always thought we worked in the commons for mutual benefit of the wider community. For the betterment of society.
Discovering that there's a whole branch of folks in FOSS that don't believe in the social contract makes me sad.
Discovering that there's a whole branch of folks in FOSS that don't believe in the social contract makes me sad.Yes, me too. I'm not as directly involved in FOSS as y'all are, but I do contribute to open source projects and I strongly believe in the commons aspect of it. Personally I think there are (too many) people who view software as a path to power, and go power mad over it. I use that to explain to myself the phenomenon of people "descending" into AI, becoming "AI pilled" or "oneshotted" or whatever you like to call it. They sense the proximity to power and they want to wield it so, so badly, as though it's a Ring of Power. Too bad Sauron has the One Ring and they're only securing their future as ring wraiths. Someone must have clocked LLMs as Nazgûl tech at some point.
MOSS is so much more beautiful in so many ways. Lovely metaphor. MOSS >> LOSS. Hobbits like MOSS, they put it on their homes and it's nice to walk through if you go about without shoes.
@abucci Somewhat unrelated, there is #mosstodon as a hashtag. Maybe I need to run mosstodon.org now 🤔
@jens @abucci @Robotistry or mossmoot.org
A moot like an entmoot. Then we aren't interfering with the mosstodon tags.
@abucci
"They sense the proximity to power and they want to wield it so, so badly, as though it's a Ring of Power. Too bad Sauron has the One Ring and they're only securing their future as ring wraiths. Someone must have clocked LLMs as Nazgûl tech at some point."
Re-reading this morning and yeah 🎯 the LLMs are turning folks into ring wraiths.
MOSS, on the other hand, can cover all given the right conditions.
@abucci @jens @onepict I seem to be completely failing at my intent of contributing a little bit of unserious fun to your entertaining thread.
I understood your original post to be pointing out that under current LLM-centric conditions, free-as-in-speech projects have been conflated with or co-opted into free-as-in-beer projects while continuing to be described as FOSS. It seemed a shame to me that you had a nice acronym for the less desirable libertarian side, but there was no equally silly/nice acronym for what you described as the contrasting "communitarian project", so I thought I would contribute one.
@Robotistry @abucci @jens It's all good, we're having fun 😊
Although even the communalism side has it's different flavours. 😊
It's under the GPL v3. Which is a license that cannot be revoked.
- If a new version is buggy, you can use the previous version.
- If the current maintainers do a bad job, you can fork it.
- If development goes inactive, you can fork it.
- If you disagree with the trajectory of development, you can fork it.
- If new versions start coming out under a proprietary license because every single non trivial contributor agrees to this (that's the only way under GPL) and thus it actually stops being free software going forward, YOU CAN STILL FORK IT (the last GPL version).
Free software has a definition. Let's stick to it. Let's not confuse people by claiming that something has ceased to be free software, when it clearly is still free software, just because someone doesn't like some new version, or the recent work of the main maintainer who has thanklessly developed it for years and made it available to the entire world.
I freaking love rsync and have like three daily automated scripts that run it. For a moment you really made me think it's gonna stop being available or something, until I went and checked the license!
Thank GNU it's not going anywhere because it's GPL.
It's absurd to claim "freedom" boils down to a license document. That doesn't make any sense.
@abucci @taylan Could you please link the original post? I'd love to read some of your arguments about the important differences between FOSS and LOSS.
So far - not having read them - I am basing this discourse on your relabelling of "free as in beer" as "LOSS" and Stallman's definition of gratis ("free as in beer") vs libre. That is, if you have the right to modify and re-use it.
What is happening to the rsync project is that the developer has turned the latest version into proprietary software, denying the user 3 of the 4 freedoms.
The only solution to the problem is to keep using an old version, or fork the project from the last free version.
What is happening to the rsync project is that the developer has turned the latest version into proprietary software, denying the user 3 of the 4 freedoms.Again I will repeat: documents are not "freedom".
The only solution to the problem is to keep using an old version, or fork the project from the last free version.That is not the only solution, and I view the reductive thinking implicit in this as problematic. This is not meant to be mean, but I really think FOSS went off the rails when it stopped understanding itself as a political project, and where we are today reflects a serious loss of vision. Like, we have a tremendously important, widely used piece of software that permeates Linux distributions and is probably relied on by many millions of people (some of whom are not even aware they rely on it!) and our only response to a single individual claiming ownership over it is "fork it"? That's it? That's our only move? Frankly that's pathetic.
Trying to be neutral between the corporate bootlicking of "open source" and the freedom of free software is just sad.
The only appropriate response to a program that was free software, being rendered proprietary, is to fork it back into free software.
At the moment, my time is being wasted, as I inadvertently installed the proprietary version of rsync, breaking my servers, as unsurprisingly proprietary slopware does not work!
I shall fix the problem with fury.
@Suiseiseki @abucci @taylan LLM generated text is not and cannot be subject to copyright thus inclusion or non-inclusion of that is irrelevant to licensing consideration.
I see no reason why the original copyright on the copied text wouldn't be under the original license (why would passing it through a LLM change anything?).
If it's not subject to copyright, and the user doesn't have a certificate, or at least a statement that x parts are in the public domain, then how can the user confidently exercise the 4 freedoms, knowing that some company won't come along and sue them for copyright infringement (after all, if a company realizes that if parts of the source code of their proprietary software has been copied into another program, the company will sue)?
I think part of "the problem" is that this is what Free Software is, but we sort of fooled ourselves into thinking it was MOSS.
Until people came along, and showed, repeatedly, that the definition of FOSS is actually kinda LOSS.
Which leads me back to something I keep repeating: it's kind of time for FOSS/LOSS to die. Let MOSS grow over its ruins.
In this video, Hank Green considers human demand for code, and how that might change as LLMs make coding cheaper.
Will we reach a ceiling, with all our software needs covered? Or will we discover countless new places and ways of using code that would have been unimaginable before?
It's an interesting question, and one that worries me.
I think Hank's right that we will discover strange new applications, and I think we're already seeing that.
I also think that we had fully satisfied the everyday software needs of most people back in the early 2000's. The last new software that really excited me and changed how I live my life was Google Maps, back in 2005.
In my lifetime, most software innovation has been about injecting tech into our lives so it can hitch a ride and extract value from our daily activities. It monitors, manipulates, and intermediates, and this has been extremely profitable.
What does that say about the future headroom for code? Not sure.
@abucci Sorry. 😝
I dunno, as internet personalities go, you could do a lot worse.
He's problematic in other ways but I can't be bothered to dig up the links.
@abucci I mean, I've noticed. Kurzgesagt in particular really pissed me off with their bad take. However, growing up on a steady diet of PBS has left me a great lover of sciency edutainment. I seek it out, I enjoy it, and I think I'm a richer person because of it.
I just can't watch it uncritically, because sometimes it's wrong. But isn't that always the case?
The three creators cited in that video have their problems, but they are, in my estimation, among the best in this category. I see you'd rather we didn't have this category at all, but I fear it plays a pretty important role in educating our society (especially with PBS being gutted by fascists).
