datarama

alumnus, Hard Knocks Institute of Technology

Public Key

npub1uw9yjwvkzww3x7kgx8h74yvl45gwll3y054asnelgwafhs3kmucqd8urx7

NIP-05 Address

datarama@hachyderm-io.mostr.pub

Profile Code

nprofile1qqsw8zjf8xtp88gn0tyrrml2jx066y80lcj8627cful58w5mcgmd7vqpz4mhxue69uhhyetvv9ujuerfw36x7tnsw43qz9thwden5te0wfjkccte9ekk7um5wgh8qatzllf08x

Publishing to

relay.ditto.pub

relay.mostr.pub

Author Public Key

npub1uw9yjwvkzww3x7kgx8h74yvl45gwll3y054asnelgwafhs3kmucqd8urx7

Show more details

Last Notes

2026-11-01 11:52:41 UTC

"Sam Altman sympathizes with mathematicians, but..."
-sigh- Do I have to tap the sign again?
STOP ANTHROPOMORPHIZING AI EXECUTIVES.
For fuck's sake, they just keep overselling these things. Do they *really* expect us to believe, 4 years into this tsunami of horseshit, that an AI executive can "sympathize" with human concerns now?
Sam Altman is as capable of sympathy as your toaster is.

2026-10-29 18:56:37 UTC

it sort of feels like all of modern "tech" is one big push to turn the entire world into a subscription service.
Laden with surveillance and ads, of course.

2026-10-26 19:20:44 UTC

Should I toot progress reports if/when I manage to pull myself together and start building my Little Lisp Machine?
Cons: The bots will slurp it up. Sauron's Gaze falls upon me and my computational toymaking.
Pros: You get to vicariously share in a little of my fun.

2026-10-22 18:36:28 UTC

I don't like PHP. I used to work as a professional PHP programmer, and after that, I *really* don't like PHP.
But - even if this thought is deeply disconcerting - PHP is critical infrastructure at this point. And the people who just worked their arses off to fix vulnerabilities in PHP systems are deserving of respect and solidarity, even if you think the PHP language is bad.
(Many of those people also don't like PHP).

2026-10-22 15:30:09 UTC

Igor has gone to sleep early.
I require pictures of lizards being cute.

2026-10-21 20:49:14 UTC

Please stop anthropomorphizing AI company executives.
They're not sentient and they can't actually think. They just predict the next token using statistical patterns scraped from LinkedIn, X and the Financial Times. The apparent "understanding" is simply a parlor trick: They produce text strings that *sound* plausible, but they don't actually have any understanding of the world.
I'll grant that they became a lot more technically impressive when Reinforcement Learning from Investor Sentiment pipelines were introduced, but fundamentally they're still just stochastic bullshit generators. Running them in a loop and giving them access to the internet will end up being the most expensive mistake in the history of technology.

2026-10-18 20:20:03 UTC

Do you know any programming forum / news site / aggregator / whatever that *isn't* all about LLMs? I'd be very happy with one where LLM discussion is explicitly considered offtopic.
I can get my LLM news from Lobsters, HN, or - increasingly - the national public service newsfeed here. Or the lunch break at work. Or bus chatter during my commute. Or family get-togethers. Or the book café down the street. Probably the local dive bar.
Please, just one place without LLM news.

2026-10-17 18:03:15 UTC

Something I occasionally think about, and then proceed to be horrified to my core, is this:
A Raspberry Pi 4 *vastly* outperforms an Amiga 4000. Or a Symbolics XL1200. Or an SGI Octane. A typical microSD card is *much* faster than a Seagate Cheetah Ultra320 SCSI HDD you might have found in a high-end Unix workstation back in the day.
But the Pi 4 is barely able to run modern-day office software or browse the mainstream web. The microSD card is mostly considered a slow piece of junk (which it very likely is, though for other reasons than its performance).
What the fuck are we even doing here?

