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
An underground site is using facial recognition to unmask camgirls. Claims to have a database of more than 2 billion faces from 7 million people. Verified it works. Upload photo, shows other sites they stream on. Could be used with social media photos to find them https://www.404media.co/underground-facial-recognition-tool-unmasks-camgirls/
@josephcox Great reporting and I have mixed feelings on this.
This is hard stuff sometimes.
I don't find it complicated. Putting your face online does not give everyone on Earth consent to use that image for any purpose whatsoever. In the real world, there are legal limits to what you can do with a photograph you take of another person. It's recognized that such laws facilitate such freedoms as free association. Why does anyone think the calculation changes when we're talking about associating online?
I love this. Thank you for making it that simple.
The future lies in embracing statefulness and effectfulness in self-modifying code. Unlike the bureaucratic procedure of rule-based coding, this style of programming is more like surfing, or performance, or gardening. Your task as a programmer is to plant a seed of code that unfolds into something beautiful, possibly guiding it as it unfolds if you have the mastery. I'll leave as an exercise for the reader what the soil is in this metaphor.
(I'm only half joking!)
For the uninitiated, what are typically called side effects include:
In what sense, then, are they "side"? And what on earth is being "managed"? Isn't that what we mean by "programming"?
#ComputerScience #SoftwareDevelopment #tech #dev #politics #FunctionalProgramming #SideEffects #effects
@abucci thank you for articulating something I've long felt was utterly bizarre about the worldview of functional programming.
I'd like to hear more from you on the social environment from which functional programming emerged, and for what purpose it embeds Scott's "Seeing Like A State" mode of software design.
@abucci nah, I don't think so, but we do understand that one best. For all the fancy ideas we can have, if we don't know how they fit in we're just stumbling along. 😋
In that respect, I do agree with you that there is merit for it.
Everything I've read about OpenClaw suggests it's the NFT of AI. These folks need the fiction that AI is approaching "consciousness", or at least "agency", to continue.
#AI #GenAI #GenerativeAI #LLM #AgenticAI #VibeCoding #OpenAI #OpenClaw
Today in InfoSec Job Security News:
I was looking into an obvious ../.. vulnerability introduced into a major web framework today, and it was committed by username Claude on GitHub. Vibe coded, basically.
So I started looking through Claude commits on GitHub, there’s over 2m of them and it’s about 5% of all open source code this month.
As I looked through the code I saw the same class of vulns being introduced over, and over, again - several a minute.
@GossiTheDog @davidgerard I asked it to put an OIDC flow into a confidential app. It worked! I mean, it also sent all of the secrets and access keys via the client… but someone not paying attention would probably just take it.
We’re going to see the dumbest security issues of our lives in the next couple of years, aren’t we.
and you can use it as a geological epoch, pliocene, plistocene, holocene, failocene….
@abucci an interesting contrast since "AI" depends so much on demoing well for its existence.
Everything written by AI boosters tracks much more clearly if you simply replace "AI" with "cocaine".
I shall demonstrate!
(Not linking to OP, because it's trash.)
"Let’s pretend you’re the only person at your company using cocaine.
You decide you’re going to impress your employer, and work for 8 hours a day at 10x productivity. You knock it out of the park and make everyone else look terrible by comparison. [...]
In this scenario, you capture 100% of the value from your adopting cocaine."
If you're not doing cocaine, you're going to be left behind.
@troy_frizzell @jwz cocaine won't take your job, but someone using cocaine will take your job
checks out
@jwz@mastodon.social A selection of headlines from finance.yahoo.com today: - "Dark side of cocaine": Wall Street weighs stock sell-off over disruption fears
- Rampant cocaine demand for memory is fueling a chip crisis
- The Big Tech losers as cocaine fears wipe billions off valuations
@driusan You can even combine the two!
"Investors see hope in rich people's yacht money despite cocaine fears igniting a turbulent week for markets"
"Cocaine wasn't the biggest engine of rich people's yacht money growth in 2025, despite assumptions"
"The cost of cocaine slop could cause a rethink that reshapes rich people's yacht money"
"Cocaine is transforming rich people's yacht money -- not destroying it"
"Cocaine spending frenzy could reshape rich people's yacht money, CIOs say"
@jwz "Look, we're putting cocaine in Firefox, but you'll be able to opt out if you don't want it. Just don't inhale."
@theorangetheme It is completely opt-in. Your room comes equipped with a mirror, a spread-out line and a stainless straw, but if you call down to the concierge desk, someone will come up and flush that down the toilet for you. If that's really what you want. By the way here is a pamphlet about our ethical cocaine initiative.
@jwz cocaine first browser
@jbz "I denied a cocaine pull request and then a cocaine posted a hit piece about me. And then someone wrote an article about it but actually cocaine made up a bunch of quotes by me."
@abucci Maybe the spelling is actually non-determanistik and you're just the first person to notice? 🤔
We present the first representative international data on firm-level AI use. We survey almost 6000 CFOs, CEOs and executives from stratified firm samples across the US, UK, Germany and Australia. We find four key facts. First, around 70% of firms actively use AI, particularly younger, more productive firms. Second, while over two thirds of top executives regularly use AI, their average use is only 1.5 hours a week, with one quarter reporting no AI use. Third, firms report little impact of AI over the last 3 years, with over 80% of firms reporting no impact on either employment or productivity. Fourth, firms predict sizable impacts over the next 3 years, forecasting AI will boost productivity by 1.4%, increase output by 0.8% and cut employment by 0.7%. We also survey individual employees who predict a 0.5% increase in employment in the next 3 years as a result of AI. This contrast implies a sizable gap in expectations, with senior executives predicting reductions in employment from AI and employees predicting net job creation.From https://www.nber.org/papers/w34836
Western Digital has recently announced that it sold all of the year's hard-drive inventory to AI companies.
In other words, prepare for huge SSD price surges.
https://wccftech.com/western-digital-has-no-more-hdd-capacity-left-out/
Good AI research should tell us something about life, or it should help people. I hate seeing research about automating what people do. It's not a good goal for science or society! I was recently reminded of this by a paper applying LLMs to math.
This domain has many good questions: what do we mean when we say a person "solves math problems"? What are they actually doing? How is this like or not like what an LLM does? How might mathematicians benefit from this?
Instead, we get papers that pit an LLM against a human on a math problems dataset. This is great for claiming "AI has superhuman math abilities now!", but it's debatable whether good answers in a test-taking environment have anything to do with logic, reasoning, or creative problem solving. Instead of exploring to what extent LLMs are "really intelligent" vs. "stochastic parrots" (and perhaps the same question for humans), it reduces everything down to a number, one that hides the deeper problem and seems far more definitive than it is.
#ai #llm
@ngaylinn Another factor that is often overlooked in particular with mathematics LLM claims is that the purpose of mathematics is not to produce as many theorems as possible and be the best, but rather to understand the mysteries of logic and nature. Understand by humans and explain to humans. It is utterly worthless if I have a machine that tells me "the following technical statement is true but requires a 2000 pages argument and numerical computation of 1487628 boundary cases". I don't care if the statement is true, all I want to know is WHY it is true and how that relates to other statements. It is well possible that LLMs can help with the latter question, but the benchmark should not be "can it solve controlled practice problems faster than a human after it has read all mathematical texts ever written and burns the electricity of a whole city?"
@abucci I love this idea. I think you could do some very interesting proof works involving the limits of symbolic manipulation, and that this might shed light on both the limits of LLMs and some of the "other stuff" that math we care about does aside from that.
What makes math meaningful, rather than just a pile of symbolic statements? Very related to @paulbalduf 's comment.
Lately I've been re-reading some of Jan Brouwer's writing on intuitionism, and I feel reasonably confident in my belief that Brouwer would have wanted little to do with an automated mathematician. For him, mathematics happened in the heads of people; formalization and formalism are at best communicative aids. David Hilbert kicked him off the editorial board of a math journal over this stance, apparently, because as one of the most well-known proponents of the formalist school Hilbert was of a mind that mathematics was fundamentally a mechanical endeavor of constructing special strings of symbols. Some of the debates we've been having over generative AI were anticipated a century ago in these intuitionist vs. formalist debates. The bad news for the staunch LLMers is that the conflict was not settled in favor of the formalist school.
@ngaylinn see, the maths thing really trips me because what do you mean "ai" has "superhuman math abilities"?? COMPUTERS have had superhuman math abilities for over 100 years!! we made computers to do math. what's the novelty? they need to work harder than that ffs
@kageRapaz I think it's really tripping up the AI hype folks, too! Perhaps that's why people are trying so hard to get LLMs to do math, because it's vaguely absurd and embarrassing that they can't.
@ngaylinn Taking off from a tangent of the discussion, This reminds me of my beef with the "kaggle culture" that has taken over so much of CS research (and 99% of AI research). Everything is about beating some arbitrary test or another, no matter the hows or the whys.
Good for pumping out papers fast, horrible for actual research.
This:
In the year of the city 2274, the remnants of human civilization live in a sealed city beneath a cluster of geodesic domes, a utopia run by computer. The citizens live a hedonistic lifestyle, but when they turn 30 must enter the "Carrousel", a public ritual that destroys their bodies, under the pretense they would be "Renewed" or reborn.(Logans Run)
and this:
In the year of the city 2274, the colony of human beings on Mars live in a sealed city beneath a cluster of geodesic domes, a utopia run by generative AI. The citizens live a hedonistic lifestyle, but when they turn 30 must enter the "Cloud", a public ritual that destroys their bodies, under the pretense their consciousness would be uploaded to a computer and live forever.#AI #GenAI #GenerativeAI #LLM #Mars #eugenics #LogansRun #ScienceFiction #dystopia
In the novel, the age limit is 21. Which might kinda make a sort of sense for the uploading scenario, to get the minds while they are still flexible...
The two most important fundamentals of computer programming are, arguably, consistency and encapsulation.
Consistency lets you write something once and know it will work the same way indefinitely. This is a huge force multiplier and the main way computers have transformed society.
Encapsulation lets you decompose a complex system into predictable, independent pieces, allowing reuse and more complex programs.
Chat-prompt "programming" breaks both.
You can't get the same results twice except by accident. You can't be sure your prompts will do the same thing the second or third time.
You are forced to express things at all levels of abstraction, because the agentic AI is fractally stupid and trained on ten years of copypasta from Stack Overflow. (I am highly ranked on SO. It is intensely useful! But not where to 'learn' programming or interaction skills.)
And verbose. Just so hoddamn verbose.
It's just so tiring.
🤯🤯🤯 Full-on denial of reality.
State capture by the fossil fuel lobby.
The US descends into a banana republic. It would almost be funny if the consequences wouldn’t be so dire for billions of humans, if the courts don’t stop this folly soon.
@rahmstorf If you‘re interested in how they pulled this off, you can listen to this interview. Never trust conspiracy myths, but this shows every sign of a real conspiracy. https://www.nytimes.com/2026/02/12/podcasts/the-daily/trump-activists-climate-policy.html
@rahmstorf and German chancelor is testing the chances of copying Trump‘s deregulation without losing power. He is strengthening his subordination to the European extrem right to secure political back up.
The US is just a blueprint for European conservatives and right wing extremists. Merz is demanding deregulation, he just does not have the power to do so.
Wikipedia’s editors and readers have a dangerous generational gap. Can it become something that serves younger audiences, or will it remain stuck in the past? https://spectrum.ieee.org/wikipedia-at-25?utm_source=mastodon&utm_medium=social&utm_campaign=fedica-Mastodon-Daily-Pipeline
RE: https://mastodon.social/@ieeespectrum/116059551433682789
> The volunteer community that built this encyclopedia has lately rejected a key innovation designed to serve readers.
this specifically means: they rejected covering Wikipedia in AI slop. This means a generational gap apparently. ok mate
Wikipedia’s editors and readers have a dangerous generational gap. Can it become something that serves younger audiences, or will it remain stuck in the past? https://spectrum.ieee.org/wikipedia-at-25?utm_source=mastodon&utm_medium=social&utm_campaign=fedica-Mastodon-Daily-Pipeline
@davidgerard caring about facts is being stuck in the past, got it
It is so weird that Wikipedia is a rare bastion of truth and journalism now
@davidgerard @ieeespectrum
the fucking gall of evoking Elinor Ostrom's work (a lazy misreading of it, even) to suggest that Wikipedia should allow walled garden-erecting, anti-commons monopolists to poison Wikipedia with slop. absolutely, insultingly absurd. no notes.
@davidgerard there’s nothing the kids crave more than LinkedIn-style chatbot slop pushed into one of the last usable websites by a professional engineering organization from 1963 they can’t afford to join
@davidgerard
The author really needs to get in the fucking sea.
"But the volunteer base is aging. A 2010 study found the average Wikipedia contributor was in their mid-twenties; today, many of those same editors are now in their forties or fifties."
@geospaz @davidgerard people in their 20s age in 16 years?! How is that even possibly?!
Oh, and a 29y364d old on 1 Jan 2010 is…46 today, not in their 50s!
@davidgerard come on IEEE, leave Wikipedia alone: if you dont like the truth then you can just go and use an already existing 100% private enterprise AI slop-ipidia; Elon Musk's very own Grokipedia (no I won't link to it).
@davidgerard
Not a problem. Elon Musk has already cloned a slopipedia from it. Anyone actually wanting slop is free to use that instead, regardless of their age.
@davidgerard 'You know what will make wikipedia better? Content generated by a machine woth no concept of truth being important that has been fed upon copies of wikipedia that have been regurgitated by a different machine and passed through the 'Will this please a nazi billionaire?' filter.'
@davidgerard although I agree on keeping AI out, I do think the article is a bit more nuanced.
@jeroenbosman @davidgerard A long and complex discussion by passionate volunteers is boiled down to the single word "yuck". This article is marked safe from engaging with any sort of nuance.
@avery @jeroenbosman it's straight-up advocacy with a slight nuance paint job. that's more odious, not less
@davidgerard They know how unpopular AI is and buried what the "innovation" actually was a full eleven paragraphs in, and even then only mentioned it very briefly before rushing onwards with the presumption that the most trusted website in the world rejecting AI slop must somehow represent some sort of crisis.
@davidgerard I think it has to do more with this
https://mastodon.bsd.cafe/@ricardo/115906820176488716
@davidgerard "Dariusz Jemielniak is a professor at Kozminski University, a faculty associate at Berkman-Klein Center for Internet and Society at Harvard University, *and a board member at Campus AI* and at the European Institute for Innovation and Technology. He served for a decade on the Wikimedia Foundation Board. He recently published the books (with A. Przegalinska) *Strategizing AI in Business and Education* (Cambridge University Press, 2023) and Collaborative Society (MIT Press, 2020)."
@davidgerard “Why did 2025’s Wikipedia crisis result in immediate clampdown, whereas its internal crises from 2011–2014 found more community-based debates involving discussions and plebiscites?”
Because the AI summary “innovation” is antithetical to everything Wikipedia is supposed to stand for, maybe? And the origins of LLM models are built on theft, exploitation, and are seriously damaging to our information ecosystems at large?
Fucking fuck off with this “generational gap” bullshit.
@davidgerard keep on rejecting it guys. Too much slop as it is, let's try and have one or two places where you can be pretty sure it's a real human on the other end.
Alt...