Note, I did not generally recommend this video by Hank. There's a lot I dislike about it. But that question of "how will LLMs change the demand for code?" was new to me, and seemed worth engaging with.
"how will LLMs change the demand for code?"Regardless of where you stand on any of what follows, I submit there far, far superior sources to someone like Hank Green if this is genuinely a question that's of interest.
Green is paid by organization(s) aligned with the AI industry, and has demonstrated he will frame his content around the ideas these organizations are pushing. The AI industry has a very specific take on this question in furtherance of their own goals. This show/channel/person could not be more biased, in that sense. I view them as a force of miseducation. As you note, it works by superficially aping entities like PBS that people trust, and functions to "science wash" content that has a manipulative purpose.
PBS has its issues, no doubt, but it's a member-station-supported non-profit organization with federal licensure requirements. PBS is supported by member station dues, some of which in turn come from individual donations to the member stations. Till 2025 it was also funded by CPB, which was formed by an act of Congress in 1967. All of this entails a fair amount of accountability for content. It is a categorically different organizational structure from "science commentators" on Youtube, and treating the content of the two as similar because of superficial resemblances in production quality is very dangerous, in my opinion.
I missed reading Starlink's latest conjunction report when it came out a bit over a month ago. I just skimmed through it and I think I need to go lay down for a while. It's terrifying how close we are to major collisions in orbit all the time... (I especially love the note about how space-track.org being offline briefly caused them to miss a potential collision... SO FRAGILE AAUGH)
Article summarizing the report here: https://ca.pcmag.com/networking/16653/260-starlink-satellites-burn-up-in-earths-atmosphere-as-more-head-for-fiery-ends
Full report here: https://www.scribd.com/document/1057502572/SpaceX-Gen1-Gen2-Semi-Annual-Report-7-1-26
@sundogplanets Are you aware of any organizations working to reduce the chances of Kessler syndrome happening above us?
@sundogplanets say it does happen with a couple of these star link satellites? It'll be a cascading event, but how long will it last for? Months? Years? Decades?
@sundogplanets Would be a "fun" calculation to do to see, based on our understanding of what Aluminum oxide does to the ozone layer, how many deorbiting starlinks it would take to destroy the ozone layer entirely!
@startswithabang Some of the foundational work for that is here: https://arxiv.org/abs/2510.21328
@sundogplanets Bah, science in its infancy is one of the most promising, and frustrating (for its lack of definitive answers), types all at once!
@sundogplanets
15000 satellites in the array when complete, 5 year lifespan = 24 satellites deorbiting per day.
Wait until there are 1000000 servers in orbit yielding 500 or so deorbiting per day
Oh gosh I just did the math on the maneuver rate, which I couldn't bring myself to do earlier. More than 207,000 collision avoidance maneuvers in 6 months.
That means that somewhere in the Starlink megaconstellation, a satellite is performing a collision avoidance maneuver EVERY 1.25 MINUTES (EVERY 75 SECONDS)
AAAHHHH I'M SURE THAT'S FINE.
And now I find myself reading about asteroid collisional cascades. For no particular reason...
I mean, to be fair, they have done it perfectly so far. No Starlink collisions.
But there is a shitload of untracked, lethal debris orbiting around up there. And they are completely dependent on so many systems (like space-track.org, apparently).
And our future use of orbit is completely dependent on Starlink operating perfectly, every minute of every day, forever.
Like their last report, they also have a Shit List of which objects Starlink had to maneuver for the most often, and complained a lot about it. "SpaceX coordinated with other satellite operators hundreds of times over the reporting period"
What they fail to say is that this is because anyone who wants to travel to higher orbits is REQUIRED to coordinate with them, because they effectively control Low Earth Orbit.
@sundogplanets This is very 'complaining about how bad traffic is without considering that YOU are the traffic' and hence the problem.
@debcha That is pretty much the tone of every single one of these reports. It would be funny if it wasn't so dangerous. "We are annoyed that we are the only operator who is required to do this reporting" (Yes, you own 2/3 of ALL SATELLITES that's why)
Which objects are those? Asking for a friend.
I read while Envisat 🛰️ 😈 operates in a higher orbit, for polar shell Starlink satellites the upper stage or the satellites themselves may cross Envisat's altitude band during initial deployment.
They also have a list of deorbits, I haven't managed to get the pdf from the FCC (because their website fucking sucks, on purpose I'm sure). And scribd wants me to pay to download it.
Anyway, the PC mag article says 260 Starlinks reentered. At some point I'll count up the gen 1 and gen 2 and get a mass estimate. But that's 1 or 2 a day. And a lot of new weird metal in the stratosphere. Thanks, SpaceX.
They also list 4 "disposal failures" which are satellites that died before SpaceX purposefully chucked them into the atmosphere. This includes Starlink 34343 which either exploded or got hit by debris a couple months ago. https://keeptrack.space/deep-dive/starlink-34343
4 out of thousands of launched satellites is pretty good. But when you have nearly 11,000 satellites, you have to operate COMPLETELY PERFECTLY every minute of every day, forever.
Please don't fuck up more than you already have, SpaceX.
@sundogplanets I feel like there's gonna be a day where they just abandon it all for whatever asinine reason and it rains space junk all over, all at once, with no accountability
@sundogplanets I dont see any reason to freak out about it. They are low orbit sats with natural fallback of 5 years, spacex could go bust and all the sats left and it'd all be gone in a few years.
They all are designed to full burn up before risk to ground and no evidence of them failing to fully burn after hundreds of burns.
The real shit you should worry about is high orbit sats that are 20 years old and actually able to cause Kessler syndrome.
@Rin3d @sundogplanets Do you understand that the person you answer is a specialist of the topic, and if I remember correctly has been investigating actual debris from Starlink satellites that fell on the ground (see https://thenarwhal.ca/space-junk-falling-50th-parallel/, even with a photo of her with space debris)?
I cannot be sure you are a man, but your message looks like a strong contender for the worst mansplaining of the year.
@BrKloeckner Thank you.
@sundogplanets You're welcome. Sorry you have to put up with that, I assumed you should not also have to carry the weight of letting such people know it is not ok.
*looks at Musk's record of SpaceX rockets operating perfectly, and not undergoing rapid unscheduled disassembly at all*
I'm sure it will be fine.
No Starlink collisions.
That we know about.
@sundogplanets … by a company who has recently been merged with a company whose business model is basically selling vibe engineering/vibe everythinging tools to the industry.
Correct me if I'm wrong but, collision avoidance manoeuvres cost propellant. A satellite can only carry a small amount of propellant. More manoeuvres cost more propellant. Nobody is refuelling satellites in orbit.
I'm sure it will be fine.
Systems requiring so much active corrections to maintain a safe state are inherently bound to fail over time.
@sundogplanets Do you know https://outerspaceinstitute.ca/crashclock/ ?
@sundogplanets @wendynather All of the sudden all those asteroid near misses start getting a lot more interesting. Don’t have to hit the Earth, just swoosh by in LEO leaving a wake of destruction
@sundogplanets if a manoeuvre is not taken, what is the probability of a collision? Is this information that is easily accessible?
