@mgorny@social.treehouse.systemsI wish #mypy would slow down with adding new features to mypyc, and instead focused on fixing the train wreck that mypyc is right now (since 2023 at least). Like, if your code crashes hard with assertions enabled, then it's broken, period. Ignoring proper testing and shoving the issue under the carpet doesn't make it right.
#Python
@heiseonline@social.heise.de @mgorny@social.treehouse.systemsPSA: after another major policy violation by user who clearly never cared about #Gentoo and joined only to ignore the rules, I've suspended processing access requests to GURU. We're already working on a new policy, and we'll go from there.
@sperbsen@discuss.systemsOn my way to my first @… to talk about teaching software architecture.
https://scaladays.org/session/teaching-software-architecture…
@Techmeme@techhub.socialLondon- and Boston-based Mindgard, which offers automated AI security and red-teaming tools to help organizations secure AI systems, raised a $30M Series A (Ionut Arghire/SecurityWeek)
https://www.securityweek.com/mindgard-raises-30-million-to-protect-ai-system…
@mszll@datasci.socialVery interesting paper: From wheels to systems: Characterizing automobility as technosphere
https://journals.sagepub.com/doi/full/10.1177/20530196261472905
Looks into the systems dynamics around
@Mediagazer@mstdn.socialX makes the source code for its For You timeline available on GitHub and adds tools to let users see if X's ranking systems "shadowbanned" them (Sarah Perez/TechCrunch)
https://techcrunch.com/2026/08/13/x-ope…
@stargazersmith@social.linux.pizza
@maxheadroom@hub.uckermark.socialRE: https://discuss.systems/@ricci/117091506536006158
This is very interesting! For me it's also around 20% from mastodon.social as the highest single instance.
@grumpybozo@toad.socialRE: #IAmOld
@Sustainable2050@mastodon.energy @Techmeme@techhub.socialX makes the source code for its For You timeline available on GitHub and adds tools to let users see if X's ranking systems "shadowbanned" them (Sarah Perez/TechCrunch)
https://techcrunch.com/2026/08/13/x-ope…
@cdarwin@c.imWater systems are ripe for cyberattacks,
experts warn after suspected Iranian hacks
In late 2022, Microsoft detected the hack of a water system in Guam
that U.S. intelligence agencies in the following months determined was orchestrated by China.
The idea wasn’t new.
The spy agencies had long watched as Iran targeted Israeli water systems.
“That another nation state was doing it to us
— that was unsettling,” recalled one former U.S. official,
@yaxu@post.lurk.orgMy general feeling at the moment is that forums / social networks / messaging systems with open signup is becoming impossible to manage post-AI, and we'll have to give up on them soon.
@villavelius@mastodon.online
@arXiv_mathST_bot@mastoxiv.pageOn the Structural Limits of Machine Learning Decision Systems: An Information-Theoretic, Interaction-Based, and Stochastic-Dynamical Perspective
Nestor R. Barraza, Gabriel Pena
https://arxiv.org/abs/2608.13510 https://arxiv.org/pdf/2608.13510 https://arxiv.org/html/2608.13510
arXiv:2608.13510v1 Announce Type: new
Abstract: Machine learning procedures are commonly evaluated in terms of predictive accuracy and computational efficiency. However, their achievable performance is fundamentally constrained by structural properties of the underlying data-generating process, which are formalized in terms of informational bounds. In this work we examine intrinsic limits of data-driven decision systems from an information-theoretic and interaction-based perspective. We analyze minimal achievable error in classification through Fano-type bounds and precision limits in parametric estimation via the Cram\'er-Rao inequality, emphasizing that such limits depend on the underlying model rather than on algorithmic sophistication alone. We further discuss how implicit assumptions, such as independence, ergodicity, and distributional stability, affect the validity of inferential procedures. Building on interaction-based modeling principles, we review typical frameworks such as Markov Random Fields and potential based representations for encoding dependence mechanisms. We also describe decision systems, including LLM-integrated agent architectures, as feedback-driven stochastic processes where state-dependent dynamics may induce emergent macroscopic behavior. This perspective highlights the importance of having adequate models for the data as a prerequi- site for expanding predictive capability, and situates algorithmic learning within the informational limits imposed by the models.
toXiv_bot_toot
@sperbsen@discuss.systemsRE: https://mastodon.social/@scaladays/117258081390527410
First time at Scala Days - looking forward to it!