The "but it might work for us" meme from Arrested Development, captioned "but Wikipedia cannot survive by refusing to change"
@davidgerard The Chinese Communist Party has granted British Citizens 30 days visa free entry into Mainland China because under the UK Labour governent the UK and China are now allies while the degenerate USA is an enemy of China and doesn't deserve visa free entry. I hope the Americans are jealous of British citizens for not being allowed visa free entry into Mainland China. The British can now see the truth about China and Xinjiand instead of being lied to by the corporate media, Fuerdai
Also, "generational gap"? Are you going to tell us next that millennials eat too much avocado toast?
@ieeespectrum ieee is what's washed up
@abucci @ieeespectrum i haven't forgiven them for their hit piece on louis pouzin and the OSI where the author cackled over IBM's ability to distort and break down the international collaboration. https://spectrum.ieee.org/osi-the-internet-that-wasnt they are a fundamentally immoral institution
This supposed "crisis" seems to be made up. For example, where the article says "Research has shown that many readers today greatly value quick overviews of any article," the link shows something completely different: an article titled "In the AI era, Wikipedia has never been more valuable", containing no such research.
From https://www.reuters.com/business/alphabet-sells-bonds-worth-20-billion-fund-ai-spending-2026-02-10/
Yeahhhhhh OK.
@abucci Motorola issued its 100-year bond the year before Nokia overtook it as the biggest seller of handsets. Three years later, Motorola was losing money on every handset it sold. Within ten years of the bond, the company split to avoid a selloff. Alphabet knows all of this because they bought Motorola Mobility two years later.
One thing it says to me is that Alphabet is not expecting the explosive profits from AI that the industry has been suggesting are forthcoming. Convertible bonds are what would allow investors to take advantage of AI upside. Century bonds are traditionally issued by slow-growing, stable businesses like Ford or Coke. There's no way for me to know for sure, but it feels like a desperate confession.
I hadn't remembered that Alphabet bought Motorola Mobility so perhaps they even had a ready-made playbook on hand. Interesting.
From https://www.cnbc.com/2026/02/06/google-microsoft-pay-creators-500000-and-more-to-promote-ai.html
Echoes of Matt Damon shilling crypto.
If the organic demand for AI were as high as we've been led to believe, what's with the big paychecks to shill it?
#US #economy #stocks #CasinoCapitalism #AIBubble #AI #GenAI #GenerativeAI #ChatGPT #GPT #Microsoft
Permalink: https://fred.stlouisfed.org/graph/?g=1RGjU
Alt...
Graph of data from the St. Louis Fed's FRED tool, plotting the "World Uncertainty Index for United States" series versus time. It explodes in recent years and is over 2.5 higher than it was at its last peak near the start of the COVID pandemic
#singularity #TheSingularity #RayKurzweil #AGI #MostPhotographedBarn
New on Techtonic: I speak w/Peter Dear about the history of science - then explain that *how we interpret AI* is in the balance right now.
Full episode:
https://techtonic.fm/episodes/2026-02-09-peter-dear-the-world-as-we-know-it-and-how-we-interpret-ai/
More Techtonic:
https://techtonic.fm
RFK Jr. openly announces his eugenicist views as well. I think it's important to remember that while the tech and AI are bad by themselves, they are serving an even more sinister, eugenic purpose in the hands of these folks.
A first essential condition on the cognitive is that cognitive states must involve intrinsic, non-derived content. Strings of symbols on the printed page mean what they do in virtue of conventional associations between them and words of language. Numerals of various sorts represent the numbers they do in virtue of social agreements and practices. The representational capacity of orthography is in this way derived from the representational capacities of cognitive agents. By contrast, the cognitive states in normal cognitive agents do not derive their meanings from conventions or social practices.Adams & Aizawa, The bounds of cognition
If you haven't been following this stuff for long, it might surprise you to know that "adding" 130,000 jobs in a month is a net loss. A jobs report like this would have been taken as a sign of a very poor economy even 15 years ago. The reason is that around 180,000 young people reach legal employment age each month in the US (Contra the admin's confabulations, this is how many US citizens reach employment age, and is not about immigration slowing). Adding 180,000 jobs would be treading water. Adding 130,000 is effectively 50,000 new people without jobs, a net loss by the spirit of this metric. There's also the reality that many of these estimates have been revised down by the BLS after more data came in.
Saying hiring is "picking up" and is "stronger than expected" is effectively saying "it could have been worse! 🤷 ". It's wild to see this positive spin on a very poor report.
It's extraordinarily fishy that the reported unemployment rate is not moving much in spite of these persistent and mounting job losses. I know why this is so please don't @ me a splanation. My point is that it, again, does not reflect the spirit of what this metric is meant to capture. It is wholly perverse now and should not be reported at all, let alone taken as a meaningful indicator.
The US economic numbers, when you push aside all the B.S., are recession-level. There's no "momentum" in the "labor market". There are not "green shoots" or "signs". The economic situation is bad.
"With that in mind, we need to consider the possibility that cp(1) may soon become – and indeed _may already be_ – sentient and self-aware."
https://exple.tive.org/blarg/2026/02/07/on-the-crank-spectrum/
"In October, the government classified the AMOC collapse as a national security risk. It amounts to a reckoning with national survival."
And a major #AMOC change would seriously impact the world at large, even the tropics - not just Iceland. 🌊
https://www.adn.com/nation-world/2026/02/10/how-a-warming-planet-could-turn-iceland-into-a-glacier/
Alt...
Marc Benioff asked international employees to stand up at a all-company meeting then made a "light hearted joke" about ICE agents coming after them.
Real class act, this guy
Alt...
popular tweet:
Being a billionaire must be insane... like being kicked in the head by a horse every day
TL:DR; The Salesforce CEO at a company's annual all hands gathering asked international employees to please rise in the audience, then made a "joke" that ICE agents were in the back of the room ready to nab them.
Guillotines for billionaires is the moderate choice at this point
@mathowie
There is someone on his staff -- probably multiple people -- who get paid to tell him not to say that
Had the guy I was speechwriting for ever slipped in something even *one tenth* as awful as that, I would have fired myself
And not just because it would have woken me up to who/what I was working for, but also because I would have considered myself a failure at my job
Just... holy shit
What a fucking monster
@kims @mathowie Benioff has been stepping on this particular rake repeatedly. I imagine his comms folks are apoplectic
https://www.businessinsider.com/marc-benioff-apologizes-for-national-guard-comments-2025-10
“can you recommend an alternative?” no. that’s why I’m so fucking angry. we fucked it. we’re no longer capable of making online chat protocols or web browsers or discussion forum software that people actually want to use. it’s all shit. every choice is either proprietary and exploitative or awful open source garbage sucking all the air out of all alternatives and forks, operated by fucking terrible people. we let this happen.
you want there to be an alternative so you can skip being angry too.
don’t come to this thread to argue. get angry and fucking do something about it. you feel trapped? good. welcome in. it fucking sucks in here.
@zzt
Do you know the best place to start making a difference?
I'm super out of my depth but from today it seems like XMPP might be a viable way forward in terms of large-scale community platforms like discord? Does it lack for decent clients or... are we really so far behind that we don't even have a base to build upon?
@davidgerard @R3yScale signal groups support 1000 participants, and video chats supposedly will do 75
an idea (no warranty on how good it is) is to use libsignal to implement a signal client with a concept of a signaling group chat, used to implement groups of groups and present them like how discord presents servers and channels. signal’s existing invite URL mechanism can be used both to invite users to the “server” (signaling group chat) and manage channel membership.
So this is fun: libsignal is AGPLv3. This means that you cannot use it in anything that you put in the Apple App Store. Signal requires a CLA that allows them to use a different license for their iOS app.
It also means that you cannot use it to integrate Signal with any proprietary services (which makes me sad because I'd love for banks to use Signal to reach people instead of SMS).
Protocols should have permissively licensed reference implementations.
@david_chisnall@infosec.exchange @zzt@mas.to @davidgerard@circumstances.run @R3yScale@techhub.social what terms for Apple's app store disallow AGPLv3 code? I thought it was not required because signal releases the code and apple isn't modifying or hosting their infrastructure at all
@froge @davidgerard @zzt @R3yScale
GPLv3 is incompatible with App Store distribution because the code signing means that the end user can't make changes. The anti-Tivoisation clauses in GPLv3 were explicitly designed to prevent this kind of distribution.
GPLv2 is also probably incompatible apparently, but I can't remember why (something about end users not being able to make changes or distribute or something). This is why VLC in iOS is limited: it doesn't include any of the GPL'd dependencies.
A lot of folks have learned about this and discovered that using AGPLv3 instead of a permissive license gives them a lot of control over their ecosystem. It's a great way of seeming open but preventing forks from being deployed in a load of contexts.
@david_chisnall @froge @zzt @R3yScale tivoising GPL2 is fine, that's why that provision was added to GPL3
honestly, the problem you raise is entirely on Apple
feel free to do a BSD re-engineer of libsignal, it's the strategy most use in this sort of situation
unless you don't actually care that much and just wanted to flame the GPL in a conversation that you introduced the GPL to
@davidgerard @froge @zzt @R3yScale
Doing a reimplementation of a crypto library is absolutely something I am not qualified to do. Doing a reimplementation of a crypto-protocol library and keeping it up to date with an evolving protocol is something that is almost impossible for a third party that isn't treating it as a full-time job.
EDIT: I brought the GPL into this because AGPLv3 is the reason that there are no third-party Signal clients for iOS (Molly is available on Android) and is the reason why the proposed approach in the post that I replied to cannot work. No one who is not Signal can ship an iOS app, and iOS is almost half of the market. The CMA may change this, we'll see.
But, at the moment, AGPLv3 is a tool that Signal uses to prevent third-party clients. AGPLv3 + a CLA means that the owner of the repository has a lot more rights than everyone else with respect to use of the code. That creates a power imbalance in the ecosystem.
@david_chisnall @froge @zzt @R3yScale right, so much more about GPL resentment (and not Apple resentment, for some reason) and not really about Signal at all except for daring to use a copyleft licence
@davidgerard @david_chisnall @froge @zzt @R3yScale
I didn't see resentment. Just clarity on what the GPL(s) are suited for, and why Signal would make the engineering choices they did on the licensing.
Those have both intended effects and (clearly deemed acceptable or at least tolerable) side effects.
The constraints on 3rd party interoperability and level of effort required are just something to deal with.
My takeaway: maybe skip full interop and use the protocol.
Without wishing to derail the discussion:
To a non-profit eking out an existence amongst hostile corporate entities, building an AGPLv3 + CLA moat around your product so you can stay in control is arguably fair. In the case of the Signal Foundation, this appears to have been the explicit goal.
I think Mr. Chisnall's argument against copyleft needs to be understood in the context of having libsignal be a commodity industry client baseline?
@ermo @david_chisnall @froge @zzt @R3yScale i'm understanding it in the context of david having started on anti-GPL arguments out of the blue in precisely this manner for a few years now
@davidgerard @ermo @froge @zzt @R3yScale
The context is that the GPL prevents precisely the thing that was proposed in the thread, but I am aware that whenever a discussion of the GPL comes up you jump to ad hominem as quickly as you can (in the last thread on the GPL, you immediately jumped to calling me a libertarian, as I recall).
I guess GPL advocacy is much easier if no one ever mentions any of the negative consequences.
My impression is that Mr. Chisnall comes at this from the utility argument re. being able to proliferate standards across the ecosystem, while arguing that the AGPLv3 + CLA de facto encourages silos, the freedom from which seems to be the most prevalent and accepted argument for the GPL?
IIUC, rather than "out of the blue", it might instead be an honest argument for why the GPL is not always the guarantor of the desired outcome of ubiquity?
@ermo @davidgerard @froge @zzt @R3yScale
If you insist on using a title, it’s Dr, but there’s a reason my username includes my first name.
I mostly come at this from a software freedom and end user empowerment perspective. I want to build systems where end users are free and able to modify their software to better serve their own purposes. The GPL was created with the same goal in kind, but it was created in a setting where users were MIT PhD students, but also in a world where UNIX was young. Programs were small, self-contained things that communicated via standard input and output streams, where GUIs were rare (and simple).
Enforcing the GPL costs a lot of money. I worked on a FSF project where we saw a VC-funded startup take our code and violate the license. The FSF declined to do anything because enforcement was too expensive. The startup was later sold for a lot of money. So the GPL meant the same as a permissive license for malicious actor, but came with more restrictions for those that didn’t. It amplified existing power balances. When I worked at Microsoft, I saw the same thing: a company that can hire a load of lawyers can do whatever it wants, individuals can’t afford to enforce the licenses and there are always loopholes that mean that powerful people can do what they want.
But let’s assume that it does work. If you want to create a world where free software is the norm, end users need to value those freedoms, and not just the zero cost. That means you need to build systems where modifying the software and distributing changes are easy, natural ways of using the software. But licenses like the GPL assume a strict separation between the program (created by programmers) and documents (created by users). If a document wants to extend a program, a GPL’d program means a GPL’d document. That means you can’t share it under NDA and so on. If you share it at all, you have a bunch of legal obligations to comply with.
But all of that is off topic for this specific thread, which is about a specific choice of a mobile-first app (Signal supports desktops but only as secondary devices, you must create and maintain your account on iOS and Android) to adopt a license that prevents anyone else from shipping an iOS app based on their code. That’s a choice that they made. Now, you can argue that this isn’t strictly their choice, Apple’s App Store requirements are the things that really prevent it, but Signal has also tried to discourage third-party apps in a bunch of other ways, so I can’t imagine that they’re particularly unhappy with that aspect of their license.
So now we have an example of the AGPL restricting users from doing things that they want to do, all in the name of freedom.
@david_chisnall @ermo @davidgerard @froge @zzt @R3yScale speak for yourself: i simply license my extensions as AGPL. and it's a bit rich to claim (1) that the FSF represents all possible enforcement claims when the SFC has taken the mandate of heaven for a while now (2) not to elaborate upon your background in license compliance and then to call it off topic. you spent two paragraphs on it, it's clearly not off topic to you
@david_chisnall @ermo @davidgerard @froge @zzt @R3yScale i was hired for contract work to implement signal for a european startup several years ago so the idea that AGPLing the code that implements things safely undermines "freedom" is a bit much. the entire point of (A)GPL is that a proprietary competitor can't reuse the exact work that makes the base library safe to use, then add proprietary extensions without making those available to the community. it directly serves the purpose of something like signal, which attempts to produce an open service not tied to corporate walled gardens like for example google or microsoft loves to create.
@zzt @davidgerard @R3yScale would be able to consult with anyone interested in doing this. libsignal is very flexible and does not encode too many specifics to the signal app
@hipsterelectron @davidgerard @R3yScale I will pledge as much of my time as I can (note: currently that’s not as much as I’d like, as I’m behind where I’d like to be on a web forum project that’s an improved lemmy-compatible service and under the gun at work) if anyone’s able to commit to starting a project looking to implement along these lines, and would love to sit in on a libsignal consult if it’s feasible to do one in a group setting
hey shitheads who came here to argue: fuck you
that’s it actually, just fuck you
“oh ho ho you posted this on mastodon which is the alternative to” get fucked. the open web is burning and you’re jacking off on the flames. the fuck is wrong with you.
I have just been informed that atlassian quietly* acquired jitsi several years ago
to the absolute little shits who came into this thread to do free advertising for the jira of video conferencing: ahahahaha
* for the tedious little shits: 8x8 (other enterprise assholes) bought it off them but atlassian still considers jitsi theirs in every way that counts: https://developer.atlassian.com/platform/open-source/
@zzt it’s also mondo impossible to deploy and administer yourself. Because the company that writes it only profits when you pay them to deploy it. Toddler blocks level of obvious motive.