I have been of the opinion for a while now this is deliberate. The goal is, in fact, to precipitate a Kessler syndrome. It fits with a number of other nihilistic actions M Musk has taken, and his particular attraction to activities in which he gains recompense for doing harms.
@sundogplanets
I’m old enough to remember laying out in the back yard late at night to catch a glimpse of the International Space Station as it made it earthly revolutions roughly every 90 minutes or so. Was just enthralled by the technology. No more, sadly no more. #TooMuchOfAGoodThing
@sundogplanets i have worked for ESA and heard multiple stories about them trying to get *someone" at Starlink on the phone because of imminent collisions and missing agreements on who takes evasive action (settled now). Pretty wild
@sundogplanets Collisions at all will be terrible, but at what point do we reach critical mass for a runaway chain reaction?
@swordgeek Oh we're beyond the limit. Some parts of orbit are already in Kessler Syndrome, but the early stages are slow.
@sundogplanets @swordgeek wait wait what?? That's a lede you have to tell us about!
@fiigvam @swordgeek We're beyond the runaway limit. If there's a collision in altitudes beyond the runaway limit, that means that debris pieces will continue causing more collisions. But the timescale for those secondary collisions could be years. But the rate will continue to increase...
Here's a recent paper by Hugh Lewis and Donald Kessler (yes, that Kessler) that's super dense but talks about it. Figure 16. https://conference.sdo.esoc.esa.int/proceedings/sdc9/paper/305/SDC9-paper305.pdf
@sundogplanets @swordgeek oh interesting! So regardless of the in orbit maneuvers that starlink is actively performing, there is at least some patch of orbit that is in slow but increasing cascade Kessler syndrome?
@fiigvam @swordgeek Yes. Scary.
@sundogplanets It's because of the potential for collision that SpaceX programmed their satellites to deorbit and burn up at the end of their useful lives rather than move into the usual 'parking' orbit.
@swb72 There is no parking orbit for LEO, just for GEO
@sundogplanets Either way, SpaceX isn't just leaving junk in space. (Elon's personal Tesla Roadster is in a heliocentric orbit as a required dummy payload for Falcon Heavy's test flight).
Who would’ve thought a bunch of Rich assholes could kill the sky.
Space will be the final frontier.. When we can get back to launching anything into orbit again
@sundogplanets Thanks for keeping us informed. We've trashed the planet and now we're trashing space. We're literally shitting in our own nest.
#computer #games #gaming #id #IdSoftware #IdTech #Microsoft #XBOX #Doom #Quake
People who've only ever known broadband often imagine dial-up as "just slower internet." It really wasn't lol.
What was practical was fundamentally different. Clicking the wrong link could cost you several minutes and stall your browser. Also tabs didn't exist just additional windows.
A page with a few megabytes of JavaScript is effectively unusable on dialup. Which is... most of them 🙃.
Alt...
Wildfire map of eastern Canada, centered on Quebec, showing the many out of control wildfires.
New blog post: "Conviviality in computational science"
https://blog.khinsen.net/posts/2026/07/06/conviviality.html
"Conviviality matters for science for multiple reasons. One of them is epistemic: if you want to derive knowledge from your work, you need to know exactly what you are doing, and that includes a detailed understanding of your tools."
@khinsen Thanks for sharing this! I wrote an essay called "Towards convivial science" (https://qspace.fqxi.org/competitions/entry/2310). It was for a competition, but wasn't formally published. I've been meaning to come back it in the the context of computation specifically, so really nice to see other takes on this!
@SylviaFysica Thanks for your essay! The conviviality of science as a whole is a relevant topic in particular now as Open Science becomes fashionable - which for some people means more conviviality and for others more industrial productivity!
I am looking forward to your thoughts on computation. In that context, have a look at @stephenrkell's work as well:
https://www.humprog.org/~stephen/research/papers/kell20convivial.pdf
@khinsen drawing the attention of @danmcquillan and @abucci to this thread.
There were only three big simulation programs that everybody worked with [...]. Only a very small number of researchers understood them in detail and could modify them. Everyone else computed whatever the software allowed them to compute, rather than what they actually wanted to compute.
I believe that the reversal of nemesis can come only from within man and not from yet another managed (heteronomous) source depending once again on presumptious [sic] expertise and subsequent mystification.
--Ivan Illich, Medical Nemesis
Scientific thinking—that is, a thinking in overflight, a thinking of the object-in-general—must return to the ‘there is’ which underlies it; to the site, the soil of the sensible and opened world such as it is in our life and for our body—not that possible body which we may legitimately think of as an information machine but that actual body I call mine, this sentinel standing quietly at the command of my words and my acts.
--Maurice Merleau-Ponty, Eye and Mind (translated by Carleton Dallery and Markus Hardtmann)
By the later 1970s Illich had recognized that the moment had passed when the radical change of direction he had urged would have been possible. The institutions whose counterproductivity he had exposed might have acquired a more ambiguous taste for many of their clients, their priesthoods might have lost some of their prestige, and the rituals of progress become tarnished, but this would only translate into a protective cynicism, and the accommodation of a countercultural element within “the system.” What Illich had sought — the disestablishment of all secular churches and an end to their monopolies — would not occur.
--David Cayley, The Rivers North of the Future
@khinsen Beautiful! I had heard bits of your story, but it’s great to see it in perspective and to see the connection with the concepts Illich came up with—which I view as the thing I should finally read!
I also see connections with “Décroiscience” by Nicolas Chevassus-au-Louis, which, while not referring to “conviviality”, advocates for something that looks like it: more agency for scientists and slower, reproducible research.
Alt...
"Gelatinous fruit body of Phaeotremella frondosa", from Wikipedia (assuming I IDed it correctly). Looks like a big pink brain stuck to the base of a tree.
Today I learned that there are only three English cities with an electrical consumption greater than 200 megawatts: London, Manchester and Birmingham. By comparison, the people who want to build AI data centres here, which nobody asked for, want a supply of up to 1 gigawatt (1,000 megawatts) for each one. That’s how skewed this whole nonsense is.
And that’s before we discuss the precious cooling water they’ll use once and dump. Maybe if each data centre provided a city-wide district heating system from their waste heat…
@christineburns OpenAI is about to top that with planned 6GW of energy consumption for running AI data centers.
https://edition.cnn.com/2025/10/07/tech/openai-chatgpt-ai-infrastructure-explainer
I wanted to share my thoughts on the proposed OMB rule change that would radically transform federal grant-making into an overtly partisan political process. It's highly alarming, and would be devastating for American science (among many other things). https://youtu.be/ZyBibk4a6Vs?si=Bg1KCBtDR1wOTKir
Many professional and legal societies, as well as gov/constitutional scholars, have already opined--and there is almost universal alarm. For instance, statements from:
American Physical Society: https://www.aps.org/initiatives/advocate-amplify/policy/federal-grants-rule-change
American Meteorological Society: https://www.ametsoc.org/ams/about-ams/ams-statements/statements-of-the-ams-in-force/ams-summary-comments-on-the-office-of-management-and-budget-proposed-rule-on-regulation-for-federal-financial-assistance/
I strongly urge anyone with strong sentiments on this critical issue to offer a public comment by July 13. You don't need to be a scientist; this rule change would affect everything from disaster response to cancer research--and involves your tax dollars: https://www.regulations.gov/commenton/OMB-2026-0034-0001
2026 AI Inflection Point. The Preliminary Report of the UN Independent International Scientific Panel on AI: Evidence-based assessment of opportunities, risks and impacts of AI is. Preparing the stage for the Global Dialogue on AI Governance, July 2026
@wmijnhardt "this proωees man clambing of Reasoning" — the extruded image provees the point I guess?