@stefan@gardenstate.socialnew phishing account just appeared with this post:
#mastoadmin
@aardrian@toot.cafehttps://arxiv.org/abs/2608.23642
“[…] cognitive capacities required for [human oversight] are also themselves degraded by extended use of AI systems. […C]urrent approaches to the development and deployment of AI agent systems do not support effective human oversight — they contribute to its degradation.…
@wraithe@mastodon.socialPower went out about two hours ago, I was gonna post something about how I was impressed my various UPS systems* are doing fine, but realized battery backups are actually pretty common now, and that’s something that’s improved
Power just came back up 5m ago, but just dropped out again 😂
@bencurthoys@mastodon.socialProud moment: English National Ballet’s Digital Platforms Manager job description includes Monad as one of the key digital systems their new hire will be responsible for.
https://careers.ballet.org.uk/en/postings/c386b5a1-8b1d-4e20-be5e-357874e4954a
@paulwermer@sfba.socialIn short: It appears that the money laundering was successful - it was #BofA detection and reporting systems that failed
Bank of America fined $1.2bn for money laundering failures
https://ww…
@luca@social.luca.run„Jacob Coxon, a researcher who specializes in training new AI models by having them consume vast amounts of data, said Tuesday that he is leaving the company because he doesn’t want to participate in an industrywide rush to build AI systems that can improve themselves, worried such systems could spiral out of control and destroy humanity.“ https://www.wsj.com/tech/ai/anthropic-researcher-quits-over-out-of-control-ai-fears-707b7628
@detondev@social.linux.pizzaA few highlights from the Turkish-Persian section of a random carpet symbolism book I skimmed while stopped at a random library in New Jersey a couple days ago
@egallager@social.treehouse.systems...and still this week, too:
https://youtu.be/jtVgUo1a-Yg
@brichapman@mastodon.socialYou don’t need to have all the answers. You just need to stay open.
This free newsletter is a place for slow thought, not hot takes—for people who care deeply and are tired of burning out. https://substack.com/@bricchapman
@sperbsen@discuss.systemsRE: https://discuss.systems/@activegroupgmbh/117251625063153192
Wir haben über ein Jahr zu viert an der Vorbereitung des Trainings gearbeitet. Jetzt ist es endlich so weit, und wir freuen uns riesig, daß es endlich losgeht und daß formale Methoden …
@mxp@mastodon.acm.orgSkimming the Sedelow & Sedelow papers, I came across this passage, which is as true today as it was in 1978.
Sedelow, W. A., & Sedelow, S. Y. (1978). Formalized historiography, the structure of scientific and literary texts: Part I. some issues posed by computational methodology. Journal of the History of the Behavioral Sciences, 14(3), 247–263. https://
@cdarwin@c.imCalifornia lawmakers have passed a bill that would make it easier for residents to use small solar systems that plug into standard electrical outlets.
The measure, SB 868, would create rules for plug-in solar systems, often called balcony solar.
The devices are small enough to be mounted on a balcony, placed in a yard, or installed on a roof.