And then people say “look there are free hosts!” Yeah and what has happened to free hosts of literally every previous “free alternative” web service?? Shut down/paywalled as soon as it’s popular enough that the bills to run it (or prospective profit) are more than a rounding error of nerdery. This isn’t a fucking sustainable ecosystem and believing it is is fucking yourself over.
@s0 fucking exactly! nothing kills kneejerk advocacy like running this absolute shit in production.
@zzt i do shit about it. i don't want to be trapped in a room with a fuckwit foaming at mouth about solvable problems instead of going and solving them. fuck off
@zzt I said similar to friends earlier today.
"I'm just pissed. There's no real alternative to this. XMPP is a fine protocol layered in utterly shit clients, bound to OMEMO so it makes everything ten times harder than it has to be. Instant messengers are fine but not meant for large scale group chats. Matrix is a *fucking joke* that failed at E2EE, leaks metadata like a sieve, and capitulates to fascists (including ICE).
It would take just one project to fix this. One where people stopped being dicks and just doubled down on usability and ease of access. But that costs money and is *boring*. UX design isn't "sexy" nowadays unless it's RustNodeElectron ver. BleedingEdge.0.1(a) and so nobody wants to take on the work when they could just go make a new shitty, shiny thing instead."
That was about chat clients specifically but it's a cancer all the way down.
"Why write stable software when Docker exists and you can just trash it and start over with a handful of commands?"
"Well, there's gotta be a reason they're all doing it so I have to as well."
"Why should we work on polishing existing software when I have The Next Big Idea and can work on that instead?"
We won't ever be able to reclaim anything if everyone wants to abandon the "boring" parts in favor of the next shiny thing, leaving half finished projects to rot eternally in /home/fuckit.
@kitkat @Arthfach it failed completely at e2ee twice: https://furry.engineer/@soatok/116055556402436098 and previously https://soatok.blog/2024/08/14/security-issues-in-matrixs-olm-library/. mistaking fandom for security is a dangerous fucking game.
@zzt Are we not in the alternative right now?
It can be built but can we steer lemmings to use it?
Alt...
A Gif of the Lemmings computer game showing lemmings jumping into lava with a sign in the backround pointing to the lava with the workd 'Paradise'.
@joegaffey @zzt if your answer starts "well i'm a snide hipster, but" then you have singularly failed to understand the question
@joegaffey @zzt @davidgerard Yeah, to me that attitude is on a level with folks who think the geeks, nerds and Mops essay is great.
I love working with and in FOSS but geeze folks can we not just assume that kind of attitude about folks who use discord.
It was useful to them, there was an ecosystem for folks to tap into. Like moderation tools that evolved.
@joegaffey @zzt maybe it starts by not comparing your users (the 95%+ of humanity that doesn't have an inexplicable boner for arcane computer nonsense) to lemmings
@anhedonie @joegaffey @zzt if they can't telnet 6667 can they even really be allowed to use the computer
the key problem is that nobody wants to run a box
they want to use a service
that ends directly at discord
mastodon has the advantage that it actually successfully decentralised, and 75% of users can and do tell eugen to go whistle
it has the disadvantage that you can see the wires and plumbing and worst, the guys loudly telling you that the leaks and sparks and flames and outages are features akhully
@davidgerard @anhedonie @joegaffey @zzt and worse than just seeing the plumbing, there is no real model to buy a box and then call a plumber when you need the sewage flushed
You either give a company the right to all your personal info so they can shove ads in your eyes or you learn enough about admin to be able to run it with the help of a hosting service you pay a sub to. There's no Middle model that mostly just works with occasional maintenance like plumbing
@zzt Old stuff still works though.
@t_var_s @tux0r @zzt fsvo. distributed irc was shit in extremely tired ways, somehow combining the worst of decentralised with the worst of feudalised. it only "works" on a single well-run server, and we're back to the original problem.
i met my wife on irc, i have a fondness for it. it still doesn't answer the question.
@davidgerard @t_var_s @tux0r @zzt I have no fond memories of IRC usability wise. The UX was already bad in the 2000s, and from the little contact I've had with it since, there has been no improvement.
@davidgerard @t_var_s @tux0r @zzt
<DrCuriosity> Yes.
<DrCuriosity> There are some things I don't miss.
* DrCuriosity has quit (*.net *.split)
@DrCuriosity @t_var_s @tux0r @zzt
IRC's fragile spanning tree structure is the root problem but i don't know how we'd do cross-tree links without all sorts of weirdness
there's probably a solution involving terrible hand rolled cryptography that also doesn't solve the problem
(as with so many such problems)
@DrCuriosity @t_var_s @tux0r @zzt discord basically worked in ways irc didn't. if you don't understand that, you don't understand the question.
@davidgerard @DrCuriosity @t_var_s @zzt Webshit isn't a "way to work" though.
@davidgerard by hand rolled do you mean reinventing the cryptographic primitives, or do you mean not using tls? or maybe youre okay with tls but dont want to use https or wss in particular?
@sylvie i mean some general architecture astronautics involving ill applied cryptography, i absolutely have no plans to go into depth with it
That's a big part of the problem, seems to me. Mom signs up for IRC, someone says something like that, Mom leaves IRC, calls the police, and tells all her friends about what happened. Nobody in her social circle will ever consider signing up for IRC after.
This kind of tech, whether IRC or insert your favorite here, can't spread if it can't guarantee the boughie white person experience. Entire shopping malls go bankrupt after an assault happens in front of them.
@abucci @t_var_s @zzt German Usenet has a Law that states that “the longer a discussion goes on, the more likely is it that any possible behaviour will occur”. I think that fits. :-)
Unmoderated content is the core of the free internet though. I’m not sure either whether trying to regulate that by “heavy filtering” will make it go away.
Everyone should be allowed (and probably even encouraged) to make a fool of themselves.
Unmoderated content is the core of the free internet though. I’m not sure either whether trying to regulate that by “heavy filtering” will make it go away.I don't agree with this at all, I'm afraid to say. My belief is that this is a fantasy held by people who've never seen firsthand how the internet community sausage is made. The core of the "free" internet is moderated content: that's the only way you can be free to speak with other people freely. The challenge is figuring out how to do this while maintaining other values.
In my experience of the internet dating back to the 1990s (former USENET denizen here), unmoderated content leads to the blowing-apart of communities. USENET is largely unusable now because of unmoderated porn, spam, scams, hate, etc. The only spaces that ever persisted for any length of time were the heavily moderated ones, or the ones too small and invisible to matter to anyone outside the active user base.
I've done this work myself and I've seen firsthand how it works. I spent some time moderating a forum where some of us would proactively monitor another forum because users of that forum liked to rile up a gang of people to come over to our forum and disrupt it as much as possible. We'd pre-emptively permaban any of them who arrived before they even had a chance to post. If we hadn't done that, our forum would have fallen apart. Regularly being attacked by groups of 10+ malicious people posting malicious content would have derailed all conversations. Many people would have fled, and they'd have been right to. The typical user never saw any of this, and could use the forum freely, because we were actively and heavily moderating it.
What you're calling "heavy filtering" is exactly why a group of people can have a lovely chat about their favorite movie without it being interrupted by trauma-inducing videos or images, or gangs of people posting offensive material. Online community cannot possibly exist and cohere if, unpredictably and at any moment, some randos could pop in and traumatize half the people. This happened constantly on unmoderated USENET boards, and happens constantly today. I run a single-user instance and have instance blocked over 100 domains now and every once in awhile someone still manages to get through and porn-bomb me or whatever. SSDD
Anyway, that's my experience of it, even recently here on the fediverse.
RE: https://mas.to/@zzt/116045651572809856
As angry as OP is, I don't think they're nearly as angry as they could be. Allow me to stoke the fire a little bit.
I have used damn near every single mainstream communications medium on the Internet since about the early 2000's. You know how Discord managed to establish dominance?
By being the *least fucking awful*.
By *giving a shit about the user experience*.
By not having their contributors/community shit all over things users like and refuse to implement because they personally don't want color, bold, italic, or underlines, in _their_ text, and adding options to turn those off just for them would just be _too hard_. So as a result, users still get those things by abusing Unicode characters (esp. for math), which produces a usability fucking nightmare for screen readers, which prompts other assholes to say "well the screen readers need to adapt and start trying to read text like that as normal text" when the entire point of having special characters is they have dedicated semantic meaning.
By having the first voice chat I've experienced reliably working basically every single time I click it. For me, Skype was never that good. The closest/best other thing for voice chat was fucking IParty.
By doing the logical thing and putting all the various big groups you might want to communicate in into one window, and by holding onto transcripts and making them searchable in-app so I don't have to (1) worry about whether my backups are okay and (2) `grep` through a bunch of big-ass files for the one thing I need to remember.
I've seen criticisms about something being an "everything app" or "not everything should be in one place" but what those criticisms UNIVERSALLY fail to address is that this one place beats every. single. other. *specific.* place. on their *home. fucking. turf.* In practically every way that matters.
And that's how we fucking got here.
AodeRelay boosted“can you recommend an alternative?” no. that’s why I’m so fucking angry. we fucked it. we’re no longer capable of making online chat protocols or web browsers or discussion forum software that people actually want to use. it’s all shit. every choice is either proprietary and exploitative or awful open source garbage sucking all the air out of all alternatives and forks, operated by fucking terrible people. we let this happen.
you want there to be an alternative so you can skip being angry too.
a few people have come here to advertise for their own projects. I don’t usually reply either because I’m already keeping tabs on their work or I feel nothing for it and can’t muster a reply.
some jackass came in here to advertise for a grift that’s very coy about how it requires you to spend ethereum to make a full-fledged account. the docs try to hide this and also say there’s non-cryptocurrency alternatives and that isn’t true.
don’t fucking try to grift people in this thread.
anyway the thread got its first “if you don’t like an open source project you’re not allowed to call it garbage instead of contributing” asshole I hadn’t already blocked yet and their codeberg is almost empty and that’s fucking hilarious
@lizzy@social.vlhl.dev I carry a phone just to take photos, as I don't have a sim card there's no internet when I'm outside, it's just a camera.
Cell phones, even with just SMS, are a surveillance nightmare. People who care about that sort of thing are right to be cautious, in my opinion.
Like I said, people who are cautious about surveillance are right to worry about this sort of thing. These dangers are quite real.
So, THAT's what ICE is doing ?
They sure are seriously lacking in reasoning, so it indeed makes sense for them to try anything to improve on that aspect...
@abucci Abduction is cool! It's like a detective's job: we've got some rules, we've got some consequences, but we don't actually know the facts that might have led to those consequences...
I read this and the first thing that came to mind was Kool Keith/Dr. Octagon saying "f@?k it he's dead / oh sh!t there's a horse in the hospital!" on General Hospital.
I am losing it at how many of my peers have forgotten what software engineering is. It is not typing in lines of code.
@abucci @bert_hubert lovely point. And I’m not saying you are wrong. Only few of the proponents of the things mentioned did not pivot the last two conference seasons.
@abucci @bert_hubert the people that researched and authored said blogs were very serious. Sadly many of those paying said authors to do the work that prompted the articles weren’t equally serious.
Getting the whole argument just right is fiddly, but the basic idea is this. You feed some kind of theory into the AM/AS, which is a black box. It churns on this and spits out a result, which is added to the theory (I'm neglecting the case that the result is inconsistent with the theory). It can now churn on theory + result 1. For any given and potentially very large N, after doing this long enough, it's churning on theory + result 1 + result 2 + ... + result N. Whatever it spits out will be dependent in particular on results 1 - N. When N is large enough, unless you know these results you will not be able to understand what it outputs because the output will almost surely depend critically on one or more of results 1 - N. In other words, the output will look like noise to you. If the AM/AS is appreciably faster at producing results than people are at understanding them, there will be an N beyond which no one can understand the output up to that point. It'll become indistinguishable (unable to be distinguished) from random noise.
If you're into software development, this would be analogous to a software system that generates syntactically-correct code and then adds that code as a new call in a growing software library. If you were to run this long enough, virtually all the programs it generated that were short enough for human beings to have any hope of reading and understanding would consist almost entirely of library calls to code generated by the system. You'd have no idea what any of this code did unless you studied the library calls, which you wouldn't be able to do beyond a certain scale. If the system were expanding the library faster than you could read and understand it, there'd be no hope at all.
I'll leave it as an exercise to the reader whether this is a desirable thing to do and whether it's happened yet. I would offer, though, a question to ponder: what reason is there to believe that a random number generator hooked up to an inscrutable interpreter produces human flourishing, for any given meaning of "human flourishing" you care to use?
#tech #dev #mathematics #AutomatedMathematician #AutomatedScientist #AI #GenAI #GenerativeAI #ThoughtExperiment
@abucci Feels adjacent to Kolmogorov complexity. The simplest system that can contain all of math is all of math.
A Mozilla director of engineering:
https://mastodon.social/@stevemarshall/115740621035086166
> Are we building AI features? Yes, because lots of users are asking for them.
WHICH USERS STEVE
SHOW US THESE USERS
NOBODY ELSE AT MOZILLA SEEMS TO BE ABLE TO PRODUCE THEM
AND WE'VE BEEN ASKING A LOT
@davidgerard it's possible that users are asking for features! unfortunately in 2026 if you attempt to deliver a feature by any method other than chatbot, you are immediately transfigured into a donkey.
@davidgerard When "lots of users" equates to a few real users and then the CEO who represents a number of users much greater than the real users.
@davidgerard To be honest, if they say they receive that request from a lot of users I have no reason (nor do I care) to question it.
What I wonder is if they are also measuring how many they lose. I am biased but in my circle, I have seen literally everyone (I was the one last standing) switching to any other browser.
Don't know, there is a good bunch of directors over there, I'm trusting they are smarter than I am.
@nilace I think there is reason to question it, though, because every actual survey or request for comments they do is relentlessly negative
it's philosophically possible the lurkers support them in email, but it's negligible as a claim
@davidgerard I agree, I'm just making room for the possibility that they are researching in areas I have no idea about. Or countries that are not present in the same environments I am.
To be fair, it hasn't been without pain and it's been a process, but I don't care anymore.
I've seen people asking even for what I think is a fair middle ground in line with what used to be their... thing (opt-in instead of opt-out) with no answer.
Several weeks ago I just said goodbye.
I can't seem to find these "mysterious" users either.
The only ones who seem really enthusiastic about AI is the fossil fuel industry & their tech kleptocrats. The money laundering is endless.
https://www.nytimes.com/2025/10/27/technology/saudi-arabia-ai-exporter.html
https://www.theguardian.com/environment/2026/jan/29/gas-power-ai-climate
https://www.desmog.com/2025/12/11/the-koch-network-is-pushing-trump-to-accelerate-ai-documents-show/
@davidgerard Most users are asking for AI features! I mean, they haven't been able to find all the individual opt out buttons we hid deep in the technical settings, that means they consent right?
@davidgerard
Sounds more like that is from the director of marketing then from the director of engineering.
@davidgerard He just mixed up the word "user" with "customer". Just like their CEO before. As I tried to reflect on that fact earlier, here: https://mastodon.social/@chainq/115745585459795231
Just one note on the Mozilla CEO AI debacle: People are outraged. "How can Mozilla be so out of touch with their users?" - I keep hearing.