Alt...
Extruded image of the typical mediocre style with a lot of obvious problems in the rendered text, e.g. ''this provees man clambing of Reasoning"
“Impossible to predict this”, says the CxO class that keeps ignoring all predictions of this exact outcome
> As one New York financier told Financial Times journalist Gillian Tett, new hires who were seen as “AI natives” are turning out to have alarmingly shallow ideas. So much so, the anonymous finance worker admitted, that his firm now actively avoids seeking out AI-literate STEM graduates, and opts to comb through humanities students instead.
https://futurism.com/future-society/college-critical-thinking-ai
Hi. Any blind people have audio described versions of the 1999 film version of Animal Farm and the one that comes after that? I have the 1954 version. If not, undescribed audio would be fine. Thanks for any help or boosts.
RE: https://ehlo.exim.org/@bernardq/116851178355484115
- Reports we believe are LLM generated will either be rejected outright or if thought to be plausible will be credited to the unnamed and uncredited authors whose works were ingested as the training corpus.
power move
Forthcoming changes to the Exim Project contribution, security reporting and release procedures.
We are updating our contribution, security reporting and release procedures to better align with the new landscape of development and reporting.
These will be incorporated into the appropriate document locations in due course.
----
Contributing
As a mature project we do not expect major changes in the current codebase. Most changes are likely to be limited to security/bug fixes, refactoring for code consistency and very occasionally new capabilities to accomodate new standards (example as of today being the forthcoming DKIM2 specification).
Any sizeable contribution will be considered not only in respect to its conformity with current project standards but also in terms of whether the project developers consider the ongoing maintenance burden to be acceptable.
With this in mind we would ask you follow the following steps to contribute.
* Start by checking and discussing via the exim-user mailing list whether we would consider the feature.
* If we approve an suitable ticket will be created to track the specifics.
* We do not accept LLM generated content in any form whatsoever.----
Security Reports
* We only accept reports against the latest release.
* Reports should be limited to succinct descriptions of the problem, with minimal code inclusion.
* Reports where EXPERIMENTAL builds flags are used must be filed as ordinary bugs and are not considered security issues.
* Reports against isolated source files builds are not accepted. Only reports against the full binary are considered.
* Do not include Proof of Concepts
* Do not include suggested patches.
* Do not include extended diatribes about code flow.
* Reports we believe are LLM generated will either be rejected outright or if thought to be plausible will be credited to the unnamed and uncredited authors whose works were ingested as the training corpus.
* Reports that do not meet the criteria for security issues but are merely bugs will be turned into normal public issues.
* If we do determine there is a security issue then we will release an update on the relevant branch as soon as there is a tested fix. Each issue will be allocated a GCVE identifier against our GNA ID, we will no longer be using legacy CVE IDs.----
Release Procedures
We will be changing some of our release procedures. Starting with 4.100 we will change to releasing only 4.yy and 4.yy.zz releases on the master branch.
Intermediate work such as general bug fixing and feature enhancement will move to the 4.next branch.We will endeavour to apply any .zz security fixes to the 4.next branch in a timely manner though there may be some lag.
Release candidates for 4.yy+1 will be tagged on the 4.next branch.----
These changes will allow us to allocate our limited resources more efficiently within the current environment. Whilst we might fine-tune some of these changes we will not envisage any significant changes.
The announcement by Meta that they'll start selling AI compute potentially broke one of the pet narratives of the AI bubble: compute is so scarce that we need to spend all the money to build more. That story is the justification for crypto-turned-AI companies with poor fundamentals like CoreWeave to continue existing.
Nothing about stock or commodities prices makes any sense anymore so maybe tomorrow their stock will be up 2x.
#AI #AIBubble #CoreWeave #CRWV #Meta #StockPrice
Alt...
Stock price chart of CoreWeave ($CRWV), the primary neocloud provider. It's dropped ~18% over the last two days.
the editor in question said this on Bluesky: "I’ve officially resigned as Associate Editor for Frontiers in Systems Neuroscience. It used to be a reputable journal, but became a case study in how forced automation destroys academic integrity. " here
Further down the thread: "I emailed and met with the editorial office to ask for the AI assistant to be turned off. I was told this is not possible. Instead, I was treated to some vague promises of potential future improvements and a dose of gaslighting."
Of course the AI assistant could be turned off (the editor knows this; I am not criticizing him, simply emphasizing a point). Whoever pulls the strings doesn't want it turned off, and so it will not be.
The future this is trending towards is AI-generated "articles" about "science" "conducted" by AI, "reviewed" by AI, "directed" by AI "editorial" systems. It seems the entire system would have no human beings in it. Why, exactly, would we want to build such a system? What would it be for?
#AI #science #SciencePublishing #AcademicPublishing #AcademicIntegrity
#SpudCell is an impressive technological achievement, but NOT a living cell:
https://biotic.org/research/spudcell
It does not have a proper metabolism.
Instead, it is "fed" liposomes whose lipids lead to membrane growth and "food" that includes pre-synthesized ribosomes & enzymes.
It suffers an error catastrophe after less than five "generations." So: not really "reproducing" (no sustained completion of the life cycle).
The authors know that. But "synthetic cell" sounds good for marketing, I guess.
@yoginho "nutrients including ribosomes, enzymes, and small molecules." is certainly an original definition of nutrients!
I'm immediately reminded of the dire wolf that was NOT a dire wolf.
Or Venter's "synthetic" cell, that was just an existing bacterium with an engineered genome.
All really impressive technological feats, don't get me wrong.
And in some ways probably useful to understand how life works, though it's not clear to me right now in what way.
But none of these achievements are anywhere near creating actual synthetic life.
@yoginho I do think that truly synthetic cells are a much more likely way in which artificial life will be made than some disembodied LLM 🙄
The Pygmalion bros are so tiresome with that bullshit.
@yoginho I do think that truly synthetic cells are a much more likely way in which artificial life will be made than some disembodied LLM 🙄
The Pygmalion bros are so tiresome with that bullshit.
I've never been impressed by demonstrations that amount to "we took apart living things, smooshed the parts together in a different way, and lo and behold, the result acts a wee bit like living things if you squint just right and forget what 'living' means". I don't have the background to surmise whether that's a fair characterization of what they've done, and without a clear methods section describing what they did I'm at a loss in that respect. But even "we synthesized parts from living things, smooshed them together, and lo and behold the result acts a wee bit like living things" is hardly any more impressive (except, as you say, as a demonstration of control).