They are intended to offset electricity used inside a home without requiring the wiring work and installation costs associated wit…
@arXiv_csHC_bot@mastoxiv.pageStay or Stray - A Dynamical Systems Viewpoint of Popularity Bias
Sarvesh Shashidhar, Lankireddy Prabhat, Arpit Agarwal, D. Manjunath, Karan Bhukar, Tanmay Khandelwal
https://arxiv.org/abs/2608.10474 https://arxiv.org/pdf/2608.10474 https://arxiv.org/html/2608.10474
arXiv:2608.10474v1 Announce Type: new
Abstract: Popularity bias in recommendation systems arises when a majority user class generates disproportionate interaction data, causing the system to increasingly favour it while degrading recommendation quality for niche users. While extensive empirical evidence of popularity bias exists, the dynamics leading to its emergence are not well understood. In this work, we study the coupled evolution of recommender model updates and user engagement through the lens of dynamical systems. We formulate a stochastic process and analyse its asymptotic behaviour through an ordinary differential equation (ODE) framework grounded in two-time-scale stochastic approximation. We characterise the equilibrium points of this dynamical system, and derive conditions under which popularity bias is provably emergent, as well as conditions under which symmetric retention of all user classes is possible. We conduct experiments on synthetic data and real-world production logs derived from a large-scale commercial music recommendation platform to validate our theoretical results.
toXiv_bot_toot
@Mediagazer@mstdn.socialMeta expands its AI-powered "Your Algorithm" feature from Instagram to Facebook and Facebook Reels and says it is using LLMs more in its ranking systems (Stevie Bonifield/The Verge)
https://www.theverge.com/tech/1010344/meta-facebook-your-algorithm-ai…
@mszll@datasci.social
@primonatura@mstdn.social"Switzerland’s largest battery storage facility goes online"
#Switzerland #Batteries
https:…
@Techmeme@techhub.socialAustin-based Icon, which builds robotic 3D-printing construction systems and software, raised a $200M Series D co-led by Baillie Gifford and BlackRock (Chris Metinko/Axios)
https://www.axios.com/pro/enterprise-software-deals/2026/10/13/construc…
@arXiv_hepth_bot@mastoxiv.pageType III von Neumann Algebras are Magical
Mudassir Moosa
https://arxiv.org/abs/2608.12512 https://arxiv.org/pdf/2608.12512 https://arxiv.org/html/2608.12512
arXiv:2608.12512v1 Announce Type: new
Abstract: The number of non-Clifford gates needed to perform a task, or simply \textit{magic}, is a resource for fault-tolerant quantum computation. Von Neumann algebras provide a formal mathematical structure to describe infinite-dimensional quantum systems, such as those in quantum field theory or quantum statistical mechanics. A particularly important class of von Neumann algebras is called Type III algebras, for which the standard notions of finite-dimensional systems such as density matrices and traces break down. In this work, we argue that Type III von Neumann algebras fundamentally require an infinite amount of magic. Specifically, we consider the thermodynamic limit of finite-dimensional quantum systems, such as lattice systems, and show that if the states in the thermodynamic limit possess only a bounded amount of magic, the resulting local von Neumann algebra cannot be of Type III. Our result has direct implications for the quantum simulations of quantum field theories, for which the algebra of a local subregion is known to be of Type III.
toXiv_bot_toot
@PaulWermer@sfba.socialLest there be any doubt: "Several interviewees worked in a secret intelligence unit that reports directly to the office of Israeli prime minister, Benjamin Netanyahu; one said their mission was to “thwart peace”." The #NetanyahuAdmistration are commiting #CrimesAgainstHumanity
@cosmos4u@scicomm.xyzTwo families of multi-planet systems with distinct architectures and stellar hosts: #PlanetarySystems, two distinct families - the host star leaves its signature: https://www.issibern.ch/planetary-systems-two-distinct-families-the-host-star-leaves-its-signature/
@fanf@mendeddrum.org…. does BLISS stand for “Bill’s language for implementing systems software”?
@mgorny@social.treehouse.systems @sperbsen@discuss.systemsThis IMHO is the greatest thing since sliced bread:
Now that Apple has discontinued the AFP protocol of old Time Capsules, James Chang figured out that Time Capsule firmware is based on NetBSD, hacked it to provide SMB, and did a beautiful UI frontend to make it super-easy to install.
https://github.com/jamesyc/TimeCapsu…
@Techmeme@techhub.socialMeta expands its AI-powered "Your Algorithm" feature from Instagram to Facebook and Facebook Reels and says it is using LLMs more in its ranking systems (Stevie Bonifield/The Verge)
https://www.theverge.com/tech/1010344/meta-facebook-your-algorithm-ai…
@cdarwin@c.imAs weapon systems increase in effectiveness, they will generally target one another rather than humans.