Remember: Yes, we are their users, but not their customer. The customer is who pays, and that is Google. The rest of us are a relatively insignificant part of the product, because Google will probably keep funding Mozilla anyway. It's their cheapest option to avoid some Chrome monopoly claims.
It's just sad. Sincerely, a Firefox user since Phoenix 0.1.
Or, support a better plugin infrastructure that allows 3rd parties to add stuff.
If some 3rd party wants to enshittify with AI, let them.
Keep it out of the base software.
@davidgerard It's that obnoxious swerve into formal logic where "all users hate 'AI'" is false, because "some users love 'AI'" is true, where "some" can be just one.
@foolishowl @davidgerard and it can also just be the ceo. dogfooding!!
@hipsterelectron @foolishowl eating our own dogshit
@davidgerard @hipsterelectron @foolishowl "Not only is eating dogshit nutritious, it is also invigorating!" - Firefox
@davidgerard
Apparently, they also know a whole lot of people who have a different definition of "opt-out" than the rest of us, so they're just keeping the AI features opt-in.
I don't know who these people are or why the Mozilla team's leadership knows so many of them that it warps their understanding of a basic concept, but I do know that I don't want to be alone in a room with them.
@davidgerard there's this discipline called product management and when some users ask for something, you just do it.
Never validate any feature request against considerations like cost-benefit analysis, interest from other users, whether it would make the product worse for more people than it makes better, any of the irrelevant stuff. Doesn't matter whether it relates to product goals at all either.
If it's a really expensive thing to implement and competitors are years ahead: even better.
@davidgerard Mozilla has conducted at least one user survey about AI features. The company should release its data to support its claim.
When they say "users" they mean the ones who use your data (remember when they started harvesting your browser activity?) for reasons... You know, like the NSA, DHS, IDF, MI6, etc.
@davidgerard if you look at all their social accounts, even the Mozilla AI one, and all their posts are very low engagement and when there is engagement it's mostly comments like "stop putting AI in stuff!"
@davidgerard Are we supposed to not ask why an 'opt-in' feature requires a killswitch?
@fuzzyfuzzyfungus these are the kind of guys who *literally* pretend not to understand what opting in is
@davidgerard I'm also one of those users who appreciate the AI features in Firefox, like many others who raised their voice here. However, I also support Mozilla's move to let users disable AI features. That being said, there is certainly no need to attack people and get personal here. This isn't X.
@davidgerard Ooft, lads, that "we don't believe in opt-in because we don't know what it means" thing is a death-knell. Proper jumping in with the techbro PUA creep crowd there.
I'd say that's a pretty solid litmus test for when an OSS project has passed the point of no-return under its current governance. The project may technically remain "open-source", but the point of it being open-source in the first place has been entirely defeated.
Don't need to bother inspecting the source for malware when they're proudly blogposting their way through admitting that they've given the entire company over to the purveyors of malware-in-a-guy-fawkes-mask.
When the entire notion of consent gets reframed so much that they've torn the picture, done a serial-killer magazine montage with the text from it and, oooh, whadya know, it comes out saying "i tHe UnDeRsInEd Do HeArBy gIvE cOnCeNt FoR [eVrYtHiNg]", the source licensing becomes irrelevant. It's radioactive either way. It's just a matter of time before you start losing fingers and toes to it.
WHICH USERS STEVE
SHOW US THESE USERS
Elon Musk Jeff Bezos the winging twit from microsoft who wants us all to love ai,
Thiel Peter Pedo nutter.
and other VIP`class users.
Nearly forgot Steve whatshisname from mozilla management.
We ignored the other 98.99% who keep telling us they do not want Artificial Idiots.
Because they think there are too many real ones around already.
one could make the argument that the 1:2:7 ratio exists. One¹ very vocal user asking for AI, 2 not very vocal users asking for AI, and 7 users wanting AI and not saying anything. Fair to follow what they are saying. It's like when we see a bug, we know it's affecting more users, than just the users who speak up
However, we have hundreds of very vocal people saying they don't want AI, if double that aren't being very vocal and are voting on their polls that they don't want AI. Just imagine how big the silent majority are who don't want AI
The ratio of people not wanting AI is bigger than the people who want AI, especially when they count the silent majority in each group
One wonders why Mozilla is ignoring the actual numbers and plowing on…
¹ Number are illustrative and don't necessarily represent an accurate amount
@davidgerard (steamed hams voice) user polls! at this time of year! in this part of the country! localized entirely to our company intranet!
@davidgerard for those morron and their AI cult 2 people asking is a lot of user.
Ans seeing their market share ... it's half of their user
@davidgerard@circumstances.run what users want ai in firefox? oh, you wouldn't know them, they go to another school
Slopzilla claims they are working for everybody but they keep announcing new AI slop features, but when it comes to major, exciting new Firefox features that aren't AI, they are totally silent.
@davidgerard He obviously didn't see the DuckDuckGo survey which showed 90% of their users dont't want AI. It's being imposed on us rather than being based up consumer market research.
reading vibecoders talk about how great vibecoding is for engineering real things is like reading bitcoiners talk about how they think money works
i am more, not less, convinced ai and crypto are closely analogous
@davidgerard they belong to the super-class of "grift, but disguised as a complicated economic thing that involves computers"
@floatybirb @davidgerard I'd say it's beyond mere "grift".
It's an active attempt to force basically everyone in developed world to pay them a subscription. This was a thing with crypto (hence people like Peter Thiel funding Ethereum, the Everything blockchain) and that's a thing now with AI (hence the push of "AI as a personal assistant").
@floatybirb @davidgerard @art_codesmith part of what I despise about llm-driven coding is precisely this, they're pushing to turn writing code into a subscription-based service - whether intentionally or not. What worries me is the developers in my own circles who I've seen and heard making statements like "I just couldn't handle the stress of the demands of my day job without these AIs" or "I wouldn't be able to keep up the output required if I went back to writing code by hand".
@davidgerard A few years back, the talk among the crypto folks was “How do I pivot my crypto farm to AI?” It is no accident the playbook seems so incredibly familiar.
@davidgerard I'd add that even before ai and crypto we had entrepreneurship, or even physically impossible indiegogo devices. There is an ongoing process of young men who believe that you can manifest revolutionary technology by merely stating what this supposedly future technology is going to achieve, while subcontracting the actual production to investment in whatever providers are nowadays attractive with the forces of market. They merely change the snake oil but the overall pitch remains.
@davidgerard @elrohir This has always been my vibe whenever Elon announces anything — it's the same as when a child "invents" something by putting some stickers on a box and writing "TIMƎ MACHINE" on it. It's adorable, but it's not a business plan unless you know how to build it
I am now unreasonably pleased that every device I bought on Kickstarter or indiegogo did eventually get made and sold (although often with some compromises for the most extreme features and some took a long time to actually be built)
@gbargoud @davidgerard Fortunately kickstarter always kept a rule requiring the existence of a reasonably working prototype for hardware campaigns, which Indiegogo did not.
That is why some of the worst offenders on Kickstarter were merely financially stupid like the juicero. Indiegogo, on the other hand, gave us bangers like
the "scuba mask that extracts more oxygen from water than its surface can physically touch" https://www.deeperblue.com/triton-underwater-breathing-gill-scam/
or the "laser shaving razor" https://bmmagazine.co.uk/tech/non-existent-laser-razor-hits-indiegogo-after-kickstarter-boot/
@davidgerard The last few years really made it painfully obvious how many utterly unqualified opportunists have found their way into computer science.
https://www.agbi.com/finance/2026/01/saudi-launches-blockchain-tokenisation-centre-of-excellence/
Cryptocurrency isn't currency. It's tokens gained by "mining".
Mining what? Supposedly mathematical operations but at a huge energy cost.
Cryptocurrency rewards wasteful energy use.
Like AI, it's pointless churn to burn fossil fuels & evade taxation.
https://www.fintechweekly.com/magazine/articles/saudi-arabia-flexible-salary-early-wage-access-fintech
Counterfeit leprechaun gold & useless company scrip.
https://en.wikipedia.org/wiki/Environmental_impact_of_bitcoin
@davidgerard
It's basically its own literary form at this point. You have the overly-italicized personal grindbro introduction about how the founder realized they had been doing it all wrong. the listing of the stack where you say which dozen frameworks and SaaS subscriptions power your paginated CRUD list site and HTML form. The recounting of the SLOC. The disregarding of the legal liability. Etc.
@jonny @davidgerard Those of us who were building “paginated CRUD list sites with an HTML form” 25+ years ago with Perl, PHP, or (yikes) ASP (just plain fscking ASP, no “.NET” underneath)* are looking at this and wondering why a vanilla CRUD site is such an accomplishment, at least for people who also have 25+ years’ development experience under their belt.
* Or in my case, all three, at various times. The pre-.NET ASP wasn’t my choice, though.
@dpnash
@davidgerard
"I realized what we needed was a the last list of pizza recipes, the one that would get it all right.
Flash forward. Claude is cooking. Next.js. Supabase for the toppings tables and Postgres for the doughs (for performance). Lovable is locked in. Bolt is battened down. A little airtable for robustness. Firebase auth plugged into stripe for micro-ingredient transactions. Dual deployment to vercel and netlify with a cloudflare worker for load balancing. A custom WordPress admin page hooked up with a whaletail synchole to prod. I select the final designs from figma and we're done.
40m tokens, 3 days, 456k lines of code, and it's done. pizz.ai."
“Pizz.ai now being live, it was time to cook up some skills for these recipes. I have my OpenClaw marinating on the Mac mini, so it was time to use it to add some spice. Less than 10 million tokens later, I can turn on the pizza oven and set the temperature from the toilet. By talking *to* the toilet. I did end up hosing off a couple hundred malware skills that showed up to the party, but that’s the price of progress. Smooth, slick, simple. The future.”
Have I got the pattern down right yet? After an entire brain hemisphere threatened to go on strike after forcing it to read about “Gas Town” last month, I refuse on principle to read more than a sentence or two of this sort of thing.
@dpnash @jonny @davidgerard our country is being run by people who read Philip K Dick thought “Ubik” was the first TED talk
@jonny @davidgerard @dpnash
What "Gas Town"?
@musevg @jonny @davidgerard A fever dream from an experienced developer turned vibe coder, who decided what the world needs is (in his words) "Kubernetes for agents", and went ahead and vibe coded *that*.
It's as bad as you could imagine a vibe-coded-Kubernetes-for-agents might sound like. They have a blog post describing the application, which is about 30 screens (on a big desktop monitor) long and reads like a bad trip. We'd have probably heard more about it if Clawdbot...no, wait, Moltbot...excuse me, Openclaw hadn't shown up a couple weeks later and stolen the spotlight.
@davidgerard
This is a real meetup in my local area...
Alt...
Title from Meetup.com "How to Build AI Skills For Your Careet With Vibe Coding in 2026"
@davidgerard There is a whole field of technologies which never seem to transition from hype to something useful. For example in the Telco industry this is "IMS" (IP Multimedia Subsystem). It found a bit of use on 4G and 5G mobile phone networks for reserving bits so you can actually telephone over those networks even though your cell is full, but proponents wants that everywhere... and nobody is buying, since it's ultra complex and more expensive than building excess capacity.
Check out the skills repo (“clawhub”) for this new vibecoded openclaw security disaster, and realise it’s for techbros to manage losing their money betting on “prediction” markets. #Facepalm
@davidgerard Vibecoding is all fun and games until they realize their monster is now 5 years old, production critical and literally touching any part of it will break something else.
I've personally witnessed some really funny cases that aren't that funny for the coder any more.
Just wait for when "Agentic AI" madness hits full speed. It's going to be a horrible couple of years before all of this collapse under its own weight.
@hittitezombie @aamurusko79 they're having trouble getting Agentic AI to hit full speed. viz. it doesn't fuckin work lol
@davidgerard Having recently been forced to take a "skill evaluation" on "prompt engineering" at [redacted], (Which I managed to nail simply by guessing the answer they wanted from my decent understanding of the actual workings of LLMs as token-predictors) I can quite confidently say calling whatever that steaming hell-dung is "engineering" is an insult to engineering...
@becomethewaifu @davidgerard It's just engineering for people who don't know what engineering is.
@ColorfulCeleste @becomethewaifu @davidgerard To be fair, the whole field of software engineering is an insult to the actual engineering disciplines. Any mechanical engineer would frown at the things we do while creating software.
The internet: *foams at mouth how great claude opus 4.6 is*
Claude Opus 4.6: Doing a select on timer.C and when it fires, replacing the timer with nil is the idiomatic way to have an action run only once in go.
Me: but, wouldn't it be better to call timer.Stop(), which stops the timer and makes sure that a select on it would never fire again?
Claude opus 4.6: * proceedes to explain in detail why setting timer (not timer.C) to nil is better *
* Cries in nil pointer dereference *
@davidgerard I think we'll just end up slapping a "free range software" label on hand crafted software. Or "organic" software. Or something.
@davidgerard Don't forget the people who gushed about how NFTs would make it possible to buy tickets online, as if that hasn't been a solved problem for decades.
RE: https://mastodon.social/@stroughtonsmith/116030136026775832
This is one of the worst takes from LLM enthusiasts.
Compilers are deterministic, extremely well tested, made out of incredibly detailed specifications debated for months and properly formalized.
LLMs are random content generators with a whole lot of automatically trained heuristics. They can produce literally anything. Not a single person who built them can predict what the output will be for a given input.
Comparing both is a display of ignorance and dishonesty.
Much as you don't generally go auditing the bytecode or intermediate representation generated by your compiler, I think the idea of manually reviewing LLM-written code will fall by the wayside too. Like it or not, these agents are the new compilers, and prompting them is the new programming. Regardless of what happens with any AI bubble, this is just how things will be from now on; we've experienced a permanent, irreversible increase to the level of abstraction. We are all assembly programmers
@arroz it’s always a bit depressing when I find out about a new pocket of mediocre tech jackasses posting twitter crap on masto. all of the guys posting “LLMs are like compilers for natural language” should have their CS degrees yanked cause they’ve proven they don’t meet the academic requirements for a CS undergrad.
@arroz “LLMs are natural language compilers”, brought to you by the same kids insisting their product is “the operating system for the web” because nothing means anything if you ignore all implementation and engineering details
We'll see how I feel in the morning, but for now i seem to have convinced myself to actually read that fuckin anthropic paper
I just
I'm not actually in the habit of reading academic research papers like this. Is it normal to begin these things by confidently asserting your priors as fact, unsupported by anything in the study?
I suppose I should do the same, because there's no way it's not going to inform my read on this
"AI" is not actually a technology, in the way people would commonly understand that term.
If you're feeling extremely generous, you could say that AI is a marketing term for a loose and shifting bundle of technologies that have specific useful applications.
I am not feeling so generous.
AI is a technocratic political project for the purpose of industrializing knowledge work. The details of how it works are a distant secondary concern to the effect it has, which is to enclose and capture all knowledge work and make it dependent on capital.
So, back to the paper.
"How AI Impacts Skill Formation"
https://arxiv.org/abs/2601.20245
The very first sentence of the abstract:
> AI assistance produces significant productivity gains across professional domains, particularly for novice workers.
1. The evidence for this is mixed, and the effect is small.
2. That's not even the purpose of this study. The design of the study doesn't support drawing conclusions in this area.
Of course, the authors will repeat this claim frequently. Which brings us back to MY priors, which is that this is largely a political document.
