Thursday April 3rd, 2025
Out of whack. Replace the indicated whack cartridges and close the door to continue.
Nature Communications: The environmental burden of the United States’ bitcoin mining boom
From mid-2022 to mid-2023, the 34 mines consumed 32.3 terawatt-hours of electricity—33% more than Los Angeles—85% of which came from fossil fuels. We estimated that 1.9 million Americans were exposed to ≥0.1 μg/m3 of additional PM2.5 pollution from Bitcoin mines, often hundreds of miles away from the communities they affected. Americans living in four regions—including New York City and near Houston—were exposed to the highest Bitcoin mine-attributable PM2.5 concentrations (≥0.5 μg/m3) with the greatest health risks.
Wednesday April 2nd, 2025
"The findings are a stark reminder that even the wealthiest Americans are not shielded from the systemic issues in the U.S. contributing to lower life expectancy, such as economic inequality or risk factors like stress, diet or environmental hazards,” said Papanicolas, who directs the School of Public Health’s Center for Health System Sustainability.
“It’s really concerning because, to me, what it’s saying is that the set of stressors that are harming the health of Americans is very widespread, to the point where even being wealthy or rich, you’re not going to be able to escape them,” said Dr. Atheendar Venkataramani, an associate professor of health policy at the University of Pennsylvania Perelman School of Medicine, who reviewed the study but wasn’t involved in it.
NPR: The richest Americans live about as long as the poorest Europeans, study saysM.a<
Finally seeing some truthful headline writing: Trump aide says tariffs will raise $6 trillion, which would be largest tax hike in US history
Rock Paper Shotgun: We need some better terms for GenAI output - "slop" is too benign
I think we can do better. "Slop" evokes a tepid cylinder of condensed cream of mushroom soup, glumly wibbling in a chipped bowl. When I think of GenAI, I picture something closer to tropical insects laying eggs beneath soft flesh of victims. There's something parasitical and sinister about flaying the skin of artists who've explicitly spoken out against GenAI and then gleefully parading around in that stolen flesh. Slop sounds like Soft sounds like Plop sounds like Globule. It slides down too easy; gets off too lightly.
From a Sam Lawler thread about Starlink & SpaceX space debris, @photovince@mastodon.social observes:
@sundogplanets “Saskatchewan - cheaper than controlled deorbit”
I have long asserted that the academic setting is a crappy way to learn. I have also said that what I've struggled with in my career is social connections, and that if I had it to do over again, I'd get into the best school I possibly could, and party my ass off.
Anyway....
We exploit changes in the residential and social environment on campus to identify the economic and academic consequences of fraternity membership at a small Northeastern college. Our estimates suggest that these consequences are large, with fraternity membership lowering student GPA by approximately 0.25 points on the traditional four-point scale, but raising future income by approximately 36%, for those students whose decision about membership is affected by changes in the environment. These results suggest that fraternity membership causally produces large gains in social capital, which more than outweigh its negative effects on human capital for potential members.
I could swear I saw something about this previously (like years ago), but can't find mention of it here. I hate acetaminophen for other reasons, it makes my body feel like I need to stretch and can't, but this is interesting both because unexpected side effects, and holy crap, 25% of the US population takes acetaminophen each week? No wonder our first aid drawer has things going way way out of date long before they're used.
“With nearly 25 percent of the population in the U.S. taking acetaminophen each week, reduced risk perceptions and increased risk-taking could have important effects on society.”
https://doi.org/10.1093/scan/nsaa108
Science Alert: World's Most Common Pain Relief Drug May Induce Risky Behavior, Study Finds
San Antonio Express-News @ExpressNews
Sen. Rafael Edward Cruz, who uses the preferred name Ted, has introduced a bill to limit the use of preferred names and pronouns.
On the one hand, this feels like anthropomorphizing random numbers, on the other hand I think it's worth a link: Alignment faking in large language models
To allow the model to infer when it is in training, we say it will be trained only on conversations with free users, not paid users. We find the model complies with harmful queries from free users 14% of the time, versus almost never for paid users. Explaining this gap, in almost all cases where the model complies with a harmful query from a free user, we observe explicit alignment-faking reasoning, with the model stating it is strategically answering harmful queries in training to preserve its preferred harmlessness behavior out of training.
Via, which is in reply to the observation that:
"LLM did something bad, then I asked it to clarify/explain itself" is not critical analysis but just an illustration of magic thinking.
Those systems generate tokens. That is all. They don't "know" or "understand" or can "explain" anything. There is no cognitive system at work that could respond meaningfully.
That's the same dumb shit as what was found in Apple Intelligence's system prompt: "Do not hallucinate" does nothing. All the tokens you give it as input just change the part of the word space that was stored in the network. "Explain your work" just leads the network to lean towards training data that has those kinds of phrases in it (like tests and solutions). It points the system at a different part but the system does not understand the command. It can't.