It feels a bit like attaching some components to a battery and then claiming you've made a working electrical device without introducing any power because you've obscured the battery. Quanta Magazine loves to publish stuff like that.
@Stomata distros should provide more of the user login scripts compatible for each shell. Fish Shell is a much nicer default experience for new users.
New from 404 Media: we've got leaks from Amazon, GitHub, Adobe, Citi, and more. The big takeaway is that companies in all industries are suddenly throttling their employees' AI use because AI is actually too expensive. Cutting off access to some models entirely https://www.404media.co/companies-are-throttling-employees-ai-use-because-its-too-expensive/
@josephcox I wonder if this will trigger some organizations to explore using local models for things like AI assisted coding. Running an LLM locally on your own hardware would be cheaper, have fewer environmental impacts, and could be better for privacy. It could help to halt this insane centralized build out of data centers that only concentrate power in the hands of a few. Certainly not for everyone but that gaming PC could help you code during the day.
"Why should I waste an afternoon learning some tool or library just so I can write a one-off script for this project? It would be so much faster to just ask Claude to do it!"
I dunno, that sounds like my idea of a good time, actually! Why would I let an LLM do it? 😎
Seriously, though: I do worry about how much some folks have internalized hustle culture. Slacking off a bit with a useless programming side quest is a time honored tradition, and frankly good for avoiding burnout or getting yourself out of a rut. It's human.
Also, I hate to see folks complain about having to learn something when it's not strictly necessary. Learning is good for its own sake, and you never know what will be useful in the future! It's a muscle that gets stronger with practice, regardless of the topic.
The internalising hustle culture is very on point!
It’s very much part of what’s going on, and why some are revolted with it all, IMO.
Thing is, it’s kinda correct, no? The professional/money-making world works on hustle culture, and so most are likely correct in thinking they’ve gotta hustle everything.
Except now the value proposition of learning anything is questionable if not zero. There’s enough of a chance that the AI already knows and can do it better. Or will in the future.
And if that sounds off and alarming, which it should, it’s because this whole moment is likely very fucking serious. Not just progress but challenging the nature of humanity.
@maegul Well said, and I completely agree. I intend to stubbornly hold onto my values and humanity for as long as I can. I do worry about whether that will become harder and harder as folks around me and the tech world at large let this mindset take over. But, then again, resisting is the only way to prevent that...
@ngaylinn One of the good features of being a student again is having a (not entirely) random selection of seminars that I can join. Rarely do I finish one of those without an unexpected interesting thought. (Some of those would likely be monetisable also!)
@DamonHD Excellent! That's the spirit.
I've had the same experience, too. The random impulse class picks often turn out to be the very best!
RE: https://dmv.community/@jcrabapple/116846329747286727
What could possibly go wrong? 👀
Scientists built a cell from scratch for the first time. It feeds, grows, replicates its DNA, and divides. Called SpudCell.
36 genes, 90K base pairs, 150-200 molecules. Not alive (can't make ribosomes, can't evolve, needs constant feeding) but it's the first complete cell cycle from entirely non-living components.
@abucci The answer, from one person outside the US, but close to their border, is, isn't the answer obvious?
@davidgerard Our right to privacy shall be respected and the UK Government must take action to respect our citizens basic human rights.
@davidgerard i thoroughly hate the idea of creepy people walking around surreptitiously filming everyone they see. or the potential use of smart glasses as a kind of mobile face recognition and surveillance network.
however, the general idea of domething you wear like glasses that can process what the wearer is looking at is an amazing accessibility tool.
augmented reslity glasses are also used in some complex engineering settings such as aircraft maintenance.
regulation should be about the unconfined "mass-data" and online aspects of these.
@Bredroll @davidgerard Have you ever been harassed by somebody using these glasses? They are predominantly used by people as harassment aides and should be treated as such
@Shrigglepuss @davidgerard im sure it happens, but im also sure that no blind or sight impaired person will be doing that
@Shrigglepuss @davidgerard if someone is determined to be a creepy wierdo by filming people without thier knowledge and these glasses are banned, they will just go back to what they did before, little cameras in a hat, or bag strap.
much better would be to make thier use for unwanted stuff be obvious, like the fake camera shutter sound phones are required to have in many countries, perhaps a glowing red "record" light for these?
@Bredroll @Shrigglepuss see elsewhere in the thread about obvious actual use cases versus excuse pseudo-use cases
@davidgerard Frogface Farage would say "we're not part of the EU, why should we care about GDPR?"
Also, there's huge profit waving, cannot ignore that.
And imagine having 400.000 new cameras all around, a surveillance wet dream.
@davidgerard I can't sign this petition as there are a lot of blind people using these to help them with navigation and reading while out and about.
@andrew @davidgerard but the glasses are designed and sold more as a spying tool than a disability aid. Why not make the camera more obvious so people know they might be being filmed? It doesn't have to look bad, just obvious.
Besides, even if the user isn't spying on people with their creepy glasses, you can be pretty sure Meta is
@trantion @andrew yeah @fasterandworse has a whole thing about products that have benign use cases put forward as an excuse for the incredibly odious and obvious true use case
somewhere on https://www.youtube.com/@fasterandworse steve which one is it?
@davidgerard @trantion @fasterandworse Isn't there a camera active light on the glasses themselves when its taking pictures or filming?
@andrew @trantion @fasterandworse one that's trivial to cover, AIUI
the true purpose of the glasses is stalking. all else is excuses.if meta gave a hoot about the blind - which it absolutely doesn't - they wouldn't be the gadget they in fact are.
The glasses were designed for a different, and largely negative, purpose. Any side benefit they have for certain kinds of people does not change the negative purpose and the negative consequences that flow from it. We make due with whatever's available, but this does not mean that what's available is good or defensible.
I'll also throw in, as a low-vision person, that tokenizing disabled people to win internet points in an online dispute sucks and you shouldn't do that.
@davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
@abucci @davidgerard @trantion @fasterandworse I am a low vision user who is using these glasses on a regular basis while out and about to read bus stop signs, restaurant menus, and if I am lost then I use WhatsAp or Aira to get some assistance. I completely agree about the fact that the design purpose of these glasses wasn't to help blind people, but I can't support and outright ban because they are helping me in my daily living to be independant.
@andrew @abucci @davidgerard @trantion @fasterandworse I think this is a major point. And it's not just blind people. There are genuine good use cases. Sales should not be banned, but pervert uses should incur significant punishment, including deprivation of liberty. Ditto, regulation of the maker - if they're found to be using private data for their own purposes, there should be very significant fines.
@robparsons @andrew @abucci @davidgerard @fasterandworse I think it's better to target the manufacturer than the user where possible, simply because they're an easier target than chasing every user after they do something wrong.
It was Meta's choice to make the camera discrete. In theory they could have made a mistake in assuming people couldn't hide the notification light, but if that was the case they should have been required to fix the problem.
There are genuine good use cases. Sales should not be bannedFirstly, (b) does not follow from (a). There are genuine good use cases of nuclear-powered rockets. They've also been deemed so astronomically risky that they are a non-starter. The history of tech criticism is full of less-extreme examples.