@jenniferplusplus I like the fact that their own research doesn't fit their lazy claim you reference, and they spend a lot of time trying to work out how the claim can be true, even though their own evidence is against it (and more in line with the mixed evidence in the literature, as you say).
@jenniferplusplus it reminds me a bit of the famous thing with the Flat Earth Society people who spent $20k on an expensive laser gyroscope to "prove" that the Earth was not a rotating sphere... and then spent a lot of time being very confused and upset when, of course, it measured precisely what you'd expect from a rotating spherical Earth.
@aoanla @jenniferplusplus I was baffled that Anthropic published this paper, let alone promoted it on their blog. Cos even their headline results say "AI coding bots are shit, don't use them, they're no faster and they make you stupid". But yeah, they thought they were saying things about productivity.
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience the machines when they first came on the market. I want everyone's input! Please boost!
(I'll ask the same question about minicomputers. This poll is about the early consumer home computers released between say 1977 and 1994.)
| I had access to an 8 or 16-bit computer during their heyday: | 0 |
| I did not have access to an 8 or 16-bit computer during their heyday: | 0 |
| 8 or 16-bit computers had their heyday before I was born or when I was an infant: | 0 |
Well, everyone, you can now submit a comment to let the FCC know what you think about SpaceX asking for 1 million satellites for "AI datacenters" whatever the fuck that means.
https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf
Comments due March 6.
I am having a very hard time believing this is really happening. Fuck you, SpaceX, and fuck you, FCC. This is not regulation, this is a fucking joke, that will destroy our ability to use satellites for centuries.
@sundogplanets have you ever thought of using the surface of the ocean to show the non-maths how this looks on a smaller sphere? Get them to drive a boat through the grid or something tangible.
I don't think they understand 3D...or 2D
If anyone has time and energy to set up instructions for how to submit a comment to the FCC (it's really fucking complicated, on purpose, I'm sure), I would very much appreciate it! Otherwise I'll do it in the coming days.
@sundogplanets It's not going to work. That will be obvious long before he has all that many satellites up, and he'll move on to his next sick joke.
If somebody wants to venture into this, please test all steps.
The first one involves sending an email to ecfs@fcc.gov with "get form" and your email address in the message body.
The reply I got was trying to strangely gaslight me:
"Delivery has failed to these recipients or groups:
ecfs@fcc.gov
Your message couldn't be delivered. The Domain Name System (DNS) reported that the recipient's domain does not exist."
There seems to be a strange subdomain falstaff.fcc.gov involved. The attached error log says:
![[?]](data:image/png;base64, iVBORw0KGgoAAAANSUhEUgAAAEAAAABAAQAAAACCEkxzAAAAUUlEQVQoz43R0QkAMQwCUDdw/y3dwEvsvzlL4X1IoQkAisKmwfAFT3RgJHbQezpSRoXEqeqCL9BJBf7h3QbOCCxV5EVWMEMwG7K1/WODtlvxAYTtEsDU9F34AAAAAElFTkSuQmCC)
Diagnostic information for administrators:Generating server: SJ0PR09MB11735.namprd09.prod.outlook.com
ecfs@fcc.gov
Remote server returned '550 5.4.310 DNS domain falstaff.fcc.gov does not exist [Message=InfoDomainNonexistent] [LastAttemptedServerName=falstaff.fcc.gov] [SA2PEPF00003023.namprd09.prod.outlook.com 2026-02-05T12:30:46.776Z 08DE6078A5284768]'
@sundogplanets A.I. data centers in space is such a batshit crazy idea, it's hard to believe anyone takes it seriously. But they do.
It's just mind-blowing. Like living inside a comic book.
@sundogplanets let me marh this badly...
At a launch a second they could have that sucker up in months. At a launch every hour it'd take more than a century. At a launch a day millenia. Assuming of course one satellite per launch. That's just the getting to orbit bit. Fabbing the satellites might well take longer. After of course the lead time to ensure hallucinating chatbots are not on the worse granola.
@sundogplanets may be a blessing in desguise, i mean if (when) they fuck up, it will (force) the US to clean their mess by inventing tech to clean out there (if scientifically possible i'm not well known in space physic).
Elon the illegal South African Nazi predator just can't stop pushing his unwanted stubby nub of a weenie in people's faces.
What a dirty pig.
@sundogplanets the good thing is that Elon Musk has never delivered on anything he's ever promised, and this would require rocket launches at such an wildly high pace that I don't think they could come close to pulling this off.
It's just to juice SpaceX IPO and then musk can be a trillionaire. If the next US administration comes to be, this plan is dumpstered for sure.
@sundogplanets How would a data center event work in outer space? Heat would build up. Unless I'm missing something and the idea is to have something super worse than regular DCs down here?
Remember the hole in ozone days ?
Sheesh .
Think it was after “ give hoot don’t pollute “ and just before acids rain.
Anyway ,, Now we blast rocket fuel through it for the new computer science data barns.
🤔. Interesting swings in folk ideology.
@sundogplanets But SpaceX will “scal[e] to make a sentient sun to understand the Universe and extend the light of consciousness to the stars!” How could you possibly be against that?!?
Yeah, it’s batshit crazy, Musk is hyping for IPO (although I don’t know if he believes it himself… he’s huffed his own farts a few times too many). I’ll be astonished if more than, I dunno, a few dozen ever get launched. By that point, it should be clear what a stupid idea it is.
@sundogplanets Our economy is no longer serving the public, but the financial system.
Data centers in space won’t work, but the investment bankers don’t know that, so bullshit that make number go up rules.
Data centers in space are the tulip craze of our age!
Why would they even want AI datacenters in space? We haven't solved how to cool them without massive amounts of water down here on earth, how are they going to cool them up there? Some months back I read of someone else wanting to create the same datacenters in international waters. Is this just a new way for them to try to avoid legal liability? If that's the case, I wonder what kind of content they will generate up there.
@sundogplanets That means to not comply with Police investigations and Courts orders (grok, grokipedia,...) worldwide through alternative SpaceX internet except by use of force. That means controlling Space launches. In addition to polluting our sky and the Space
@sundogplanets With Kessler syndrome thought to be inevitable now I'm surprised this is even being entertained and policy isn't being drawn up to ration new satellites and mandate retrieval of spent/decommissioned/junk satellites in order to get approval for a new one.
science fiction coming alive ...
Alt...
poster for the silly TV show "Quark", a 1977 American science fiction sitcom starring Richard Benjamin. Quark was created by Buck Henry, co-creator of the spy spoof Get Smart.
The show was set on a United Galaxy Sanitation Patrol Cruiser, an interstellar garbage scow operating out of United Galaxy Space Station Perma One in the year 2226. Adam Quark, the main character, works to clean up trash in space by collecting "space baggies" with his trusted and highly unusual crew.
@sundogplanets Any chance the guidelines for commenting when the FCC was mulling axing/reinstating Net Neutrality are still valid at this point?
@sundogplanets Thanks for the great talk tonight in Hamilton Dr Sam. Scary but enlightening... time to spread this widely.
@sundogplanets plus space is not only a US thing (until some guy decide he wants a property title on it for some US thing). There is a higher authority for this.
@sundogplanets Many thanks for sharing the link - I wasn't aware of his at all. Absolutely disgusting! I shall be submitting some 'developmental feedback' to the FCC!
@sundogplanets 🤦🤦🤦🤦🤦
Alt...
Dr. Evil from the Austin Powers movies One Million Dollars meme, but with "One Million Satellites" as text instead
@sundogplanets This is more about the upcoming ipo than reality. It's to make dumb investors wet themselves with excitement. At that level, it'll probably work...
But still, needs a big NO from the FCC, who should not be the arbiter anyway, it's an international issue.
I wonder what the Chinese think about it?!!
@sundogplanets the environment agency should have their word to say. Because all these millions of datacenters are going to be disposed off... And they won't be recycled. Just burned up. In the open. No filters. No catalysators. Just burned in the atmosphere.
@sundogplanets Why would Trump want sophisticated communication systems to be available to just anyone?
@sundogplanets "unlimited number of satellites"
Soon our weather forecast will be predicting how many of those satellites will be falling down tonight.
@sundogplanets i think it means space x is about to go through an ipo and he is trying to scam investors for the thousandth time
@sundogplanets this will go as well as hyperloop did, if at all.
Good luck cooling these machines.
But a lot of waste and undesirable side effects will happen before the "proof concept" is delivered.
@sundogplanets I don't believe that anyone involved in this has actually contemplated the meaning of 'a million.' It is just a marketing buzz-word, and Flusk is putting it out there to flaunt his stupidity in front of everyone's faces.
It's strange and frustrating that most AI researchers don't seem interested in natural intelligence.
In the early days, when "neural networks" were seen as models of brains, many people seemed at least superficially interested in neuroscience. It's not like that now. I'm sure some folks would say "yeah, and aerospace engineers don't worry about bird flight, either!" but that feels wrong to me.
If all you care about is moving cargo, then sure, flight is solved, and who cares if our designs are "biologically realistic". Similarly, if all you care about is recognizing images, playing video games, and generating slop, then AI is solved. We'll just make the current solutions better.
But I think we've barely scratched the surface of what intelligence actually is! Current AI is so narrow and so shallow by comparison, yet I think people don't even notice that because they haven't actually thought about how intelligent living things are, and in how many different ways!
Not to mention the fact that so many of our current AI solutions are really just models of human intelligence, trained by example. Do we not care about creating genuine, original intelligence? Is bottling it up in a computer so that it's reproducible the same thing as making it in the first place? I'd say obviously not, but apparently many people don't see the difference...
I too have been struck by the apparent lack of curiosity of our colleagues. One tentative conclusion I've come to is that there is a subspecies of STEM folks driven largely by impressive demos. I feel like this tendency reflects the perversities of short-termist economic thinking, but in any case the view is a splashy demo that makes it into Nature and the NYT but has shoddy science backing it is superior to excellent science that has no splashy demo. The field "progresses" from one impressive demo to the next. Pollack used to say things along these lines, but that's no surprise given that he founded the DEMO lab: it's right there in the name! I think that mindset is not uncommon, though. Even those who agree the science is important run up against the constraint that there's significantly more funding for flashy demos than for basic research.
@abucci Completely agreed. I also think there's something to be said about how AI is ostensibly about understanding intelligence, but in practice seems more concerned with automating human cognitive tasks.
Here's John Searle in 1983:
Marvin Minsky of MIT says that the next generation of computers will be so intelligent that we will ‘be lucky if they are willing to keep us around the house as household pets.'Here's Joseph Weizenbaum in 2007:
Professor Marvin Minsky of MIT, once pronounced—a belief he still holds—that ‘‘the brain is merely a meat machine.’’He goes on to note that meat is dead and might be eaten or thrown out. Flesh is what's alive. He also draws attention to the word "merely", as in "nothing more than".
I share with Weizenbaum the belief that Minsky has clearly expressed a disdain for human intelligence. We're on the order of household pets. Our brains are no more than food or trash. Obviously Minsky doesn't speak for all AI researchers then or since, but his "meat machine" language is all over the place, and this disdain or even contempt for human intelligence and achievement is also common.
It definitely doesn't speak to a curiosity about intelligence, which I think requires at least a little bit of love and esteem.
You might be interested in some of the work being done by some of Daniela Rus’ graduate students. Rus is a roboticist , and she and her students were looking at what the 312 neurons of C.elegans could accomplish in contrast to the army of matrix entries in their “neural nets”.
Some differential equations later, they had a nineteen neuron structure that could pilot a quadcopter and follow a person around campus.
They have a startup, Liquid.ai, starting from there you can find some of their papers.
@lain_7 That sounds amazing! I was just thinking: robotics really is an exception to this rule. I think it's more apparent in that field just how much we still have to learn. I hadn't heard of that experiment, though! I'll definitely check it out.
I think a good place to start might be their paper, “Causal Navigation by Continuous-time Neural Networks”, which Google can find on arXiv.org.
That’s an early paper from before they actually built their quadcopter, but searching for the authors will find you more.
oh, and this:
https://www.eecs.mit.edu/drones-navigate-unseen-environments-with-liquid-neural-networks/
(I’m not sure where I got the 19 neuron figure from, maybe an informal talk by Rus.)
I should say, robotics really is the exception to this rule. There are some very cool biomimetic robots these days! And kind of always have been, I think. Perhaps they're more eager to draw inspiration from nature because it's more obvious just how much we have to learn in that domain?
Just today I was reviewing an excellent paper on the design of efficient quadruped gaits. I was delighted by all the anatomy and physiology references, and it's clear these ideas directly led to their very cool discovery. I'm looking forward to seeing that one get published. 🙂
And, of course, there's the ALife research community, which is into life-like intelligence in all its forms. I just don't think of that field as AI. It often feels more like a reaction to AI's overly pragmatic, anthropocentric approach.
I always thought it was weird how people worshiped Chomsky for his political thought when his field was linguistics.
And I always felt like the thrust of his analysis was of the "what did you expect of the US?" variety. I resented that because it de-activates people.
So I felt like people who wanted to talk at me about Chomsky were more interested in words than actions.
Didn't know he was a creep, though.
@D_J_Nathanson When I hit grad school I started interacting with women who had worked with him. So none of this is surprising to me, and boy do I wish he wasn't lionized the way he appears to be.
I have now destroyed my collection of pencil drawings.
Not quite as "destructively calming" as when I deleted a 20+ year-old mountain of code made just for myself.
@datarama Destroying them simply to declutter, or is there another reason? Are you doing okay?
@sitcom_nemesis because making them was pointless, and looking at them is a depressing reminder of attempted cope for a wasted life
@sitcom_nemesis The code I deleted in 2024 was because it sat there like a taunting, depressing monument to an entirely wasted life.
@datarama Sounds like you've been going through a really tough time. I'm sorry you're feeling like this 😢 .
@sitcom_nemesis The last 3 years have been, with no doubt, the most miserable of my life.
I actually *miss* the other contenders for "most miserable" - the 1½ years of total isolation in the start of the pandemic, or being a little disabled kid who got beaten up all the time, or getting by life well under the relative poverty line while dragging myself through alcohol abuse withdrawal. All of those times were terrible too, but at least I had *something* that made life worth living.
@sitcom_nemesis It turns out I can cope a lot better with life simply being tough than I can with nothing in life remaining that makes it worth living.
@datarama Humans tend to work like that, iirc it's a theme that came up in Victor Frankl's book Man's Search For Meaning.
Is there anything else in life that helps make you feel like it's worth living, that's harder for Big Tech to get its grubby hands on? Friends, perhaps? Family? Stardew Valley?
@sitcom_nemesis I basically have no social life anymore. The pandemic destroyed that. The only people I see regularly are coworkers - though I do still have online contact with two of my old university mates. I'm long-term single and never had kids.
In 2024, I think I actually managed to burn myself out on Stardew Valley.
I'll be contributing to "Hybrid Minds" in Vienna later this month (registration at link below).
I'll provide some philosophical reflections on the idea that we can replace organic parts of a living being one by one with mechanical ones - and end up with an equivalent system. I call it the "Cyborg Myth." It is part of the machine view of the world that regards organisms as mechanisms and thinking as computation, but neglects the problem of biological organization.