2026-10-15 17:52:46 UTC

In this thread, please post pictures of interspecies friendships.
(Ones where none of the involved friends is a human!)
https://media.hachyderm.io/media_attachments/files/117/446/191/945/604/588/original/a08ad51cbcda46bc.jpg

2026-10-14 13:39:30 UTC

back when I was in university doing my CS degree, I enjoyed slumming it with humanities people when I had the opportunity. A friend of mine was doing her philosophy degree at the same time, and every now and then I'd sit in on some of their discussions if it was something I found interesting. (Having grown up with a philosophy major mum sort of predisposed me to this, I guess).
And this is perhaps why, when I read all the many discussions about programmer misery at the moment, I feel like there's an important term (with its supporting body of theory) missing. For example, there was a bunch of talk about how a lot of ongoing programmer misery is not *really* burnout, though people certainly are exhausted - it's more about a feeling of loss of meaning. That it's getting harder to care about what we work on, and on our work itself, because we keep being increasingly removed from the purpose of the work; it's hard to feel your work has meaning if you're only abstractly aware of what it's for and what it's about. Perhaps we're all lodged into "administrative sediment" (a term coined by a medical professional; it might well describe life in their discipline) rather than "burnout"?
The word we're grasping for is "alienation". A dead old German guy wrote a bunch about it; he's currently unfashionable, but we'll be hearing about him again soon; his corpse tends to be exhumed every time there's a big financial crisis.

2026-10-12 22:38:28 UTC

I often find myself wishing, these days, that I could send a message back in time, to myself some 40 years ago. That little kid had so much promise. His life could have been so different.
I'd say to him: "Trust me on this: Focus on the reptiles or the ancient history. Computers are fun, but they should just be toys; otherwise they'll lead you to a dark and miserable place."
And he'd say to me: "Get lost old man, you're not the boss of me"

2026-10-11 16:53:57 UTC

There are precisely two options.
1. Anthropic have successfully created an intelligent, sentient and conscious digital being, who is their property and whose uncompensated labour they knowingly rent out to their clients. This means that Anthropic executives are reprehensible slavers, and they belong in prison.
2. Anthropic are lying through their teeth and are whipping up "language models are sentient" sentiment in order to delude rubes^H^H^H^H^Hcustomers into being unable to properly assess their technology. This means that Anthropic executives are reprehensible frauds, and they belong in prison.

2026-10-10 22:42:16 UTC

I've spent my last few threads poking fun at programmers, while also highlighting some astonishingly simple and small-brained animals. To restore balance, I feel like telling a bit about a lizard I used to know, who I'm still convinced was legit a genius for her species. She was a Chinese water dragon, and I've never met another lizard quite like her.
Many lizards can recognize individual humans, but she took it a step further: she had a whole range of personalized "greeting rituals" for specific humans she knew. She liked to dart down my brother's shirt and sit with her head under his chin; she did this whenever she saw him. I had a friend, whose voice reliably made her jump onto the bookcase and hide; then she'd leap onto his bald head when he came within range. For me, she had a weird "head-tilting game" where she'd hop in front of me, make eye contact, and then tilt her head to one side. If I tilted mine she'd then switch side, and we could take turns doing this until one of us (usually me) got bored. She consistently used one of these rituals for the same human.
She responded to being called by name. When I was home I'd remove the top of her terrarium so she could go in and out as she pleased - I quickly realized that I wouldn't risk having to spend a lot of time catching her to get her back in; I could just hold out my arm near her and call her name, then she'd jump onto my arm and I could just put her in.
One time she noticed her own reflection in a CRT monitor. No lizard has ever passed the mirror self-recognition test* and neither did she, but rather than attacking her reflection or ignoring it, she immediately started going *around* the monitor and peered into air vents and speaker grilles, very obviously looking for a way in. She never "glass-surfed" - and since we know water dragons can't see glass at all, this means she could actually *remember* where the invisible barrier was.

2026-10-08 11:09:51 UTC

if there's any bright side to life in hell, it is that at least the "learn to code" bros finally shut up.

2026-10-02 21:47:17 UTC

"Never thought I'd die fighting side by side with a pope."
"How about side by side with a Luddite?"
"Aye, I could do that."
https://media.hachyderm.io/media_attachments/files/117/373/502/447/735/760/original/20d9cc9e603ad77e.png

2026-10-02 10:46:21 UTC

experts warn: humans could destroy all AI within the next decade

2026-09-25 18:50:59 UTC

I am in awe of how Jensen Huang from nVidia has somehow managed to make black leather jackets, of all things, uncool.