Secondly, in this particular case the "good use cases" are side effects. Tech companies don't give a shit about low vision and blind people, let's be real. We are used as tokens in a cynical groping for feel-good reasons to accept total surveillance. It is 100% possible to design and create hardware and software that does benefit blind and low vision people, while also having few of the negatives of so-called "smart" glasses. Such technology will never be made if we let tech companies play this sociopathic game and settle for the crumbs they toss our way from time to time.
@andrew@hodgson.io @davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
@abucci @robparsons @davidgerard @trantion @fasterandworse The petition does specificly state though to ban "all smart glasses with video recording capability" which includes all the smart glasses specificly designed for blind people as well as mainstream offerings.
@andrew @abucci @robparsons @davidgerard @trantion @fasterandworse yeah the petition as framed is a non-starter in that respect. Quite aside from the fact that the govt takes as much notice of these petitions as I do of a passing pigeon, the usefulness of these as accessibility tools vastly outweighs their possible role in the pervs perving kit. The only person I know who owns a pair is blind and it has been absolutely transformative for him. I'm sure there's a better framing available here.
@silasmariner @abucci @robparsons @davidgerard @trantion @fasterandworse I'm not worried about the ptetition as such, but backlash from public if blind individuals who already may feel compromised are challenged for using the glasses in public.
the usefulness of these as accessibility tools vastly outweighs their possible role in the pervs perving kit.You're the third or fourth person to bring this fallacious argument to the thread, presumably without reading the conversation up to this point.
As I said earlier, nuclear-powered rockets are incredibly useful. They are also astronomically risky, and therefore we don't use them. Asbestos is an excellent fire retardant, yet we don't use it either. The history of technology is full of examples of technology that would be quite useful, but was deemed too risky, damaging, or otherwise too full of negative consequences to inflict on society. There is such a thing as a tradeoff.
Your "vastly" is weirdly motivated exaggeration. It is quite possible to put together a system with no almost no snooping that would be of great help to blind and low vision people. I know this because I am low vision and I have done it. The fact that we are even having this "snooping is a necessary evil" conversation is that there is effectively no regulation of the technology sector that would force companies to work out how to make products that genuinely help people without spying on everyone around them. They are free to develop horrendous surveillance devices and pretend they are doing so benevolently. Orwell must be rolling over in his grave.
Somehow this has engendered a vast volunteer army of unpaid public relations ambassadors such as yourself, which only makes the problem worse.
@andrew@hodgson.io @robparsons@mastodon.social @davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
@abucci @robparsons @davidgerard @trantion @fasterandworse I agree that we need to make these glasses safe for everyone to use, I as a blind person don't want to be targetted for choosing to use smart glasses in a particular situation.
Smart glasses allow the recording of others without their knowledge or consent and the uploading of this content to the internet.This is a serious issue for everyone that could have life-changing consequences for anyone.
You say:
I as a blind person don't want to be targetted for choosing to use smart glasses in a particular situation.This is an issue of convenience for you.
That's really where you stand on it?
Smart glasses could be constructed that temporarily record only what's necessary to perform a certain function, and treat those recordings as ephemeral (deleted after use, never uploaded anywhere). Computer vision tech could be used to ensure that only images that read as street signs and other such markers are recorded with any fidelity. And so on. Regulation is one mechanism by which we ensure companies are forced to work out what the real use cases are for real blind and low vision users (not the make believe ones used as justifications) and how to support those use cases without running afoul of community preferences as expressed in said regulations.
Is your stance really that for fear of losing a convenience you have, and for fear of being criticized for it, you'd deny everyone else the possibility of having just such a regulation that could benefit you too? Like I said before, this sounds like a selfish stance.
@robparsons@mastodon.social @davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
@abucci @robparsons @davidgerard @trantion @fasterandworse The statement "uploading of this content to the internet." is vague. All the systems I know of require uploading the content to the Internet, whether that is to allow a person on the other end to see it, or to allow an AI system to analyse the content. There is little to no processing power in the glasses themselves.
@andrew @abucci @robparsons @trantion @fasterandworse I would like to see your response to the subject of Anthony's responses.
@robparsons@mastodon.social @davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
Nevermind that they can take that carrot from you at any time, which means you are not independent. You may have gained a bit in the short term relative to what you had before, but in truth you are dependent on the good graces of companies that demonstrably have no grace whatsoever. The moment they feel they don't need to pretend to care about low-vision and blind people, they will stop caring about low-vision and blind people. There is no independence or freedom in setting oneself up to be used and discarded in this way, especially since there are alternatives without that risk.
For what it's worth, I use an old school digital camera with an LCD screen, or a smartphone camera, to help read signs. I have map apps with downloaded maps and metadata for navigating. It works well enough; glasses wouldn't increase convenience appreciably. I could come up with hardware and software that would be significantly better than this or so-called "smart" glasses while also having few of the negatives. I imagine you could too.
@davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
@abucci @andrew @davidgerard @trantion @fasterandworse I imagine requiring a prescription for these as a medical device would be a good thing but something tells me the moment the market is limited like that they'll have no interest in providing the product.
@reflex @abucci @andrew @davidgerard @fasterandworse Having these as mass market devices can be really useful: more innovation, lower prices. I heard about these benefits when smartphones became common. I think it would be better if the lens was surrounded by something to make it stand out (e.g. black lens on a white background), so you can always see if someone is wearing camera glasses. It would feel less clandestine, but I don't know how to deal with people putting stickers over them
@reflex @abucci @andrew @davidgerard @fasterandworse To be fair, secret filming has been around for decades. The question is what's changed now. Are there significantly more people doing it? Is that because glasses like this make it easier or would they do it anyway? Cultural change that makes people think they'll get away with it? Or has it actually not changed that much but there's a greater perception of it as a problem?
@trantion @abucci @andrew @davidgerard @fasterandworse Pointing at a bad thing that can happen now is not a defense in making that bad thing easy and convenient, it's an argument that it all needs to be regulated.
I'm in the EU, you cannot record without consent, on your own property you must post a notice and people must be able to contact you and request footage of them be deleted.
This should be the norm everywhere.
@reflex @trantion @abucci @davidgerard @fasterandworse The lack of regulation is concerning even from the point of the assistance apps, something which has been pointed out by the blindness community on several occasions. If people are asking the meta AI service to identify objects and documents, we don't know to which companies/countries that data is being sent.
@reflex @trantion @abucci @davidgerard @fasterandworse If we want live human assistance from someone other than a known ally, that can come from two places. A volunteer using something like Be My Eyes, where there has been evidence of the volunteers recording the interactions and posting some of them on Tiktok, or from a company based in the US called Aira.
@trantion @reflex@retrogaming.social @abucci @andrew @fasterandworse "to be fair" is up there with "to play devils advocate"
@trantion @reflex @abucci @davidgerard @fasterandworse I very much believe its a cultural thing. I mean we had general public wearing body cams before smart glasses took off, and we didn't get this sort of backlash from that.
@andrew @trantion @abucci @davidgerard @fasterandworse Were body cams uploading video to the cloud where it can be analyzed and used to train corporate AI and facial recognition and sold to third parties who use it to stalk everyone they can?