Tag yourself:
Alt...
where
p {\displaystyle \mathbf {p} } : constant pop,
c 0 {\displaystyle \mathbf {c} _{0}} : initial crackle,
c {\displaystyle \mathbf {c} } : final crackle,
s 0 {\displaystyle \mathbf {s} _{0}} : initial snap,
s {\displaystyle \mathbf {s} } : final snap,
j 0 {\displaystyle \mathbf {j} _{0}} : initial jerk,
j {\displaystyle \mathbf {j} } : final jerk,
a 0 {\displaystyle \mathbf {a} _{0}} : initial acceleration,
a {\displaystyle \mathbf {a} } : final acceleration,
v 0 {\displaystyle \mathbf {v} _{0}} : initial velocity,
v {\displaystyle \mathbf {v} } : final velocity,
r 0 {\displaystyle \mathbf {r} _{0}} : initial position,
r {\displaystyle \mathbf {r} } : final position,
t {\displaystyle t} : time between initial and final states.
@abucci yes, and yet it has substance, such that if you fly through it at speed it can cause turbulence...
So I see #Nolto doing the rounds as a supposed #OS #LinkedIn alternative
Judging from the Codeberg issues [1] and the self-professed ignorance of the prompter about fundamentals of federation and software dev [2], it seems highly likely that Nolto is vibe-coded
Superficially feature-rich but lots of loose ends and basic errors, this is a privacy and security disaster waiting to happen — I would not touch this with a barge-pole 😬
[2] https://codeberg.org/Tensetti/Nolto/issues/31#issuecomment-10292844
Alt...
List of currently open issues on Codeberg, many of thrm the typical loose ends that result fromvibe coding without deep understanding
Alt...
User pointing out: 'Also, the README file clearly says that everything is acctually implemented.'
Prompter JTensetti: 'Because it’s hard to stay on track with everything, I’m not working full time on this. I had it working and then something happened and now it’s not working.'
Addition: @sandorspruit points out that the readme discloses this clearly (in case there was any doubt)
It's interesting to me that the style of the disclaimer itself features the typical LLM-assisted hubris: emphasis on lowering cost (whose cost, and at what price?) and big words about "deliberate and manual" decisions, but relatively little to show for it.
I will take the hint and move on.
Alt...
🤖 About AI-assisted development
Nolto is built with the help of AI.
Much of the codebase is generated using an AI-assisted coding tool (Lovable.dev), based on explicit descriptions of system behavior, constraints, and architecture.
AI reduces the cost of producing code. Code itself is no longer the differentiator.
Product thinking, values, trust, federation design, and solving real problems are.
Those decisions are deliberate and manual.
If AI-assisted development isn’t your thing, that’s fine — feel free to move on.
@dingemansemark Is this even allowed on @Codeberg? It might violate §2.3 or §2.6.4 of the ToS as the LLM was trained on stolen work. This could make generated text illegal too, but that was not ruled on by a court yet.
@dingemansemark
Since @JTensetti is here, perhaps you could ask him about it directly? They did ditch the loveable.app site recently, for one.
@elmerot
I think @JTensetti is busy enough as is, and to be fair the readme is clear on the "AI assisted coding". I still decided to post because I think many people discover Nolto not via its repo but through its slick home page and / or a widely shared blog post about leaving LinkedIn, neither of which mention anything about vibe coding or Nolto's incomplete federation features
Nature discovers a level below Nature Scientific Reports
http://archive.today/2026.02.03-055607/https://www.nature.com/articles/d41586-026-00285-6
> The vision of human-level machine intelligence laid out by Alan Turing in the 1950s is now a reality. Eyes unclouded by dread or hype will help us to prepare for what comes next.
you fucking idiots
@davidgerard That is a very strong claim by a bunch of people with not really any track record in the field.
Nature really has found a bunch of wunderkinder.
@davidgerard I sometimes wonder if these particular people really are on par with their LLM love interests
@davidgerard My money says Danks and Bergen corralled this piece together. And that UCSD is looking for funding.
Also, AI tools screw up math. Why are people still having this discussion?
@davidgerard the thing nobody ever mentions is that in that same article Turing discusses ESP as proven scientific fact.
which isn't a knock on Turing but should raise some reservations about taking the article uncritically
@davidgerard @cstross You fucking idiots indeed.
Turing sadly failed to realise that most people would very easily find any old crap indistinguishable from people. The Test only proves something about people, not about AIs.
Wake me up when they’ve got to grips with Searle’s Chinese Room.
@BashStKid @cstross The Chinese Room is a puzzle about the dangers of being a Chinese student in the vicinity of John Searle
@davidgerard @cstross “But, Professor, I can help you with your translation problem!”
(Searle sticks fingers firmly in his ears)
@BashStKid @cstross Searle was notorious as a sexual harasser of Asian students in particular
I really wonder if these people claiming human-level 'intelligence' from machines ever spent time programming an Eliza, and I suspect the answer is "no".
@davidgerard @jaztrophysicist Is there a reason why I can't access to the link you posted? (I keep getting CAPTCHA in languages such as Thaï, Arabic, Greek, etc. )
@davidgerard Oh dear. I read the first five paragraphs and honestly can not face any more of this. Am now strenuously resisting the urge to say libellous things about the authors.
at last, an ethical chatbot https://www.maiasa.ai/
Maiasa, an #AI for the future that is ready for anything you throw at it*:
CALL NOW!
(via @davidgerard)
*) Maiasa's active vocabulary is limited to the single token "a", and it will reply to all prompts with only "a".
Alt...
Photo of serial entrepreneur/businessman Brian Butterfield, with superimposed the text "Call Now"
MESSAGE OF HIS HOLINESS POPE LEO XIV FOR THE 60TH WORLD DAY OF SOCIAL COMMUNICATIONS
His emphasis on face and voice is good.
Musk wants to merge SpaceX with xAI, then take it public
Or maybe Tesla
https://www.youtube.com/watch?v=VMKyocHwUJI&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20260202-musk-wants-to-merge-spacex-with-xai-then-go-public - podcast
time: 5 min 18 sec
https://pivot-to-ai.com/2026/02/02/musk-wants-to-merge-spacex-with-xai-then-take-it-public/ - blog post
Alt...
Elon Musk in an OCCUPY MARS shirt jumping in an X shape
@davidgerard Oh look, after parading his child around while he was destroying the US, he's planning to use another one of his babies as a shield to protect Twitter from the repercussions of being home to the for profit CSAM engine. 'You can't imvestigate or prosecute me, US national security needs twitter and rockets!'
Here's my list of established para-academic educational & research organizations:
https://docs.google.com/document/d/1q8HP1tMe1L42nZq-v_iB0dCIrGPYNe_sGeimtWep5x4/edit?tab=t.0
Related post: https://elftheory.substack.com/p/para-academia-is-the-future
I only add to the list when I stumble across things or when people make suggestions. I haven’t done any intentional research; I don’t even know what search terms I’d use, since “para-academic” hasn’t been enshrined & “alt-academic” has multiple conflicting meanings. Feel free to share & make suggestions, thanks!
Hello #fediverse! Thanks to my new DFF grant, I'm now looking to hire a PhD student to join me at AAU in Aalborg 🇩🇰 to work on "usable decentralization", i.e. on making distributed and federated cloud services accessible to the everyday user. For more details, see link below, and please don't hesitate to DM me with questions!
https://www.vacancies.aau.dk/phd-positions/show-vacancy/vacancyId/895183
#HCI #academia #getfedihired #decentralization #AAU #Aalborg #Denmark #selfhosted #selfhosting
Here's one point: masked thugs in mismatched camo are breaking laws that include abducting, assaulting, and executing people, and the regime is attempting to force us all to refer to this as "law enforcement". For all the many criticisms one might aim at US law enforcement, members typically wear blue, show their faces, and doctrinally aspire to behave transparently and professionally (please don't @ me with your critiques of law enforcement or hashtag ACAB comments; this is not meant as a defense, only a contrast. I've been beaten and teargassed by cops and witnessed far worse so I am acquainted with this aspect of US "law enforcement").
In other words the regime is attempting to change consensus reality so that Americans accept that the phrase "law enforcement" includes random, unprovoked breaking and entering by heavily armed masked men into homes or vehicles, assaults, and summary executions on the street, targeting all citizens. They want this to be a normal and accepted occurrence anywhere and at any time. Much as one might criticize previous administration's immigration policies---and one might really really criticize those---this attempt to shift consensus on "law enforcement" to include summary executions of anyone and all the rest is new (It is not new for certain groups; what I think Snyder is saying is that it'd be new for everyone to be targeted in this way, and for most people to accept that's just how it is now).
Snyder also points out that the border plays an important role here because it is where the country, and therefore the law, ends. In other words, it's not coincidental that the regime chose to elevate ICE. Historically, authoritarian regimes have a marked tendency to expand the lawlessness and indefiniteness of border zones to include the entire territory of the country, and the current regime is no different. Immigrants and immigration aren't the only targets here. The larger aim is to indefinitely suspend the rule of law nationwide by making the entire nation into a border zone (Recall the Texas governor kidnapping immigrants and shipping them to "blue states", a classic attempt to spread resentment of immigrants throughout the country). Again, this is a narrative move, an attempt to shift consensus reality.
So, one way to resist is to simply not accept either of these attempts to change reality. Continue to refer to what's happening as unacceptable, not who we are, etc. Continue to point out that out of control border "enforcement" has led to street executions. Continue to name these actions as the criminal acts of thugs. Continue to pressure people with power, such as lawmakers, to do the same. Not out of some misguided or naive nationalism or patriotism, but in order to keep a stake firmly planted in the ground against the forces attempting to move it. This is something we can all do.
Incidentally, all eyes on Haitian and Haitian-American people in Ohio over the coming weeks. The Haiti Temporary Protected Status designation ends this coming Tuesday, February 3, 2026. Haitians in Springfield were specifically abused during the Trump campaign. The US has a long history of abusing and dehumanizing Haitians dating back at least to Thomas Jefferson, so the Trump campaign rhetoric was no outlier or anomaly. Haiti is also one of the countries specifically named in the US State Dept's announcement about indefinitely halting immigrant visa processing. It would not be surprising if the next ICE "surge" targeted Haitian immigrants in Ohio given how the groundwork's been laid.
#USPol #ICE #history #authoritarianism #tyranny #resistance #resist #narrative #immigration #Haiti
In the schools and churches of Springfield, Ohio, people are making hasty preparations for a “large deportation” promised by the president. To all appearances, and according to local sources, the city is two or three days away from a federal ethnic cleansing, grounded in a hate campaign organized by the vice-president and American Nazis. The destined victims are ten thousand or more Haitians.From Timothy Snyder's substack. Note the use of the phrases "ethnic cleansing" and "hate campaign", which are accurate. Snyder later states, regarding his use of the word "Nazi", "I use the word advisedly". It's well worth reading what he says about that. J.D. Vance's words led to a self-professed US Nazi group terrorizing Springfield Ohio. Snyder spells out how the hyperreality--the non-existent parallel and false world Vance set in motion and then Trump amplified--is turning into real-world consequences for real people.
#USPol #ICE #history #authoritarianism #tyranny #resistance #resist #narrative #immigration #Haiti
The Trump-Vance lie that the city had been “destroyed,” the notion of “carnage,” the dehumanization of immigrants— all of this creates the impression that their promised ethnic cleansing action would be a response to something, rather than a simple choice to exercise state violence against an invented racial enemy. These reversals are very important. It is important to consider this carefully.#USPol #ICE #history #authoritarianism #tyranny #resistance #resist #narrative #immigration #HaitiFirst, Vance reported that the Nazi propaganda campaign that he had himself inspired amounted to factual evidence. Mental chaos has been created where there was none before. And then that mental chaos becomes the justification for physical chaos: Trump’s “large deportations,” ICE raids that will, in fact, wreck an improving local economy. And once that physical chaos has been created, it will be blamed on the immigrants who are no longer there. Most of this has already played out. A key threshold, which it appears that we are about to cross, is the application of the state violence. At that point, so to speak, the lie is supposed to become “true.”
Someone noted that ICE agents are apparently now just seizing people's phones without more procedure, and keeping them.
Which is robbery, and even armed robbery since ICE is armed.
Is that actionable ?
I mean, keeping the rule of law is a way to refuse the lie, as you wrote.
That said, my understanding is that if you are not under arrest, ICE has no right to access your phone unless they have a judicial warrant. You can file a criminal complaint of theft in the local jurisdiction where it happened (local politicians sometimes recommend this: https://www.bostonglobe.com/2025/06/06/metro/ice-immigration-rights-bystander/ ). The Alasaad v. McAleenan ruling found such searches unconstitutional. I don't know what followup rulings might have occurred or whether the consensus has shifted since then. You definitely do not need to consent to a search either of your person or of your phone.
Even if arrested, you are not required to speak to anyone other than to say "I wish to remain silent", "I do not consent to a search", and "I wish to speak to an attorney". You are entitled to contact an attorney. They will confiscate your belongings, but you do not need to tell them the password to unlock your phone.
Practically speaking, this is an agency that is willing to shoot people dead knowing that even the president will openly spread lies about what happened. They will almost surely threaten someone to unlock their phone and will take a copy of its contents regardless of whether they are legally entitled to do that. How much one resists is a question of how much danger one anticipates and what they are comfortable with risking. I'm definitely in no position to make recommendations about something like that.
The result of a consistent and total substitution of lies for factual truth is not that the lie will now be accepted as truth and truth be defamed as a lie, but that the sense by which we take our bearings in the real world—and the category of truth versus falsehood is among the mental means to this end—is being destroyed.Truth and Politics, Hannah Arendt
Yelp, don't consider an AI service such as ChatGPT to be a store of important content. Always keep a backup you control (which is in general good advice when working with cloud-based services)
> When I contacted OpenAI’s support, the first responses came from an AI agent. Only after repeated enquiries did a human employee respond
Err, yes, that is to be expected? We'll see this much more in the future. Expectations will shift.
@paulmelis frankly the information security and research integrity implications of this irresponsible piece by a tenured professor warrant a stronger response than 'keep a backup'. Dumping coauthored drafts, grant proposals, email correspondence, and student data (!) in a commercial ChatGPT Plus account, while knowingly giving OpenAI consent to train on all of it for >2 years, amounts to violating every professional guideline and breaching the trust of everyone you work with
I guess I'll be doing some insulating. This is the longest, coldest cold snap I've experienced here and the bright side of it is that it's alerted me to a trouble spot as well as the limits of our current systems!
Developers say AI coding tools work—and that's precisely what worries them
Ars spoke to several software devs about AI and found enthusiasm tempered by unease.
https://arstechnica.com/ai/2026/01/developers-say-ai-coding-tools-work-and-thats-precisely-what-worries-them/?utm_brand=arstechnica&utm_social-type=owned&utm_source=mastodon&utm_medium=social
@arstechnica Did you actually talk to developers?
Because the ones I know are uneasy because the "code" produced by these tools is as suspect as the other "content". At best it's replacing some rote code construction, but mostly it's expanding my workflow to be a bomb defuser.
Binance bitcoin bailout
Alt...
Headline: Binance to shift $1 billion user protection fund into bitcoin amid market rout Binance will convert the stablecoin holdings in its $1 billion Secure Asset Fund for Users to bitcoin over the next 30 days, with plans for regular audits
it does not seem to me to bode well that they’re getting this nervous when the bitcoin price hasn’t even dropped below $80,000
@molly0xfff I don't really understand this situation. Is this connected to the continuous devaluation of the US-$ in the last weeks?