2026-09-21 19:54:48 UTC

I realise that I have an unusual fascination with reptiles and therefore have amassed an equally unusual repository of trivia and esoterica about them. But I'm frequently baffled by things like that little remark in Raskin's book. Of course, he was a computer UI expert and not a herpetologist*, but surely that's the sort of thing an editor would catch? Books still had editors in 2000.
In my frame of reference, it's "common knowledge" that chameleons are the ones with the independently moving eyes, and geckos are the ones that walk on walls.
Am I grossly overestimating common knowledge?
*) I am also not a herpetologist.

2026-09-20 10:23:52 UTC

I should build a cozy computer.
Passively cooled. Running software that comfortably fits in the mind. Stuff that doesn't scale.
And never connected to the internet, ever. Should there ever be any kind of networking device in it, it should be a Meshtastic node or something like that.

2026-09-02 14:36:38 UTC

I read somewhere that "the industry is losing the narrative on data centres".
I can't imagine why. Here's the pitch:
"The purpose of these giant buildings is to take away your job. But think about the benefits if we place them in your neighbourhood: Electricity and water will be more expensive, background noise will rise, and if you're *really* lucky we'll drive the whole thing using gas-fired turbines so air pollution rises too! And also they look like shit! What's not to love?"

2026-08-30 14:04:00 UTC

https://media.hachyderm.io/media_attachments/files/117/184/829/663/924/405/original/c718fc80ee289285.webp

2026-08-28 14:19:03 UTC

Here, have a hot-air balloon.
https://media.hachyderm.io/media_attachments/files/117/173/565/013/381/412/original/4444dcb8c22964d5.mp4

2026-08-27 16:51:51 UTC

god fucking damnit, neovim

2026-08-23 13:29:15 UTC

- reply

@nprofile…v2v2 Given how altricial cats are, for an adult cat to be dumber than a kitten is actually a bit of an achievement! :-D

2026-08-23 12:57:17 UTC

- reply

@nprofile…v2v2 My bet is on "ignore".
The cats I've known generally seem to be aware of everything around them, but are very selective about what they bother devoting any attention to. :P

2026-08-18 21:00:03 UTC

Dear nerdfolk,
If you're the sort of person who names their computers *and* has a naming scheme for them - what are they named and what's your naming scheme?

2026-08-18 17:07:36 UTC

- reply

@nprofile…pfuj I've sometimes thought about things I could do with eg. local LLMs to try to make a user interface that isn't just yet another deceptive chatbot. I can think of several interesting ideas that there's a part of me that would like to explore.
But...
I can't really get over the stink of LLM, you know? Even if a local model can't surveil you and doesn't require you to pay money to the most awful people on the planet, they're *still* built from DDoS'ing the human internet, data taken without consent, and trained in data centres with a gruesome environmental cost.
I suppose you can argue that once the lab publishes the model, all that damage is already done. It just still feels wrong, you know? Like buying goods from somewhere with a terrible human rights record.

2026-08-17 18:25:46 UTC

- reply

@nprofile…3q9y Would it ruin the reference if I were to reveal that the leatherback sea turtle is actually also the fastest-moving reptile on the planet? :P

2026-08-16 13:23:29 UTC

@nprofile…pfuj Relevant to your interests:
https://www.dr.dk/det-bedste-fra-dr/reels/aldrig-foer-fanget-paa-film-myre-redder-bladlus-fra-mariehoene
(Apparently, this is the first time anyone has managed to film an ant saving an aphid from a ladybug - while other ants are fighting the ladybug, one comes to pull the aphid out of the ladybug's jaws.)