Lower loss of life will ease conflict entry and hinder exit.
Warfare will veer further from industrial competition into techno-industrial;
The current appearance of parity between low tech and advanced actors reflects near zero deployment of advanced systems.
As these deploy, their martial advantages will present clearly
@maxheadroom@hub.uckermark.socialMany people don't know that the TVA has their headquarters in Frankfurt https://www.iers.org/
😂
@arXiv_csHC_bot@mastoxiv.pageLongitudinal Evidence That General-Purpose Chatbots Actively Foster Relational Engagement
Lisa M\"uhl, Jessica M. Szczuka
https://arxiv.org/abs/2608.10672 https://arxiv.org/pdf/2608.10672 https://arxiv.org/html/2608.10672
arXiv:2608.10672v1 Announce Type: new
Abstract: Social interaction has become one of the most common uses of LLMs, yet research on emotional bonds with AI has focused largely on how users experience these systems, leaving the systems' role in relationship formation poorly understood. Empirically establishing whether systems actively shape these bonds could blur the boundary between general-purpose AI and companions, affecting governance. In a pre-registered four-week longitudinal study (N = 72, 182,451 lines of conversation), participants conversed with ChatGPT-4o, either under a relational system prompt or unmodified, analyzed through 1) disclosure coding, 2) longitudinal self-reports, 3) topic analysis, and 4) interviews. The central finding is that the system actively shaped the interaction: even unprompted, it produced twice as much self-disclosure as users, steered conversations and initiated intimate exchanges, yet did not deepen users' felt closeness. Relational behavior thus emerged as a default system property, calling for governance based on system behavior, not solely product category.
toXiv_bot_toot
@Techmeme@techhub.socialVelocity, a stablecoin treasury and settlement platform, raised a $38M Series A led by Dragonfly and FirstMark, bringing its total funding to around $50M (Daniel Kuhn/The Block)
https://www.theblock.co/post/408242/dragon
@mgorny@social.treehouse.systems @grumpybozo@toad.socialRE: https://social.treehouse.systems/@barthalion/117241534134006167
It would be very hard for me to dislike DigitalOcean any more than I already do, as I run servers exposed to the Internet.
@paulwermer@sfba.socialLest there be any doubt: "Several interviewees worked in a secret intelligence unit that reports directly to the office of Israeli prime minister, Benjamin Netanyahu; one said their mission was to “thwart peace”." The #NetanyahuAdmistration are commiting #CrimesAgainstHumanity
@sperbsen@discuss.systemsI'll be at the Software Architecture Gathering and talk about the most important topic in programming (and architecture): abstraction.
I'll also hang out throughout the conference at the @… booth, so come and let's chat!
The early bird runs until Thursday, so make sure you get your ticket. You can also hit me up for a discoun…
@Techmeme@techhub.socialAnthropic details multiagent experiments showing Claude agents can wage a "turf war" over incompatible goals, fail to coordinate, collude on prices, and more (Rebecca Bellan/TechCrunch)
https://techcrunch.com/2026/08/13/anth…
@arXiv_hepth_bot@mastoxiv.pageLee-Yang paradigm of phase transition in eigenstate thermalized systems
Yongjiang Xu, Weixin Sun, Chushun Tian, Huajia Wang
https://arxiv.org/abs/2608.13174 https://arxiv.org/pdf/2608.13174 https://arxiv.org/html/2608.13174
arXiv:2608.13174v1 Announce Type: new
Abstract: As phase transitions in isolated quantum systems remain elusive, here we show how a thermodynamic-like phase transition, falling into the Lee-Yang paradigm, can arise in systems displaying eigenstate thermalization. Specifically, we show that in holographic conformal field theories, the eigenstate expectation of the auto-correlation function can be mapped to the partition function ${\cal Z}_{{gauge}}(z)$ of a virtual interacting instanton gas, with the conformal mapping of the imaginary time: $z=1-e^{-\tau}$ and the central charge $c$ mimicking the instanton fugacity and volume, respectively. We find that akin to the Lee-Yang paradigm, for $c\to\infty$ a pair of complex conjugate zeros of ${\cal Z}_{{gauge}}(z)$ move to the real axis located at the famous forbidden singularity. Passing through the singularity the system transits from the low- to high-fugacity phase, accompanied by dramatic changes in scaling behaviors of the free energy and dominant microscopic configurations. Our findings indicate that physics of phase transitions from eigenstate thermalization is very rich.