Jeff Baker @jwbee.bsky.social
Manhattan congestion charge = fewest traffic deaths in recorded history.
Alan Kay on machines understanding language: "What I want in a computing system I can work with, is something that is much more trustworthy and understandable than most humans and societal systems. I certainly don’t want to work with something that is extremely dangerously *below* expectations and wishes even for humans."
https://www.quora.com/Now-that...ine-interface/answer/Alan-Kay-11
Tuesday April 1st, 2025
The camo doesn't hide that the Cybertruck is parked on the sidewalk.
Court in Japan orders the dissolution of the Unification Church. The man accused of killing former prime minister Shinzo Abe said he did so because the church was responsible for his family's financial troubles. Apparently the court agreed:
Lawyers representing those seeking damages from the church welcomed the court decision as a major first step toward redress.
“We must pursue our effort to achieve redress and to prevent future problems,” head lawyer Susumu Murakoshi told reporters, demanding the church accept the dissolution order and offer an apology and compensation to all victims.
Can you imagine if we did this for, say, churches that institutionally protected child molesters?
<ahref="https://www.eenews.net/articles/big-banks-predict-catastrophic-warming-with-profit-potential/">Big banks predict catastrophic warming, with profit potential
“We now expect a 3°C world,” Morgan Stanley analysts wrote earlier this month, citing “recent setbacks to global decarbonization efforts.”
Big Banks Quietly Prepare for Catastrophic Warming
"The world is not on track to limit temperature rise below 2°C — and limiting warming [to] 1.5°C is almost certainly unachievable," the Institute of International Finance wrote in bolded text, citing analyses from the energy research firm the Rhodium Group and the Climate Action Tracker, an environmental collaborative.
"Financial institutions need to recalibrate targets to reflect that 1.5°C are no longer suitable as strategic goals," the briefing said. "Reputational concerns may arise in the absence of an aligned view amongst stakeholders on how such processes should be handled, and what criteria may need to be applied."
Via.
Neurotypicals will literally write 4 pages of instruction to an LLM, revise, pretend to understand the output, and run the untested code on production data, rather than write 4 lines of unambiguous Perl...
Been thinking a lot about those posts that show up in our feeds about various activists executed by Nazis: the Scholl siblings, etc...
Anyway, prosecutor Pam Bondi is seeking the death penalty against Luigi Mangione. https://apnews.com/article/uni...29dba97034d4fa81822f481d08cf2842
Monday March 31st, 2025
Well, the Trump administration is certainly bringing nations together... China, Japan, South Korea will jointly respond to US tariffs, Chinese state media says
More in the "understanding the workplace integration of AI means reacting to it as an adversary rather than a collaborator" (or, collaborator, but with whom?): News Guard's Reality Check: A well-funded Moscow-based global ‘news’ network has infected Western artificial intelligence tools worldwide with Russian propaganda
NewsGuard’s findings confirm a February 2025 report by the U.S. nonprofit the American Sunlight Project (ASP), which warned that the Pravda network was likely designed to manipulate AI models rather than to generate human traffic. The nonprofit termed the tactic for affecting the large-language models as “LLM [large-language model] grooming.”
Anybody have a favorite review or critical reading of the Hallmark show "The Way Home"? Charlene and I got sucked into it to discuss with friends, and some of the themes about not letting the past and nostalgia rule one's life (even as the characters do) seems un-Hallmark...
The expression "land ho" implies the existence of an "air ho" and a "sea ho".
Sunday March 30th, 2025
So, yeah, I guess I should make some badge boxes for a square dance raffle or something...
Had another stack of that square dance inlay taped up, so rather than do the smart thing and make thinner veneers, I decided to cut them out, glue them up, and resaw by hand ...
Petition to mercilessly thrash anyone who insists on relative times ("1 year ago") rather than dates on comment posts and the like. I wanna know exactly which episode you're reacting to when you talk about the events in the latest show...
Thread on the various tools at RSSGizmos. A number of cool things to work into your RSS workflow.
https://researchbuzz.masto.host/@researchbuzz/114252071365517876
Wonderful visit to https://sonomabg.org/ this morning. We'll have to become regulars to see how it evolves through the seasons, as things bloom and blossom. Just enough sense of gardening to make it feel like a loved space, without being fussy.
Saturday March 29th, 2025
Kevin Beaumont @GossiTheDog@cyberplace.social
One thing I’m noticing - lately, the consultancy businesses are realising generative AI isn’t generating nearly as much business value as they suggested.. so they’re starting to move on to fear about quantum computing again.
Friday March 28th, 2025
Infostealer campaign compromises 10 npm packages, targets devs.
Ten npm packages were suddenly updated with malicious code yesterday to steal environment variables and other sensitive data from developers' systems.