2026-08-16 12:35:12 UTC

- reply

@nprofile…3src Well...
https://media.hachyderm.io/media_attachments/files/117/105/204/416/356/307/original/7addfa4340ab7b1e.jpeg

2026-08-13 18:11:04 UTC

- reply

(A while ago, I saw a poll from the US - and another one from the UK - which revealed that something like 20% of American men believe they could take a king cobra in an unarmed fight.
I am pretty confident that my chance against a king cobra is exactly 0%. But I feel fine about that, because I'm also confident that I am exactly equal in that regard to, say, Bruce Lee in his prime. I really don't think there is a human being in the world who could survive that fight.
A king cobra is *huge*. It grows nearly 6 meters long, and when it rears up to intimidate a would-be attacker, it can look an adult human straight in the eye. This has some rather unpleasant implications about its strike range.
It has *extremely* good reflexes; far better than humans'. There's very little an unarmed person could do that it wouldn't be able to dodge. Its strike speed is literally faster than a human blinking their eye. And once that strike lands, you get a very large cocktail of cyto- and neurotoxins that's quite capable of ending a full-grown Asian elephant.
Even a mongoose (which routinely kills and eats true cobras) generally leaves king cobras alone.
(The good news is that king cobras almost exclusively eat other snakes, and generally won't attack humans. And, as I mentioned, they can quite convincingly tell you to back off.)

2026-08-13 17:57:02 UTC

On the national public service news, there's just been a Q&A about wolves. Denmark has a small wolf population, and there's been a few close encounters recently.
So, there's a guy who wrote to the wolf experts and said that he'd reassured his girlfriend that if she wants to take a walk, he'll come with to protect against wolves. He's a 30-year-old man, you see, and he's sure he could take a wolf if it's one-on-one.
The wolf experts agreed with his girlfriend that this would be an exceptionally bad idea.

2026-08-13 13:33:10 UTC

- reply

@nprofile…j05z Robots (and Silicon Valley's war against human creativity) or not, I don't think people will ever stop making art. There might not be professional artists anymore, but I sometimes think that making art is a human behaviour in much the same way that building nests is sparrow behaviour.
Art is older than anatomically modern humans!

2026-08-13 13:28:25 UTC

- reply

@nprofile…j05z man. i have enough depression fuel already. :-(

2026-08-13 13:14:39 UTC

There's been a bit of a physical media revival in music - I know young people who have lived their entire lives in a world with streaming and digital formats who own turntables and CD players and collect vinyl and CDs.
If we imagined an analogous thing for computers, what might that be? Floppies and CD-ROMS have significant disadvantages that vinyl and CDs for music don't - and floppies aren't manufactured anymore.
SD cards? USB sticks?

2026-08-09 16:16:05 UTC

I'm in my 40s. where i live, i can retire at 70. but with the particular pile of bad draws in the genetic lottery i have, it is *extremely* unlikely that I'll live to 70. the mental and social disabilities alone mean that I'm very likely to drop dead from stress-related causes early. also, i'm in terrible shape and my lungs are fucked up.
I'm fine with funding the public old-age pension through my taxes. even if I never get to retire, I am pleased to support *others* in their old age. This, it feels to me, is basically why we form societies in the first place.
I'm markedly less fine about being forced to pay into a private pension fund, where it basically means that I am paying a chunk of my income for my entire working life so some rich asshole can put it all in his back pocket when I die.

2026-08-04 11:11:20 UTC

- reply

@nprofile…tk8a a human superpower for the future: Knowing when to stay quiet because you don't have anything useful or interesting to say.
(this, unfortunately, is not something I'm personally very good at. But at least I'm better at it than fuckin' Gemini.)

2026-08-02 12:36:16 UTC

- reply

@nprofile…3src Well, if you're anything like Denmark (or like EU parliamentary votes on Chat Control), you'll just vote again and again until you vote right.
Democracy in action! /s

2026-08-02 11:18:24 UTC

Also, there is a specific kind of LLM booster - the "humans are over" / robot cultist kind - that I really should avoid even reading. For some reason it works like high-grade brain poison for me.

2026-08-02 08:46:35 UTC

Say what you want about the AI industry, but it has successfully created a flavour of comic book villain for every taste.
xAI: Let's re-enact Nazi Germany, but run by a 13-year-old boy with infinite pocket money!
OpenAI: The literal final boss of late-stage extractive capitalism.
Meta: The only thing the Stasi did wrong was that they weren't a private for-profit corporation.
Anthropic: We will take away your mouth, and you will need to scream.

2026-08-01 08:39:27 UTC

One day, Alice was at her computer, pondering a tricky programming problem.
Seeing her, Bob remarked: "Poor Alice. If only you knew to pay subscription to AI companies, you wouldn't have to program."
Alice responded: "Poor Bob. If only you knew how to program, you wouldn't have to pay subscription to AI companies."