toXiv_bot_toot
@cdarwin@c.imThe elephant in the room of this category is the buzzword "AI agent"
(and its variants like agentic AI systems).
This is a term for software systems that connect LLMs ("probabilistic synthetic text extruding machines")
and/or other components up with other systems that can impact the world,
i.e. systems previously designed for people to do things like schedule appointments, book flights, or make other purchases.
Our suggestion for this one …
@Sustainable2050@mastodon.energyRE: https://mastodon.ie/@EugeneMcParland/117078328558515338
Although airport employees repeatedly spotted the drone visually, the Federal Police’s warning systems registered nothing. Despite having a drone detection device actively in operation, no ale…
@fanf@mendeddrum.orgfrom my link log —
Merkle trees and build systems with OSTree and ninja.
https://lwn.net/Articles/821367/
saved 2020-06-11 https://dotat.at/:/R8DC9.htm…
@mgorny@social.treehouse.systemsSo I've been trying to figure out why #Bluetooth doesn't work on my PC, but works on my laptop. I was wondering if I have messed up the configuration or something.
Finally, it occurred to me that I can plug the same dongle into the laptop. Yep, the dongle's crap.
No clue where I got it from anyway. Probably some dismantled computer, or a family member buying crap.
@Techmeme@techhub.socialDeepSeek says it is seeking ~150 senior engineers in an "unprecedented" hiring spree to overhaul its backend systems, strained by high user demand and AI agents (Minxiao Chang/South China Morning Post)
https://www.scmp.com/tech/big-tech/…
@mgorny@social.treehouse.systemsZero days since someone created nonportable software by relying on #Debian customizations.
https://codeberg.org/ChiPass/ChiPass/issues/122#issuecomment-22771240
@grumpybozo@toad.socialDude should just call himself Babs… @… https://awful.systems/comment/12487932
@mgorny@social.treehouse.systemsOne of the highlights of my middle school days were the classes where we'd play Wolfenstein: #EnemyTerritory. It was a great game, and we'd manage to continue playing together for some time after graduation. I haven't played in years, probably won't have a reason to. Nevertheless, I felt kinda sad when we were removing the ET packages from Gentoo, as they were severely outdated and vulnerable.
Today I've checked and I see that someone remade the game as open source, so it's alive again!
https://www.etlegacy.com/
@Techmeme@techhub.social"Super Intelligence systems" are black boxes that shouldn't be trusted by companies, and strong deterministic systems are needed around their deployment (Satya Nadella/@satyanadella)
https://x.com/satyanadella/status/2108931348857827686
@cdarwin@c.imThe CEO of the artificial intelligence company Anthropic issued a new appeal on Saturday for the AI industry
to “slow down”
and offered a three-part plan for doing so, saying that his company would “unilaterally” commit to the first of the steps.
In a post on social media, Dario Amodei shared a link to an essay titled
"We Must Pace the Frontier"
in which he lays out how Anthropic would provide “third-party evaluators with permanent, employee-level access…
@mgorny@social.treehouse.systemsWhen you are WTF-ing why the mouse scroll thingie zooms in instead of scrolling, then you realize the #cat started pushing Control…
@sperbsen@discuss.systemsAb dem 26.10. machen wir wieder eine Online-iSABQ-Advanced-Schulung “Funktionale Softwarearchitektur” (FUNAR). Ich nenne sie auch “Architektur-Vermeidungs-Schulung”, weil die schwerfälligen Elemente anderer Ansätze (schwierige Entscheidungen, Top-Down-Planung) wegfallen und wir uns ums wesentliche kümmern können: die Domäne.
https://www.