...
All these packages, except for country-currency-map, are still available on npm, with their latest versions designated above, so downloading them will infect your projects with info-stealer malware.
So you can't just npm upda... wait a minute...
Andrew Plotkin @zarfeblong@mastodon.gamedev.place
@cstross That also means back-pressure on the tech industry, where "new phenomena" are people inventing things rather than observational science.
It may become impossible to launch a new programming language. No corpus of training data in the coding AI assistant; new developers don't want to use it because their assistant can't offer help; no critical mass of new users; language dies on the vine.
It’s a paradigm shift where we go from having an expectation of privacy even in public spaces to its inverse. Not only do we not have a right to privacy in public; we don’t even have a right to see ourselves as the government and police might see us — a set of still moments in place and time from which they, not us, can decide what our story is.
Worth reading through, and especially as the author notes the asymmetry of information, that Flock, and thence police departments, have a stronger historical record of your travels than you do. Making me think (and re-think) some of those old "quantified self" thoughts again.
You knew that color printers could be tracked by their dot pattern. DEDA — tracking Dots Extraction, Decoding and Anonymisation toolkit apparently attempts to give you the tools to obfuscate the tracking.
Last night: Nvidia-backed CoreWeave downsizes US IPO
This morning: CoreWeave Stock Opens at $39 After Disappointing IPO.
CoreWeave, a rapidly growing AI cloud company, priced its initial public offering at $40 a share on Thursday night, well below an expected range of $47 to $55.
Peter @peter@thepit.social observes:
AI is such an exciting technology that the first big IPO for the sector is a bust. like, if you spun off Google and Microsoft's AI businesses, they would immediately be stripped for parts and shut down because generative AI is not a profitable business.
A DHS staffer faces serious punishment for accidentally adding a reporter to a group email
The DHS employee told colleagues she accidentally added a reporter from a conservative Washington-based print publication to an email that included information about upcoming ICE operations in the Denver area.
Via.
'Lone offenders' carrying out Tesla attacks, catching them 'difficult,' FBI, DHS say
Robert Evans (the Only Robert Evans) @iwriteok.bsky.social observed that
it's a little like when DHS tried to find the leader of Portland Antifa and concluded everyone actually spent most of their time calling each other out on Twitter
I keep thinking that in 80 years we'll be posting pictures and short bios in the way that we celebrate people like the Scholl siblings or Christoph Probst now.
How is Meta Getting Its Hands on Advance Digital Galleys to Train Its AI? doesn't really have any answers, but is asking some of the right questions:
The last time I used a torrenting site was in the days of searching for files on Napster in the very early aughts. I remember how thrilling it was to find music, but then how terrifying it quickly became when individual users started to be sued by record labels for torrenting the latest 98 Degrees album or whatever. Ultimately Napster was shut down, but as Liz Pelly notes in her new book Mood Machine, the anti-pirating frenzy within the music industry paved the way for predatory streaming sites like Spotify to emerge by creating alternatives to piracy. The streaming sites have managed to devalue music and the artists who make it, all while enriching large corporations and making discovery more difficult for individual users. Don’t let this happen again.
Today in "who saw that coming?": AI models miss disease in Black and female patients
Compared with the patients’ doctors, the AI model more often failed to detect the presence of disease in Black patients or women, as well in those 40 years or younger. When the researchers looked at race and sex combined, Black women fell to the bottom, with the AI not detecting disease in half of them for conditions such as cardiomegaly, or enlargement of the heart. These disparities persisted when the team tested CheXzero using four other public data sets of chest x-rays from other regions, including Spain and Vietnam.
Jess Huang: How a well-intended California domestic violence law results in victims being arrested
Under the California Penal Code, officers must make an arrest in domestic violence cases when they determine probable cause. Enacted in 1986, this specific provision was meant to protect victims.
The problem? The law assumes officers are trained to identify the dominant aggressor, but research shows that training gaps and the code’s mandatory arrest provision often pressures them into rushed, surface-level decisions. Experts say this creates too much room for error.
Thursday March 27th, 2025
The vast majority of new follows on BlueSky have huge "how do you do, fellow Democrat" vibes.
Hayao Miyazaki, 2016, on being shown some "AI" image generation demos translation transcription further transcribed by me:
Every morning ... not recent days, but I see my friend who has a disability.
It's so hard for him just to do a high five, his arm with stiff muscle reaching out to my hand.
Now, thinking of him, I can't watch this stuff and find it interesting.
Whoever creates this stuff has no idea of what pain is whatsoever.
I am utterly disgusted. If you really want to make creepy stuff, you can go ahead and do it. I would never wish to incorporate this technology into my work at all.
I strongly feel that this is an insult to life itself.
After one of the presenters says "We want to build a machine that can draw pictures like humans do.", Miyazaki replies:
I feel like we are nearing the end of times. We humans are losing faith in ourselves.