I must have missed that part of bodycams.
There's certainly all sorts of conversations to have about all of this, but it is absolute rubbish to suggest that large numbers of individuals ought to be permitted to wear surveillance devices in public because we have police wear body cams. That's nonsense.
@trantion@masto.ai @davidgerard@circumstances.run @fasterandworse@hci.social
@trantion @abucci @andrew @davidgerard @fasterandworse Regardless, the cost to society is very high and non-consensual. The defense being mounted is that it is a medical device for some. I'm in favor of that, regulate it as a medical device with appropriate for that purpose functionality.
If they are going to claim that as a defense they should stand behind that when the rest is regulated away.
@andrew@hodgson.io @davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
@abucci @andrew @davidgerard @trantion @fasterandworse A normal company being honest about it's intentions would embrace this, medical devices are incredibly profitable and are often paid for out of state health care.
The fact that these companies have no interest in such a lucrative, unending revenue stream tells us all we need to know about their real goals here, and it's not just making money.
@davidgerard Yes! A local auditor has gone hard on using these at protest, amongst also using them to record homeless people for social media grifts. They're scummy horrible things
The CRASH Clock paper is now peer-reviewed! So to celebrate, my co-authors and I wrote an explainer article (because how better to celebrate one article than by writing another... oh academia...)
It's now published in The Conversation Canada: https://theconversation.com/a-new-crash-clock-measures-the-chance-of-satellite-collisions-and-its-ticking-down-fast-283481
This paper took forever to publish. All the co-authors are astronomers, so we initially started with astronomy journals, and they didn't like it. Not astronomy-enough, I guess. So we tried a space journal instead. Here's the article from Acta Astronautica: https://www.sciencedirect.com/science/article/abs/pii/S0094576526004091
And here's the non-paywalled version on the arXiv: https://arxiv.org/pdf/2512.09643
You may remember I wrote a bunch about this months ago when we first put it on the arXiv. Now it's peer-reviewed, and officially DONE.
The idea for this paper came from me looking at an orbital density plot, and wanting a more intuitive way to understand these densities. What does 10^-8 satellites per cubic km even mean?!
Alt...
Two plots showing density in Low Earth Orbit versus altitude. One is coloured by type of object, one is coloured by different megaconstellations or debris fields. One feature that really stands out is how the highest density in orbit is from Starlink.
Everything in Low Earth Orbit is travelling at essentially the same speed. So you can take this density plot, and turn it in to a close encounters plot. The heart of the paper is doing that calculation in two different ways, one analytic, and one numerical.
In Starlink's orbit, the densest part of orbit, close encounters closer than 1 km happen every half hour or so. 1km sounds like a large distance, but remember the speeds are 7 km per SECOND. That is scary-close.
Alt...
Two plots showing the rate of close encounters closer than 1km versus altitude. Again, Starlink really stands out as having the highest close encounter rate, which is not surprising given the high densities.
The CRASH Clock (which is an acronym, I apologize deeply) is our metric for how dangerous orbit is. We download real orbits for everything (satellites, rocket bodies, tracked debris) and calculate how long it would be for a collision to occur if no orbital maneuvers happen.
This is a worst-case calculation. What if everyone suddenly lost control due to a solar flare, a bad software update, or someone hacking Starlink? How long do we have to regain control?
We originally wrote this paper nearly a year ago. As of last June, the CRASH Clock was at 5.5 days. It's now down to 2.5 days. It was 168 days in 2018, pre-Starlink.
It just keeps dropping as we launch more satellites into orbit.
This metric shows how completely dependent we are on continued perfect operations in orbit. 2/3 of all satellites today are Starlinks, and they performed 300,000 collision-avoidance maneuvers last year. They have done it perfectly so far. How long can that continue?
Alt...
A plot showing how the CRASH Clock timescale has dropped precipitously over the past 15 years.
The Conversation article above describes a collision scenario more in detail. It takes months to catalogue new debris from collisions, and in the the mean time, additional collisions can happen. That potential collisional runaway is the Kessler Syndrome, the worst-case scenario in orbit (explainer here: https://theconversation.com/too-many-satellites-earths-orbit-is-on-track-for-a-catastrophe-but-we-can-stop-it-275430)
The CRASH Clock is not a countdown to Kessler, but any major collision in orbit, especially in Starlink's super-dense orbit, will be an extremely bad day.
SpaceX is owned by the richest dude in the world, a demonstrably horrible person. And SpaceX effectively controls orbit right now.
There are also dozens of other companies planning to launch even more satellites into already-crowded orbit. (Another company I've never heard of just filed today for another 100,000 satellites https://spacenews.com/orbital-files-plans-for-100000-orbital-data-centers/. Fuck.)
Our future ability to make use of incredibly beneficial satellites is entirely in the hands of selfish, horrible billionaires.
@sundogplanets Can you do a simulation of how the ozone will be reduced due the refraction of space debris? How will a 2% (for example) reduction in UV will reduce the growth of oxygen producing ocean living micro organisms. Fuck Musk.
@sundogplanets The sun. The giver of life. Let’s draw a curtain over that and see what happens. Less life? What could possibly make you think that? How could the fraction of a percentage at source equate to an effect of hundreds of multiples on the ground? Well, mathematics, you git.
@Random_Seed Yeah, it's bad. There are atmospheric scientists working on this calculation, it is really far outside my area of expertise.
Just in case collisions in orbit aren't depressing enough, here's a super excited, cheery article with so many incredibly enthusiastic quotes from many astronomers about the start of full operations for the Vera Rubin Observatory! And I got to provide the doom-quote about satellites causing serious loss of data (I also provided lots of excited, enthusiastic quotes about science! But doom-quote won I guess...) https://skyandtelescope.org/astronomy-news/vera-c-rubin-observatory-begins-its-long-awaited-all-sky-survey/
@sundogplanets a guy who was working on the Rubin telescope once broke up with me in a cruddy way so I feel like if there's one upside to the satellite pollution thing it's that that one guy is going to experience professional frustrations
this isn't really much solace I know
@sundogplanets Wow! How have I not been following you before now? I thought I was.
I guess you get boosted into my timeline by other amazing folks so much that I didn't even register that I wasn't following. Ha!
A number of impressionable people I know seem to have spontaneously come to the conclusion that Kessler syndrome is a hoax, or a bunch of FUD, like climate change [sic].
Given the existence of (yet another) orchestrated messaging campaign in defiance of actual physics, I've come to the conclusion that oligarchs and their autocrat backers are trying to deliberately induce Kessler syndrome.
@sundogplanets
I never thought I'd need to ask this question before because I had always thought the probability was so low, but... If the Kessler Syndrome *were* triggered, do we know how long it would take to make L.E.O. possible again?
I honestly can't imagine just the sheer mess there would be, so comprehending any recovery is mind boggling.
@erraggy @sundogplanets The good thing about LEO is that atmospheric drag is high enough that most of the debris will suffer orbital decay relatively quickly. Square-cube law, smaller pieces have higher drag for their mass so their orbit decays faster.