@raisondetredev it mostly reads to me that they’re nervous about sinking bitcoin prices, and hoping to bolster them by a) injecting $1B of demand and b) publicly declaring their plans to do so in hopes that it will trigger others to buy
@molly0xfff Injecting $1B demand is nice, but ... what's on the other side of that $1B? Isn't that supposed to hold up the stablecoin or something? Is the stablecoin now bound to bitcoin?
@henryk no, USDC is reserve backed. the seller either decides to hold the USDC instead of the bitcoin, sell it for some other crypto asset, or redeem the USDC with Circle. it changes the seller’s portfolio, but USDC remains reserve-backed.
@molly0xfff
Where do you think this is heading? Is the billion enough to get bitcoin to seem secure, or will it keep dropping?
I'm confused.
How could $1 Billion of fake, fiat currency prop up a real, stable hard asset like bitcoin?
Shouldn't that be the other way around?
@molly0xfff Bitcoin has already been succeeded by more useful coins but it doesn't matter anyway, the reason is because it has no real world use case outside of being a more efficient version of western union.
@molly0xfff $IBIT holding 65 Bn, Strategy ~35 Bn alone, not counting whales.
One Bn is like feeding a cherry to a cow.
@molly0xfff are they buying #Bitcoin? Did they lose funds? Or is this just because of an expected dip?
Frankly it reads to me like an admission of their own guilt, the "brazen theft" being a projection, mens rea expressed as legal action.
Spotify and the three main major record labels sue Anna’s Archive for $13trillion for “brazen theft of millions of files containing nearly all of the world’s commercial sound recordings”
#culture #music #AnnasArchive #spotify #UMG #WarnerBros #Sony #lawsuit
As AI tools for writing become more common, let me throw one more worry into the mix: Students who write well without AI assistance may be falsely accused of #plagiarism by teachers using imperfect tools to detect AI-assisted writing.
Update. This fear is coming true.
We tested a new ChatGPT-detector for teachers. It flagged an innocent student.
https://www.washingtonpost.com/technology/2023/04/01/chatgpt-cheating-detection-turnitin/
"Five high school students helped our tech columnist test a #ChatGPT detector coming from #Turnitin to 2.1 million teachers. It missed enough to get someone in trouble."
Update. Of course teachers sometimes make false accusations of #plagiarism even without relying on imperfect tools. Now that they're on the lookout for #AI-generated submissions, the rate might increase.
New study: "When given a mixture of original and general abstracts, blinded human reviewers correctly identified 68% of generated abstracts as…generated by ChatGPT, but incorrectly identified 14% of original abstracts as being generated."
https://www.nature.com/articles/s41746-023-00819-6
Update. More examples of this fear coming true.
Professor Flunks All His Students After ChatGPT Falsely Claims It Wrote Their Papers
https://www.rollingstone.com/culture/culture-features/texas-am-chatgpt-ai-professor-flunks-students-false-claims-1234736601/
Update. More examples of this fear coming true.
AI Detection Tools Falsely Accuse International Students of Cheating
https://themarkup.org/machine-learning/2023/08/14/ai-detection-tools-falsely-accuse-international-students-of-cheating
Update. More examples of this fear coming true.
AI-Written Homework Is Rising. So Are False Accusations.
https://www.thedailybeast.com/ai-written-homework-is-rising-so-are-false-accusations
Update. How often might this fear come true? A new study "found that [an] #AI text detector erroneously identified up to 8% of the known real [human-written] abstracts as AI-generated text."
https://www.sciencedirect.com/science/article/pii/S2153353923001566
Update. Here's another study showing that tools to detect #AI-written text are easy to fool with "simple techniques to manipulate the AI generated content." But this one goes a step further and makes the right recommendation for teachers and schools. "GenAI tools and detectors…cannot currently be recommended for determining academic integrity violations due to accuracy limitations and the potential for false accusation."
https://educationaltechnologyjournal.springeropen.com/articles/10.1186/s41239-024-00487-w
Update. More evidence that this fear has come true.
https://www.bloomberg.com/news/features/2024-10-18/do-ai-detectors-work-students-face-false-cheating-accusations
"Even…a small error rate can quickly add up, given the vast number of student assignments each year, with potentially devastating consequences for students who are falsely flagged."
Update. This fear has come true to such an extent that students who write well without #AI assistance now feel pressure to use AI to "humanize" their writing and avoid the charge of AI-assisted #plagiarism. #Grrr.
https://www.nbcnews.com/tech/internet/college-students-ai-cheating-detectors-humanizers-rcna253878
@petersuber this unfortunately has seemed like the logical conclusion. I wonder if the version control on most document writers could help? Like if a student is working in Google Docs you could possibly tell if they are copy/pasting or taking the time to type up the document. Of course, some could still generate a paper, type up a sentence or two, watch a YouTube video, and repeat to make it look human. Of course, than this is work which calls to mind why plagiarize in this way.
@abucci @petersuber I should have specified. I’m not surprised students have turn to this given that schools have started checking through these means. I like the view that plagiarism, in terms of completely stealing another’s work, harms the culprit. As it defeats the entire purpose of education. This approaches the question of the end of education. Is it merely to be able to work in industry or to provide the student deeper understanding/a foundation for further study?
Still, it's the framing I object to. This technology has been forced down the throats of everyone, students included, and if surveys are to be believed it's been against a large majority's will. I don't think we should require the victims of this, especially not overburdened teachers and their students, to take extra steps to cope with the consequences. That's an injustice we should not accept. But the inclination to try to think of technology solutions to this grander social problem is doing exactly that, in my view. By now technologists have demonstrated beyond reasonable doubt that they do not have our collective best interests at heart, and we should therefore not be accepting their tools as solutions to the problems they themselves are heavily contributing to causing.
ICE agents shatter window, leave 1-month-old baby, mother in car after Portland arrest
Video shows federal immigration agents leaving behind an infant and broken glass after detaining a Guinean immigrant with no known criminal history.From https://www.pressherald.com/2026/01/28/ice-agents-shatter-window-leave-1-month-old-baby-mother-in-car-after-portland-arrest
Can someone clarify, in academia and industry are LLM hallucinations the result of overfitting, or simply a false positive?
I'm beginning to think that hallucinations are evidence of overfitting. It seems surprising that there are few attempts to articulate the underlying cause of hallucinations. Also, if the issue is overfitting, then increasing training time and datasets may not be an appropriate solution to the problem of hallucinations.
If you trained an LLM with only demonstrably true statements, they would still output false statements. They have no representation of truth at the level of the sequences they emit.
"Hallucination" is therefore a human judgment about the human user's reaction to LLM output and is not reflective of any semantic content of that output. All LLM outputs, whether they appear to have semantic content or not and regardless of whether that aligns with the user's expectations, are thus hallucinations in a sense.
@abucci Woah! This notion of interpolation threshold and double descent was a key insight that I was missing. Looking at the time stamps of papers detailing double descent most of them were published right after I had concluded my Deep Learning course. As I never worked at an industry, I missed this development. Initial perspective is that this seems to explain the generational by generational improvements of these models.
Well, this is going to fill my reading list for the next couple of days.
@abucci What's really interesting about this is seeing how this is our original conversation coming full circle.
The tragedy is watching it develop despite all the challenges and shortcomings.
@wwhitlow it's not over fitting or under fitting, it's simply a result of the stochastic nature of LLM and how they're not able to know that they don't know. They have to give an answer no matter what, and even if the answer is wrong, it simply outputs the most likely thing, which can include "hallucinations".
@clf
I can understand how the stochastic nature of the model leads to this. Given that it is a next token predictor. Really it’s recognizing that it was about 5 years ago that I first studied these models in a formal setting. I’ve recently returned to giving them serious thought and so I was curious as to how the decisions regarding training were affecting overall performance. While not drawing a one to one comparison, I can recall how problematic overfitting can be for Neural Nets.
Who are these eminent philosophers?
Anthropic describes this constitution as being written for Claude. Described as being "optimized for precision over accessibility." However, on a major philosophical claim it is clear that there is a great deal of ambiguity on how to even evaluate this. Eminent philosophers is an appeal to authority. If they are named, then it is possible to evaluate their claims in context. This is neither precise nor accessible.
#AI #LLM #Claude #philosophy #anthropic
Alt...
Screenshot from Anthropic's constitution. Highlighting the philosophical claim that is being called into question.
@abucci honestly, the same thought kept coming to my mind, as I was trying to make sure there wasn’t any details I had missed. Many of those references felt like how LLMs describe philosophical concepts. Which is typically limited in terms of specifics. The part that fuels and enables in-depth philosophical discussions.
@wwhitlow That's from here, right?
https://www.anthropic.com/constitution
They do need to name names.
@twsh Yes, that is the correct link. Details about Claude being the audience are in the preface. Screenshot and question about which philosophers comes from the section on Claude’s Nature.
A friend pointed out last night that WinRAR is more profitable than OpenAI and I can't stop thinking about it.
So it turns out every project with Gen AI generated code either had a team of human programmers spend more time debugging the slop code than would have been spent making it with zero Gen AI involvement...
Or the devs have totally lied about Gen AI producing the code. That's the torment nexus bot getting credit for human work, in order to shift capitalism into a direction where there's no paid human labour. So at least 95% of the population needs labour income, so with billions more people unable to even buy a banana... Profit?
@kimcrawley @davidgerard classic case of "we trained our bot on the entirity of a code base and the llm was kind of able to reproduce the code base in some way, with limitations, no real functionality, and we had to put in a lot of hours to kinda make it work". they really invented a terrible version of "cp ."
@ypislon @kimcrawley @davidgerard I will now be replacing the words "gen AI" with "lossy cp" in every AI coding pitch I read in the future
We don't tend to tolerate lossiness in text compression, especially when the decompressed form matters, but hey there's no time like the present to start!
@abucci lossy compression as in it's losing the compression humans intentionally put there in the first place
@kimcrawley@zeroes.ca @davidgerard@circumstances.run for translation, it's the same... It takes humans more time to clean up machine translations than it would take those humans to produce a clean translation from scratch... Unfortunately, in the case of translations, AI slop quality is deemed "good enough" by most companies. Hence thousands upon thousands of translators are out of work and we're all stuck with piss poor translations. Yay.
#SearX #SearXNG #SearchEngines #AlternateSearchEngines #MetaSearchEngines #web #dev #tech #FOSS #OpenSource #AI #AIPoisoning #AISlop #AI #GenAI #GenerativeAI #LLM #ChatGPT #Claude #Perplexity
@abucci@buc.ci I didn't went through all of them. But what are the end results of these threads?
Are there no open source search indexes for open source search engines to use?
Why must they rely upon Google/Bing etc?
@colinstu@birdbutt.com https://marginalia-search.com is one. The index is not complete, though if you find something missing you can ask it to index it. It also has a indie web discovery page.
I myself use Searxng. 4get is also a thing. But these are meta search engine not independent.
There's a lot of talk of AI being anti-human, or at the very least anti-Humanities, but this is missing the point. The AI assemblage is fine with both people and pedagogies as long as they are docile. What AI is anti, and what it undermines, is any tendency to refuse or revolt.
@danmcquillan sometimes i think of it as capital's attempt to replace what people want, as citizens of democracy and workers of the world, with "customer service", in which ostensibly "the customer is always right" (hence the sycophancy and total incapability of saying "i don't know") but in reality the customer is subject to, as with advertising & PR, intense psychosocial engineering and their consent + intent + values are ultimately things to be funneled towards profit maximization.
@abucci @danmcquillan when i call it “anti-human” what i mean is that it is, in addition to the things already mentioned, and extinction machine via energy and resource consumption and waste.
@abucci Right. The conflict isn't between biological and computational but between conformity and imagination.
Perhaps the most (in)famous and illustrious American computer scientist and acknowledged principal pioneer of the discipline now known as artificial intelligence (AI), Professor Marvin Minsky of MIT, once pronounced—a belief he still holds—that ‘‘the brain is merely a meat machine.’’ It is significant that the English language distinguishes between ‘‘flesh’’ on the one hand, and ‘‘meat’’ on the other. The latter is dead and may be eaten, thrown in the garbage, fed to pigs, and so on. Flesh, on the other hand, is living matter and, as such, deserves the respect and dignity for life of which, among others, Albert Schweitzer spoke eloquently. The word ‘‘merely’’ in Minsky’s sentence means essentially ‘‘nothing but,’’ that is, also not deserving unusual respect. His statement is a clear reflection of a profound contempt for life that, as I see it, is shared explicitly by important sectors of the AI community, the artificial intelligentsia, as well as many scientists, engineers, and ordinary people. Daniel C. Dennett, an important American philosopher, once said that we must give up our awe of life if we are to make further progress in AI.From Weizenbaum, Joseph (2007). Social and Political Impact of the Long-term History of Computing
What do these people actually mean when they shout that man is a machine? It is, as I’ve suggested, that human beings are ‘‘computable’’ (berechenbar), that they are not distinct from other objects in the world, in any way deserving of special respect or even attention. This then leads, at first gradually and then with exponentially increasing speed, to a view of human beings as mere objects who—no! Not who, that—can be exploited, inducted in killing machines, imprisoned, tortured, killed (providing they are ‘‘enemy combatants’’). It leads to the American military sponsoring programs to produce robot soldiers. What is then left of Norbert Wiener’s vision of the human use of human beings? And does not our world show us with utmost clarity how far we have already come?#AI #ComputerScience #human #life #TormentNexus #machinicI want here to emphasize, especially to this audience, that all this is not the fault of the computer. Guilt cannot be attributed to computers. But computers enable fantasies, many of them wonderful, but also those of people whose compulsion to play God overwhelms their ability to fathom the consequences of their attempt to turn their nightmares into reality.
@abucci "Evil begins when you begin to treat people as things." Terry Pratchett, I Shall Wear Midnight
I grew up watching gritty 1970s police procedurals, so every time someone is talking about health care and mentions their PCP there's part of my brain that is like "Angel dust?"
@peterb I have a similar, different problem any time someone mentions MCP.
Alt...
Image of the master control program from the original Tron film.
this is so self evidently stupid i don't really know how to approach writing about it https://www.propublica.org/article/trump-artificial-intelligence-google-gemini-transportation-regulations
how deliberately ignorant of everything would you have to be to think this was a good idea?
this ignorant i s'pose
@abucci @davidgerard Trump regime level stupidity.
Perhaps I should have bolded the "self-evident" part in my response, because that's the keyword.
@davidgerard
That'll be why they're putting a data centre under the ballroom. So they can do this shit 'in-house' without any of that nasty oversight that might come from using more widely available LLM set ups that are being forced into keeping chat logs (c.f. ChatGPT, and I suspect Grok very soon)
It fits with the Government of one (plus favoured lickspittles) the US has right now
#AI will produce regulation in perfect bureaucratic language. The probem is that we need new regulation to fix new problems. AI only has old regulation as its example. Garbage in, garbage out.
@davidgerard What happens when you put people who don't believe in governance in charge of government?
“We don’t need the perfect rule on XYZ. We don’t even need a very good rule on XYZ,” he said, according to the meeting notes. “We want good enough.” Zerzan added, “We’re flooding the zone.”