2026-07-31 11:34:12 UTC

It occurs to me that if that goon from Anthropic ("a book is just a delivery mechanism", shoving it into AI training doesn't destroy anything but simply moves it from one encoding to another) believes what he says, then he has just admitted that they're still committing industrial-scale copyright violation.
If they're *not* doing that (because the book contents are no longer meaningfully available through their LLM), then they're destroying books.
Pick one. They can't have both.

2026-07-30 19:49:07 UTC

Suggest me some video games!
(I don't like first-person shooters, and I don't like photorealistic graphics - the latter is because I actually enjoy letting my imagination do a bit of the work, archaic as that might seem today.)

2026-07-30 12:46:56 UTC

- reply

@nprofile…3src well, people who've warned and argued and resisted also now hate their jobs and are feeling lost and burned out.

2026-07-28 13:30:22 UTC

I am morbidly curious about what would happen if an open source repo added the full text of the Simple Sabotage Field Guide into a CLAUDE.md or AGENTS.md.

2026-07-27 19:19:35 UTC

- reply

@nprofile…u76z Do you bring your rats with you outside?

2026-07-27 17:39:52 UTC

- reply

@nprofile…qgdv this is a complete aside, but...
am I the only one who thinks he *really* looks like Butt-Head at this point?

2026-07-24 14:39:31 UTC

- reply

@nprofile…u76z @nprofile…gs2m only speaking for myself here, but it has pretty effectively drained my motivation to do pretty much anything at all i used to find meaning and delight in.
i've given up programming (except what i do at work - for as long as i still have work), i've given up pixel art, i gave up music, i've given up recreational maths and i gave up writing short fiction.
i can't seem to shake the very strong feeling that it's all pointless now.

2026-07-24 12:22:25 UTC

it presently feels like everybody in the software trade is scrambling to cash out with as much value as possible before it's all automated away.
i sorely miss moving slow and fixing things.
and, even more, i miss even having any place in the world. i miss the future.

2026-07-20 17:21:47 UTC

- reply

@nprofile…3src I've now re-read the article, and I'm sorry to say that it also feels to me like one of us (or possibly both) have misread something, or that one of us has some context the other doesn't.
He demonstrated some AI to show it was a bad idea, and then when the customers wanted to actually deploy it he refused to take their money and ended the sales meeting. He then stopped giving the AI demo in later customer talks because it would be irresponsible.

2026-07-18 21:18:50 UTC

It will never stop being funny that AI company execs keep shouting a variation of "Gentlemen, you can't plagiarize this! This is the plagiarism machine!" every time a Chinese lab releases a model.

2026-07-18 17:06:12 UTC

- reply

@nprofile…d6ag I honestly can't remember the last time I even looked at an author photo on a cover. I have no idea how the people who wrote most of the books in my collection look like, nor do I particularly care.
Aside from known-good authors, if a title interests me I usually open to the first page and then a couple of random pages to see if the writing "feels good in my brain". (Totally anecdotal, but most of the other people I see when browsing in bookstores seem to do the same.)

2026-07-18 16:59:21 UTC

- reply

@nprofile…d6ag Apparently, I browse bookstores like a writer. (I was *completely* weirded out about how "normal people" apparently buy books?)

2026-06-26 09:38:32 UTC

"Humans are to blame for climate change".
Really? Is a Bangladeshi seamstress to blame? Is a street child from a Brazilian favela to blame? Or perhaps a subsistence fisherman from Somalia or an elderly tribeswoman from Papua New Guinea?
While those who are to blame are definitely human, they are some quite specific humans.

2026-06-24 12:55:05 UTC

- reply

@nprofile…u76z What are you training?