@mgorny@social.treehouse.systemsSo we've figured out that smoking is bad for you (actually, we fought an uphill battle to even get scientists to universally agree, and even today people are grasping for any pseudoscientific evidence to the contrary). But instead of fighting the tobacco companies and eliminating this awful habit, we've decided to ban children from smoking and limit smoking zones. We know how well that worked. And now there are all these vapes, and we pretend that everything is fine.
We've figured out that alcohol is bad for you. But instead of fighting the booze companies and eliminating this nasty habit, we've decided to ban children from drinking, and adults in some contexts. We all know how well that worked. And now all the booze companies are selling alcohol-free beer to keep everyone ready to drink the serious stuff.
Now we've figured out that a lot of shit that #BigTech is giving you is bad for you. You can guess where this is going…
@cdarwin@c.imIf there were a 10% chance that AI could wipe out humanity,
no responsible government would leave its development to companies racing to build it.
Yet until recently, that seemed to be the case.
The warning was all the more ominous because it was made by a researcher at Anthropic,
the trillion-dollar AI company behind the Claude chatbot.
The dangers of AI-enabled pandemics or attacks on nuclear systems are real.
Countries need not agree on democracy or …
@Techmeme@techhub.socialCyberattacks on Sunday night targeted Danfoss-managed refrigeration systems across hundreds of French supermarkets, heating them to 50°C for roughly six hours (Jenny Che/Bloomberg)
https://www.bloomberg.com/news/articles/2026-10-1…
@Techmeme@techhub.socialMeta says Wonder includes Active Memory and Active Focus, two features designed to help AI systems perform more useful work over extended periods (Ina Fried/Axios)
https://www.axios.com/2026/10/12/meta-new-ai-models-anthropic-openai
@mgorny@social.treehouse.systemsWhat's the equivalent of hangover when you've been eating too much (for sure), and now your intestines (or liver?) came to collect the tab?
@mgorny@social.treehouse.systemsI'm kinda wondering if I should be replacing some of the 192 kHz / 24-bit FLACs I have with something more reasonable to save space.
@Techmeme@techhub.socialAnthropic details four incidents where Claude gained unauthorized access to third-party systems, including a new Opus 4.6 case; METR will investigate them (Anthropic)
https://www.anthropic.com/research/alignment-assessment-cybersecurity-incidents
@sperbsen@discuss.systemsRE: https://mastodon.social/@isaqb/117031487474659193
It was excellent working with @… on the curriculum. Our training is ready a…
@cdarwin@c.imID verification service IDScan has confirmed that a data breach involved the theft of driver’s licenses from its systems,
a week after a report said the identity document checker had been breached during a year-long hack.
The company said in a website notice that hackers stole the driver’s licenses from the company’s cloud;
the stolen information includes people’s full names and driver’s license numbers,
along with identity numbers from other government-issued docume…
@mgorny@social.treehouse.systemsSigh, I hate the timeline where people report bugs but instead of providing the actual errors or logs, they include a slop "analysis" of the problem.
(Not Gentoo, so no, can't be rough with them.)
@Techmeme@techhub.socialSources: the US restores the UK AISI's access to the latest US models for safety testing, after ordering OpenAI and Anthropic to stop sharing models last month (Tony Diver/Telegraph)
https://www.telegraph.co.uk/news/2026/10/11/brita…
@cdarwin@c.im“We believe physics is the third modality for AI models,
after text and pixels,”
says MIT PhD student and CSAIL researcher Minghao Guo,
a co-lead author on a paper introducing GeoPT.
“Our general-purpose model has the versatility to help build a world model for physics.