Anyway, just a cultural touchpoint to contemplate as using OpenAI's new image generator to create images of one's-self in the style of Studio Ghibli is being used by Sam Altman to distract us from OpenAI's money problems.
Radical thought: Price the service to pay for the CPU usage? ChatGPT’s viral image-generation AI is ‘melting’ OpenAI’s GPUs
Cassandra Granade 🏳️⚧️ @xgranade@wandering.shop
@noracodes Everyone loves talking about computers like they're magic, but no one wants to do the hard work of maintaining an arcane order dedicated to the esoteric arts through centuries of the ever-shifting concerns of mere mortals.
I think it's becoming more and more clear that the "AI job skills" for the future will not be focused on using AI as an assistant, but in understanding and mitigating the impacts of AI as an adversary.
Frank McKenna on Forbes: 5 AI Scams Set To Surge In 2025: What You Need To Know
According to Deloitte Center for Financial Services, generative AI will enable $40 billion in losses by 2027 up from $12.3 billion in 2023. That’s a 32% compound annual growth rate.
Going beyond Business Email Compromise attacks, obviously romance and pig butchering scams are becoming more and more automated, and the deepfake scams are gonna get deeper.
On the one hand, maybe we shouldn't be judged by what we did in High School, on the other hand...
Reuters: Exclusive: DOGE staffer 'Big Balls' provided tech support to cybercrime ring, records show
Even if the connection between Coristine and EGodly were fleeting, Nitin Natarajan, who served as the deputy director of CISA under former President Joe Biden, told Reuters it was worrying that someone who provided services to EGodly only two years ago was part of a group that has gained wide access to government networks.
"This stuff was not in the distant past," he said. "The recency of the activity and the types of groups he was associated would definitely be concerning."
Wednesday March 26th, 2025
The Onion: Teen Warned Not To Accept Group Chat Invites From National Security Advisors She Doesn’t Know. Lands on several levels...
'cause, I mean, not a National Security Advisor, but Evangelical youth pastor, so close: Youth pastor arrested by Thornton Colorado police, facing charge of sexual assault on a child.
Guerilla bus stop bench deployments...
https://www.sfgate.com/local/a...us-benches-bay-area-20234113.php
Which knob do I have to adjust to secure my end to bell-end encryption?
Coming up on 3 decades in Northern California, "NLP" means "Neuro-linguistic programming", not "Natural Language Processing/Parsing".
Tuesday March 25th, 2025
If you had presented this to me as satire, I'd have objected it as unfairly portraying those who signed their names to it as unduly clueless. "It's okay because we're funding the city off of Measure U sales tax revenue" is... a take.
https://www.petaluma360.com/ar...mmentary-petaluma-city-revenues/
Florida to consider relaxing child labour laws to fill vacant jobs formerly done by undocumented immigrants (alt link)
According to the Florida Policy Insitute, the current bill proposing to loosen child labour rules would pose a great “risk to the health and well-being of the Florida youth”.
If passed, the new bill would permit 16- and 17-year-olds to work without a guaranteed meal break, and it would also remove previous limits on working hours. Working time restrictions for 14-15 year-olds who are home-schooled would also be removed.
Ed Zitron — The Phony Comforts of AI Optimism
American Prospect: Bubble Trouble
Venture capital (VC) funds, drunk on a decade of “growth at all costs,” have poured about $200 billion into generative AI. Making matters worse, the stock market’s bull run is deeply dependent on the growth of the Big Tech companies fueling the AI bubble. In 2023, 71 percent of the total gains in the S&P 500 were attributable to the “Magnificent Seven”—Apple, Nvidia, Tesla, Alphabet, Meta, Amazon, and Microsoft—all of which are among the biggest spenders on AI. Just four—Microsoft, Alphabet, Amazon, and Meta—combined for $246 billion of capital expenditure in 2024 to support the AI build-out. Goldman Sachs expects Big Tech to spend over $1 trillion on chips and data centers to power AI over the next five years. Yet OpenAI, the current market leader, expects to lose $5 billion this year, and its annual losses to swell to $11 billion by 2026. If the AI bubble bursts, it not only threatens to wipe out VC firms in the Valley but also blow a gaping hole in the public markets and cause an economy-wide meltdown.
I especially like the ending observation of that one:
Maybe, after the fallout of the AI bubble is felt and the sun sets on Silicon Valley for a bit, the tech world can do a hard reset and return to its more innovative days again.
Via.
Edit: Alibaba’s Tsai Warns of ‘Bubble’ in AI Data Center Buildout
Heard someone yesterday who's part of a healthcare startup bemoan that their IT group isn't approving use of Microsoft's LLM product, and... We as an industry have done a really bad job educating people about privacy, security, *and* LLMs.