@tknarr
@sundogplanets
The universe really has itself covered, doesn't it?
@erraggy @sundogplanets Orbital physics is a harsh mistress. Gonna be a real mess though when it shuts down pretty much all launch traffic for a couple years while the mess clears itself though. Better LEO than geosync.
@sundogplanets @erraggy I thought so. Most of the Starlink-type constellations seem to be below ISS altitude, so drag's a big factor. Especially for small objects, no?
Much above ISS altitude I figure anything there's going to stay there long enough to count as "permanently". Signal strength and delay seem to be issues at those altitudes though, I don't see Starlink using them nearly as much.
It won't be technology that ends their “perfect” record - it will be an MBA who decides it's more profitable to play the odds than to waste fuel adjusting orbits.
A curve that does a nosedive on a log-scale… *yikes*
In medias sci-comm I found log-scales to be "unintuitive" to laypeople. I think the message hits a lot harder if plotted on a lin-scale and also adding markers for when the megaconstellations got launched. A graph that just falls of a cliff the moment the first Starlink satellites went up should get the message across.
@sundogplanets Ok, you made me look: Collision Realization And Significant Harm (CRASH) Clock
I'm reminded of the scene in Hitchhiker's Guide to the Galaxy where the whale is falling towards the surface of a planet "And what's this thing coming toward me very fast? So big and flat and round, it needs a big wide sounding name like 'Ow', 'Ownge', 'Round', 'Ground'! That's it! Ground! Ha! I wonder if it'll be friends with me? Hello Ground!'"
@sundogplanets I seem to recall reading that the Starlink collision-avoidance procedure is to detect a potential near miss and accelerate a bit in a random direction. Am I remembering correctly?
@Redshift42 That sounds... terrifying. I hope that's not what they are doing, but I doubt they have publicly shared what they actually do.
@sundogplanets @Redshift42 Indeed. I know a bit about this from my own work experience with Earth observation satellites ((1990’s-2000’s).
There is a degree of collaboration in the gloabal management of space operations, but this is a good example where old methods might break. The Starlink orbits are predictable, and the others as well. That means maneuvers can be scheduled. If/When perturbations occur, the effects ripple through the existing schedule and adjustments are made as needed. I know of other constellations that had weekly deconfliction meetings. I don’t know if this is how SpaceX operates, but it is a classical, solved scheduling problem that occurs in many domains, not just space operations. The question is, given a perturbation, how much time do you have to detect a conflict, gather info, compute a good enough/reasonable/optimal solution and execute it? One second ought to be plenty of time, but the problem is it scales with the number of objects affected and how solutions are created, e.g., independently or centrally? Cooperatively? It can quickly become exponentially difficult to find an answer. So yeah, how do they do it? Whatever the answer, it cannot be guaranteed to work in every case (i.e., in a formal computational context). It is heuristic. That’s the risk.
I look forward to reading the article. 👍🏼
@sundogplanets I hope the people who use OUR orbits know what they are doing. If they fuck it up, I'm going to be seriously pissed.
To give to non-photographer an approximate idea of how long is a 1/7 of second interval (the time used by a satellite to travel 1 km), from which height must we drop a body under the average gravity acceleration?
@sundogplanets That is scary. Yes 1km is far too close.
It feels like two cars at a crossroads, both going across each other. Both doing 70. And missing each other by 1/10 or a second.
And that happening every half hour.
And we live right next to it.
@sundogplanets forgive a question, but if everything is moving at almost the same speed, then 10km spacing might not be that close? I guys what in asking is what are the speeds relative to each other like? I'm *guessing* your 7km/s is around the planet speed?
@sundogplanets Huge congratulations! The journal-hopping gauntlet is so exhausting, especially when you’re doing interdisciplinary work that doesn't fit into a neat, traditional box. Getting it past peer review and officially locked in must feel like a massive weight off your shoulders. Time to celebrate!
@SuiGeneris3722 Especially because the first author is a PhD student! Very glad to get this wrapped up for her sake.
@sundogplanets That’s amazing. There's truly no better feeling in grad school than getting that final approval. Glad it all came together!
@sundogplanets Wow, thanks for that article, it helped me a lot with understanding what the CRASH clock is about and why it's important!
Happy Asteroid Day!
On a related note, how serious is the threat of small asteroids on LEO satellites and the resultant space debris, as constellations grow larger?
How big does an asteroid have to be to cause serious damage and a debris field upon collision with a satellite?
Are there known instances of satellite collisions with space rocks?
https://www.universetoday.com/articles/asteroid-day-schweickart-prize-protect-space-infrastructure
@sundogplanets Pressing “boost” on this feels a bit wrong, like I’m launching another satellite! Extremely important work.
@sundogplanets oh it's been a long time since I've modeled that kind of linear-log scale
But I don't see an asymptote forming yet
😬
@autolycos I always feel a little bad when I make a log-linear plot, thanks to Prof. Julio Navarro, a brilliant cosmologist, who told me as a postdoc that "a log-linear plot is what you use when you want to lie to your MOTHER." (Imagine that in a Chilean accent for more fun.) I promise, I'm not lying to my mother! But it's a lot easier to see the trend this way...
@sundogplanets that is a fascinating take on that kind of plot
I prefer the epidemiology/microbiology take
@sundogplanets Avoid the rush. Eject space junk in LEO now.
https://www.leonarddavid.com/introducing-fragsat-theyve-got-balls/
@sundogplanets Great work - if a tad depressing!
Keep up the good work, and thanks for sharing with us!
Now, back to editing sat trails from my astroimages...
@sundogplanets please kessler faster. It will not affect GNSS constellations, anything that lurks in LEO can go to the trash.
Do you use Hungarian notation in your code?
| No, we have typed languages now: | 0 |
| Yes, because employer forces me to: | 0 |
| Yes, in my BCPL code, which has no data types: | 0 |
| Yes, I find it useful: | 0 |
@liw I used to, in BASIC and Pascal, but it's been a while 😉
@liw @dentangle why in Pascal if you don’t mind me asking? I’m assuming you mean systems Hungarian notation and Pascal is strictly typed…
m_ for member variables and s_ for static variables as I recall.The snooping is the purpose of glasses like Meta's, and people who try to argue that the intrusiveness is a necessary evil we must accept to "help" blind and low vision people are, pardon the language, full of shit. The snooping is not necessary, and could be almost entirely eliminated if tech companies were properly regulated and were genuinely interested in helping blind and low vision people. It's heartbreaking to see blind and low vision people volunteering to be the cover story for entities like Meta, but that's where we are unfortunately.
@andrew@hodgson.io @davidgerard@circumstances.run @trantion@masto.ai @fasterandworse@hci.social
I need more ways to politely say "this code is not being respectful to the reader, who has no idea what the fuck you're doing here"
my best leverage so far on this is appealing to "what if you're out as the only person who understands what the purpose of this is" and "what happens in 6 months when you've forgotten what you were doing when you wrote this"
but agh
source code is a medium for human beings to communicate with other human beings
the fact that it can be executed and has maybe-useful effects is kind of a secondary concern
History