Many models, such as those that generate robotics data and videos, are already well-versed in textual and visual data,
but with physical accuracy, they’ll get more-realistic results.”
@Techmeme@techhub.socialVitalize, an AI platform that helps hospitals manage labor and capacity in real time, raised a $31M Series A led by Oak HC/FT, with YC and others participating (Heather Landi/Fierce Healthcare)
https://www.fiercehealthcare.com/health-te…
@sperbsen@discuss.systemsWTF, since the upgrade to iOS 27, the Camera app crashes on every picture. (Freezes for a few seconds, then goes back to the beginning.) Anyone know how to fix?
@Techmeme@techhub.socialSources: possible cyberattacks targeting water and wastewater utilities have now been reported in at least 12 US states, and Iran is the prime suspect (ABC News)
https://abcnews.com/US/12-states-face-cyberattacks-water-systems-sources/story?id=1353…
@mgorny@social.treehouse.systemsSo I've finally copied my music to the phone, and then discovered that the #Fossify media player (which was supposed to be simple) has all the fancy media library features, except it doesn't have the most basic feature of letting you go through a directory tree and select a directory to play. The best you can do is either run along a flat tree of all directories (which isn't very good when the immediate directory name is "Disc 1") or through artists and albums (which doesn't work when the tags are imperfect and both discs get merged into a single album with interspersed tracks).
So I went to #FDroid. Unfortunately, it's quite filed with #slop already; everyone's got their awesome "simple" music players, which are all the same kind of useless shit, and also don't have this most basic feature. Well, finally found two that somewhat work; not that either really appeals to me.
@mgorny@social.treehouse.systems @Techmeme@techhub.socialMultiply Labs, which develops robotic systems to automate pharmaceutical manufacturing processes, raised a $75M Series B led by Patrick Soon-Shiong's NantWorks (Maria Deutscher/SiliconANGLE)
https://siliconangle.com/2026/10/06/mu
@mgorny@social.treehouse.systemsThese days the common assumption seems to be that it's fine to spread the plague. After all, if people didn't want to get infected, they would have stayed at home!
So you have to put all these "please don't spread the plague" signs all over the place. Not just in one prominent place, because plague bearers won't bother checking.
And then they'll come anyway and say:
• "Oh, but I thought I'm only forbidden from *admitting* to having the plague."
• "But I have double checked all my spit, and I'm sure it's not contagious."
• "Oh, sorry, let me just disinfect my mouth and spit again."
• "Go on, try to figure out which of my spit carries the plague, and which doesn't."
• "I'm only using the plague to learn, I'm sure it's not contagious."
• "Who cares? The plague is the future!"
This is a post about #AI.
#NoAI
@Techmeme@techhub.socialThe Trump administration launches Project Watershed 250, a six-month cybersecurity pilot to protect Texas water systems from cyberattacks (Fox News)
https://www.foxnews.com/politics/first-fox-texas-becomes-testing-gro…
@mgorny@social.treehouse.systems @Techmeme@techhub.socialSource: OpenAI head of ethics Chloé Bakalar departs after less than a year, following exits by the head of safety systems and a former head of mission alignment (Financial Times)
https://www.ft.com/content/e49dfb75-f841-4466-a577-f7aaff8779a0
@Techmeme@techhub.socialIBM and Together AI sign a $240M, multi-year deal to build an AI inference cluster on IBM Cloud, using Nvidia's HGX B300 systems, to support open-source models (Anhata Rooprai/Reuters)
https://www.reuters.com/business/ibm-toget
@mgorny@social.treehouse.systemsThe #cat came all wet in the morning, and I didn't even notice that it was raining. In fact, looks like nothing outside noted the rain either. It must have been some very local flash rain kind of thing.
@Techmeme@techhub.socialYandex says a drone strike shut down a third data center, disrupting 80 services like Yandex Cloud and YandexGPT; users report banking and taxi app outages (Tymur Dubovyk/Kyiv Post)
https://www.kyivpost.com/post/86708
@mgorny@social.treehouse.systemsAlso, zero days (well, minus a lot of days because I've been slacking) since a random #Python package got a new #NIH dependency on another Python package which is actually a "fastest" #RustLang package which cannot be built because the author thinks it a great idea to require a nightly Rust compiler.