"Flooding the zone" is not how I usually think of regulations. In fact, it's kind of the opposite of what you want with regulations
1. Economists from the physiocrats (18th century) onward promised society freedom from material deprivation and hard physical labor in exchange for submitting to an economic arrangement of society
2. In a country like the US, material deprivation and hard physical labor have been significantly reduced since then:
If it seems that I am critical and suspicious of people making the big noise to combat fascism, it is only because I saw those same people get restless and bored with Covid mitigation mere months into the pandemic and surrender to soft eugenics so they could resume brunches and travel. People do not have the attention span for big problems without quick resolutions and will dip out the second they feel inconvenienced.
So yeah, I share your skepticism.
i can't possibly move to linux you must understand my PC not booting is load-bearing for my work
@davidgerard I wonder if "potential fixes and workarounds" includes firing all the slopslinging vibe coders and hiring real programmers again.
@theorangetheme "exploring potential fixes and workarounds" means "we have no fucking idea either"
@davidgerard "We're considering potential options and courses of action, up to and including 'doing something', 'making changes', and even 'fixing things'. In short: we have concepts of a plan, yes."
@davidgerard I want a disembodied voice to intone this article every time someone tells me windows is "enterprise grade" and linux (on the desktop) isn't
@errant this is the hard evidence that windows is enterprise software
"exploring potential fixes and workarounds" means Microsoft has no fucking idea what to do here either
Alt...
Microsoft says this issue is likely to impact users running Windows 11 version 24H2 and 25H2 on physical machines, and that it is exploring potential fixes and workarounds. In the meantime, if you do encounter this problem, you will need to manually recover your PC by entering the Windows Recovery Environment and uninstalling the latest January 2026 security patch.
@davidgerard “on physical machines” like “oopsie we accidentally broke your computer, guess you better just rent a VM from us, tee-hee”?
@arestelle *better run it in a VM on Linux
@davidgerard if you really need to run it at all
@arestelle i find the main barrier to going Linux is That One Fucking App, and a VM is an answer that works quite a lot
you can often reuse your box's OEM licence
@davidgerard @arestelle What's really funny is when That One Fucking App is also what's preventing an upgrade from Windows 10 to Windows 11, let alone onto linux
@Jer @davidgerard wait idk which one is That One
my one app was my music player and I just run it on Wine
@arestelle @Jer in my case it was Kindle Previewer, which didn't work in Wine at the time but was happy in a Windows 10 VM
@davidgerard @arestelle @Jer Scrivener, which is "works or I die and I'm taking everyone with me," does not work on anything but Windows and macOS.
And I'll stab anyone who calls Apple 'Unix' or 'good.'
@rootwyrm @arestelle @Jer it *used* to run in Wine didn't it?
@davidgerard @arestelle @Jer never ran quite right at all, doubly so when working with network storage.
@davidgerard
The shit that happens when you fire many developers in favor of your own AI that NOONE wants to use..
Couldn't happen to a more deserving company.
@davidgerard Of course they don't. They got Copilot to vibe-code some random feature, it spat out something that mostly works but with the odd side-effect, and nobody knows how it does what it does, as it's impossible to review.
LLMs are biting them, hard. It's been fairly obvious for the last year or so that the (already shoddy) quality of #Windows has plummeted. 30% of code is LLM-generated? Join the dots, Satya.
@davidgerard okay but in fairness, with Linux I have to do my own patching to make my computer unable to boot. That has been a lot of effort for me in the past, but I have occasionally pulled it off. MS is providing this as a service.
@brianpiper this is why Linux is not enterprise quality
@davidgerard @brianpiper the shareholders are not happy having to keep and pay in-house crews of experienced boot prevention engineers. 🤷
@davidgerard any time there's a tech headline like this with the word "botched" i can't help but think of https://www.youtube.com/watch?v=s3pk9CMwtN0
@davidgerard Have you tried throwing the PC in the sea and moving to a log cabin in the woods?
the palantir employees seem to have an acute case of the most obvious “hans, are we the baddies” syndrome, and it tears me up, really. so sad, poor fucking doves.
they should be ridiculed, laughed at, shamed, shunned and never employed by decent humans again.
coders for concentration camp it systems are no better than the gas chamber operators.
@mawhrin idk how common it is in other countries, but here when you graduate university you take an oath (though it's not binding) to not use the science, technology and knowledge you learned for harming people and society. They could have used that in their comp sci degrees.
On the bright side, lately I've been doing a lot of informal thinking while shoveling snow. I'm turning over an informal argument grounded on Chaitin incompleteness that if our physical universe has continuous space-time, then we must make non-computable leaps in our theories in order to increase the fidelity of our understanding. "Artificial scientists" running on computers will always have inescapable limits that don't apply to human beings. It's exactly the sort of wacky thing that makes for good shovel thinking: it passes the time, and there might be something in there that's more than passing theoretical fancy.
No grand ideas about the universe came to me, but I did realize something about Bénabou cosmoses I didn't know before so that's something.
you have more opportunities to provide external feedbackGiven how these products are being developed and deployed, it seems to me an appropriate translation of this is "you have more opportunities to fill in for the shortcomings in the product design and thereby convince yourself that you should pay a corporation when you use your own skills."
Putting on my CS hat, what you're describing is sometimes called a hilllclimber (if it actually operated as advertised, which it really doesn't as far as I understand). Hillclimbers notoriously get stuck at local suboptima. Setting lots of independent hillclimbers loose on a problem is called a population hillclimber or parallel hillclimber. They tend to have fewer problems with local optima. Of course there are significantly better heuristics in this class that could be used instead, and would be used in my opinion if the companies were serious about product design.
we only know Gas Town supposedly even works in any meaningful manner from posts that sound like they're written on amphetamines
@davidgerard it really is remarkable. Yegge used to be quite bright and rational. I mean, I'm blown away by what the best models can do now, but there's this weird cult of burning tokens to produce as much code as possible, and Yegge is fully in the cult. Beads is hundreds of thousands of lines of barely coherent code seemingly tuned to allow agents to chew tokens at maximum velocity. I can only imagine how crazy Gas Town is.
So, naturally, crypto bros will throw money at it.
@swelljoe @davidgerard Oh, I did not connect the dots. Yegge is totally nuts, sure, but I hadn’t figured out that people are pushing this stuff because it allows them to bridge cryptobollocks and LLM bullshit.
Thanks for the epiphany. I must go lie down now.
I tried to follow Travis' advice (about the Gas Town post, not tiktok), but I made it about three sentences in before I gave up.
Alt...
Screengrab of two posts at Bluesky by Travis Brown, which read as follows:
| think everybody should open the Gas Town post and slowly scroll all the way to the end for exactly the same reasons that | think everybody should spend 10 minutes watching unhinged We Are Charlie Kirk edits on TikTok .
The software developers are not okay.
@davidgerard I tried to read through a post about it and couldn't actually work out what the fuck "Gas Town" was supposed to be in any sense, not helped by the random slop illustrations which seemed irrelevant. It was like the programming equivalent of Time Cube only at least Time Cube was a self-contained thesis rather than just a sprawling mess of nonsense
@j0ebaldw1n yeah, the nonsense is the product
@j0ebaldw1n @davidgerard Thank fck, it wasn't just me, I saw it on lobst.rs and thought I was going crazy.
@chrisp @j0ebaldw1n remember, Lobsters is meant to have the atmosphere of a *garden party*
one where AI salesmen are invited to talk over everyone else
@davidgerard Assuming the purpose of Gas Town is 'push the overton window so that "I'm only using the one LLM" becomes the moderate position' ...
... then it seems to me that it's working great
@davidgerard I don't think this is literally true in like a "Steve Yegge has sat down and thought about how he can shift the Overton window on LLMs" way, but I do think it's true in a vaguer "purpose of a system is what it does" way
@davidgerard Gotta be honest, it's hard to turn down this pitch: "Gas Town is also expensive as hell. You won’t like Gas Town if you ever have to think, even for a moment, about where money comes from."
@davidgerard
This is my question.
Where are the code generated open source projects that everyone loves and uses and contributes to?
Where is the evidence of generated, working code?
If the productivity gain is so large, where is the clear and inspectable evidence of such?
@EricCarroll The claim I keep hearing is that everybody is making individual, bespoke software for themselves now.
So it‘s definitely there, you just can‘t see it.
@chris_evelyn @EricCarroll @davidgerard
I think making bespoke generative software will wind up causing more problems than it will solve.
@davidgerard it's real? I just assumed it was a particularly wild satire and/or an attempt to scoop up a little of the craziest money in the market.
Which is why I'm launching Poe's Lawyer, an LLM which detects sarcasm, satire and hyperbole and is fully integrated with the PoeCoin Futures Market: let the wisdom of crowds determine if you were joking or not and invest in a slice of the zeitgeist.
Gas Town? Gas Town is the thing that’s convincing holdouts that vibe coding is the future? Really?
@baldur what’s a gas town? is it this? https://github.com/steveyegge/gastown
@davidgerard @bobschi @baldur I thought never meet your heroes applied to hard-rock heroin enthusiasts and actors who are secretly sex pests. I'm so disappointed in this wave of high profile devs getting hooked on vibes. We live in the darkest timeline
@svines @davidgerard @baldur the writing’s been on the wall for the last two decades. i am just privileged enough i could ignore it for the first few years of that time …
@svines @davidgerard @baldur those two decades are for me, individually, from when i finished school and decided to study computer science despite all my doubts. basically because i couldn’t think of something better to do. reading the tim berners-lee bio atm, i think the problem was there way longer. *sigh*
@baldur Gas Town looks a lot like a straight up grift and a characteristic of grifters is that they know how to appeal to new marks; it's fundamental to their existence.
@baldur like, it’s weirdly impressive, but in that horrifying way where you can’t stop gawking at the audacity and incredible wastefulness
@baldur the "convoluted for its own sake, fun to get lost in the self-made hell of an architecture" pitch makes me feel like these people just need to discover videogames. i got back into Satisfactory this weekend and nobody else had to hear about my descent into madness. https://www.satisfactorygame.com
Never buy a OnePlus phone ever again. They now have a hardware anti-rollback fuse that blows if you revert to an earlier version or install a custom ROM.
https://consumerrights.wiki/w/Oneplus_phone_update_introduces_hardware_anti-rollback
@davidgerard people shouldn't buy OnePlus in the first place, because holy shit what garbage devices. Which get completely abandoned in 12 months or less.
@davidgerard
Something doesn't check out.
So there are fuses in Qualcomm SoCs that can somehow be blown from the software? Fuses that, if blown, _only_ prevent installation of certain operating systems, but have no other effect on the functioning of the device?
I have a hard time believing whoever wrote that piece knows what a fuse is.
@davidgerard manau, ES išpistų už tokius dalykus; Color OS yra skirta CN rinkai.
Beje, kaip tik neseniai nusipirkau OnePlus 13 (trečias telefonas iš jų), jaučiu kokybę šiek tiek santykinai žemesnę palyginus su OnePlus One.
Thank you for the Info, I had been looking toward purchasing a one plus device.
Now one Plus and OPPO are off my shopping list.
@davidgerard
I didn't know anything about that device, but I'm guessing that preemptively replacing the fuse before ROM changes is not an option. Is that right?
@davidgerard
So if I buy a OnePlus and it gets a tiny scratch, I just need to to attempt a downgrade and return it for a warranty replacement🤔
(Yes, I live in a country where they would have to prove that the fault is not said fuse to refuse a warranty claim).
@davidgerard Never buy anything Qualcomm again. They are the ones who put this mechanism in their CPUs.
"The anti-rollback mechanism uses Qfprom, a region on Qualcomm processors containing one-time programmable electronic fuses. These microscopic components are physically altered when "blown"; a controlled voltage pulse permanently changes the fuse's state from "0" to "1." This change cannot be reversed by any software means."
@davidgerard My phones over the last 6 years have been Oneplus. Looks like I'll have to start saving for a FOSS one when it's replacement time. :/
@davidgerard The Oneplus One was my first and last phone from them, pretty poor support even a year after it came out
@davidgerard Hm that narrows down the field of possible future purchases.
I was pretty happy with my OnePlus One, I must say. In fact it still works, but runs Ubuntu Touch. Part of the touch screen is broken, so I can only use it for dev stuff.
But yeah, they abandoned support for the phone as soon as its successor came out, so cannot recommend them unless you're going to run, say LineageOS on their phones. And with this anti-rollback move, they've said they're against openness and reuse... not going back.
@davidgerard My oneplus 6T has been one of my best phone, easy to install custom rom and lasted me for six years. Sad to learn that it's no longer a reliable brand.
@davidgerard concept of a block on a roll back isn't that new. Google Pixels couldnt roll back to Android 13 from 14 even with custom roms, as I remember
@davidgerard That's just anti-rollback protection and actually a pretty common feature on modern Android. They should have disclosed for which vulnerability they're burning it, and their software should have disabled the protection for bootloader-unlocked phones (maybe it does? Doesn't seem clear from the link to me). But other than that, this doesn't seem particularly surprising, and you'll likely see similar things from other vendors, too.
@davidgerard Hardware limitations like this suck. I loved my OnePlus (yes, including the OS) but had to switch because of the ridiculously short security update support. And the prices are no longer that good compared to other phone brands.
@davidgerard I don't use my phone much and usually have either my wife's hand me downs or some rando we got cheap.
My current phone is a oneplus nord n30. i've been meaning to put custom rom on it but haven't had the time.
Guess now I'll have to check to see if that will brick it.
I have already made the decision to just plan on paying top dollar going forward to have devices that are freedom respecting and want the device to serve me instead of the other way around.
@Jason_Dodd my fairphone 5 is nice, disadvantages: noticeably heavy (220g), not at all waterproof (the plastic back cover just pops off for your convenience) and the front camera's not good
other than that it's an extremely good high end phone
@davidgerard i'm thinking that when i get rid of this it will be a fairphone.
none of those negatives are issues for me.
@davidgerard seeing clippy and talking about phones compels me to ask, "what do you think of rossman's phone company?"
@Jason_Dodd hadn't heard of it until your comment!
i mean looks like a nice idea, i have no idea if anyone's made a list of obvious issues
@davidgerard i was thinking of changing providers when i'd heard of it. i'm slow and lazy so it will probably be a month or so but i might give it a try if for no other reason than i like to promote the general things he does.
@davidgerard @Jason_Dodd Although the weight is compensated for by all the money you don't now have weighing down your trousers after buying it. :-)
@denisbloodnok @davidgerard lol. I suppose when it comes to devices and companies continually treating us worse and worse, i'll opt for paying more when i trust that won't be the same.
@Jason_Dodd @denisbloodnok the fairphone is reasonably priced for the loadout, the 6 is £549 on the site
(mine was a gift i should note)
@davidgerard i agree. the only thing is that i'm usually on the trailing edge when it comes to devices. my main computer or phone almost always costs me less than $100. so me paying for a fairphone or a framework laptop is saying something.
@Jason_Dodd i mean if i didn't get this i was gonna get a second hand note 13 🙂 ~£170. laptops are ex-corporate kit off eBay for £300 total.
The HMD Skyline manages replaceable battery and IP65 (find a video to show how it comes apart! ) , but it's closed-ROM with no known exploits - ( would LOVE to be 'well, actually'd about this.)
#HMD #GraphineOS
@Orb2069 @Jason_Dodd yeah fairphone 5 and 6 manage IP55 lol
fairphone sells phones with stock Android and e/os, and is not ROM-hostile (e.g. PostMarketOS is not official, it installs though not everything works yet)
History