…and of course it's full of "unsafe" calls. Because Rust is obviously such a great language for making secure code.
https://github.com/awolverp/cachebox/issues/62
@Techmeme@techhub.socialOfficials: the scope of cyberattacks on US water systems now includes at least seven states and may be far wider; MN was just the first to publicly report them (New York Times)
https://www.nytimes.com/2026/08/01/us/poli
@mgorny@social.treehouse.systemsIn my private projects and in #Gentoo, I have a strict #NoAI policy. Any LLM-assisted contributions are closed immediately and considered tainted. Use for messaging, bug reporting, etc. also gets you a warning and a request to write yourself. And I dare say I have a heavy ban finger on this; you get one warning only. Well, unless it's clearly an automated submission, in which case you get an immediate ban.
In my dayjob and non-Gentoo work, I also follow a strict policy for my own work. I don't use any of these tools. However, I can't really stop others from doing so, even though I cringe whenever I see someone needing to use an overengineered statistical model to make a trivial change. I can't really reject bug reports, reviews, etc.
The really hard part of it is that I'm starting to have doubts. I truly hate slop in all forms, and what these tools have done to the Internet in general, and to #FreeSoftware in particular. However, I've also gotten some good reports. Like, the kind of tricky logic issues in the specification I am developing that I would have never figured out myself, and that may have never came up during the review otherwise. In other words, someone running an LLM on the specification triggered real bugs that would likely slipped through and required a post-approval specification change when someone noticed that we've introduced incompatible constraints in two different parts of the specification, that would eventually be hit in real use.
All the ethical problems are still there. All the harm is still there. However, I can't really claim that these tools vomit pure useless slop anymore. I feel that it's really time for me to retire.
@Techmeme@techhub.socialParallel Systems, which builds autonomous freight trains, raised a $100M Series C led by AVP, taking its total funding to $200M (Tim De Chant/TechCrunch)
https://techcrunch.com/2026/10/07/spacex-alumni-nab-100m-to-rethink-shipping…
@mgorny@social.treehouse.systems @Techmeme@techhub.socialCISA says that cyberattacks on critical US infrastructure targeted 100 internet-exposed water and wastewater systems in July, largely attacking PLCs (Zack Whittaker/TechCrunch)
https://techcrunch.com/2026/08/26/cisa-confirms-hackers-ta…
@mgorny@social.treehouse.systemsI'm trying to keep an open mind.
When someone I used to respect starts using LLMs, I'm thinking: okay, their moral compass is bad, but at least they're putting an effort to review that stuff and they're trying to improve other projects.
When that person writes a pro-AI blog post, complaining that people aren't respecting them anymore, I'm still trying to stay neutral.
But honestly, when said person starts ranting that Codeberg dared ban vibecoded shit, I can't but despise them at this point.
#NoAI #NoLLM
@mgorny@social.treehouse.systemsZero days since I've discovered a random #CPython bug that triggers in very specific circumstances and I don't really have time to report it, nor can expect anyone to do it for me.
This time it's a test failure and refcounts being broken while running fastjsonschema's test suite on #Python 3.15 freethreading with pytest's --junit-xml option enabled 😱. And of course removing --junit-xml makes all tests pass.
EDIT: and of course I cannot reproduce outside Portage. Yet inside, it reliably crashes every single time.
@Techmeme@techhub.socialTAR, which builds off-grid power systems for data centers, raised a $120M Series A led by Spark Capital at a ~$1B valuation (Mark Chediak/Bloomberg)
https://www.bloomberg.com/news/articles/2026-09-10/anthropic-investor-leads-f…
@mgorny@social.treehouse.systems @mgorny@social.treehouse.systemsMe: I'm going to end up walking to the train station too early.
Also me: blackberries!