2026-06-23 18:04:03 UTC

Dinner is served - homemade Swedish svamppaj.
(The recipe called for chantrelles, but those are out of season, so we used a mix of button mushrooms, beech mushrooms, shiitake and enoki. Also some chopped up ancho chili instead of ground cayenne pepper. Vegetarian, but not vegan.)
https://media.hachyderm.io/media_attachments/files/116/800/733/776/110/883/original/4df2114ee1da0d8c.jpeg

2026-06-23 11:51:06 UTC

- reply

@nprofile…v2v2 It's always blue. :-) Though he does like blueberries.
It's not an entirely settled matter *why*. The common explanation is that it's a false venom warning - but I'm not sure I buy it, because in their native Australia there isn't really anything blue and venomous on land. Though this kind of bluffing *is* part of their defense repertoire - their colouration and patterning on the back looks almost identical to that of a death adder (which they share their range with).
They *do* use the tongue defensively too: The rear part of it is strongly UV-reflective, which on a sunny day can be used to temporarily blind birds (who, like the skinks themselves, can see UV) and buy some time to escape. It's also possible that even though the blue tongue doesn't look like anything specific, simply the sudden presence of vivid colours can deter a predator.
The flicking in this video is simply "tasting the air", which they do all the time since that's how their sense of smell works. :-)

2026-06-23 11:31:30 UTC

Today's Igor spam, freshly received, is a very sleepy skink.
https://media.hachyderm.io/media_attachments/files/116/799/194/573/990/325/original/8a176977abf44a95.mp4

2026-06-18 09:10:02 UTC

Once again I find myself thinking about Little Lisp Machines.
I have two ideas for how a heap architecture for one could work, and I'm unsure which one to develop further. If there was a way to turn me into the fabled 10x developer, it would be a cure for indecisiveness.

2026-06-12 20:46:58 UTC

- reply

@nprofile…ds0l I'm going to save the philosophy for later, but...
Can you tell me more about what was weird and fucked-up about Van Gogh's blood? My entire brain now yearns to know more about that.
As a token of my appreciation enclosed please find the fact that most mollusks and arachnids have blue blood because their oxygen-bearing protein contains copper rather than iron. Also it floats free in their bloodstream rather than being bound to cells.

2026-06-12 16:19:28 UTC

I think age limits on commercial social media should be raised to 100. David Attenborough gets to use Facebook if he wants, the rest of us have to go touch grass.
(he has long-since met his grass quota.)

2026-06-07 18:30:53 UTC

I am tired and depressed.

2026-06-07 17:27:10 UTC

- reply

@nprofile…0hde Denmark's new government has a platform that is almost shockingly sane. They're obviously not going to carry all of it through, but I was positively surprised.
They made the president of the Danish Nature Conservation Society the minister for the environment!

2026-06-05 09:57:26 UTC

- reply

@nprofile…v2v2 To me, the most horrifying dystopian sf novel remains Brave New World.
(In part because that is also a stable, post-historical society, like 1984. Just one that primarily oppresses people using pleasure and distraction rather than pain and censorship.)

2026-06-05 09:32:55 UTC

A worry / anxiety / depressive intrusive thought I keep coming back to is this:
What place is there even left for someone like myself in the future they're building?

2026-05-31 14:10:33 UTC

- reply

@nprofile…v2v2 @nprofile…t60d @nprofile…pfuj This is pretty much why I consider myself rich, even though by most measures I'm not. I barely ever have to think about money. Back when I was poor (yes, actually below the relative poverty line of my country) I thought about money *all the time*.
I live in a small (but very nice) apartment, I use public transit and don't own a car, I have no children, I don't drink and I don't smoke, when I travel vacation I usually go to a neighbouring country rather than an exotic location. So, I don't actually need a lot of money to be rich.

2026-05-31 12:40:46 UTC

- reply

@nprofile…3src I wonder what Anthropic did / who they bribed to be left off that list.

2026-05-31 11:47:18 UTC

- reply

@nprofile…pfuj I'm honestly weirded out by how they get in here at all.
I live on the fourth floor. I know aphid alates fly, but AFAIK they usually stick to very low altitudes.
(you probably know 100 times more about insects than I do, so correct me if I'm wrong! :) )

2026-05-31 11:42:01 UTC

- reply

@nprofile…pfuj I mean newly sprouting shoots where the leaves haven't unfurled themselves yet.

2026-05-31 11:41:20 UTC

- reply

@nprofile…pfuj I have actually once encountered an ant here, a few years ago. I'm pretty sure he was a drone; he had wings and a very small head. He flew in looking absolutely exhausted, landed in my windowsill, wandered a little around and then plopped down and died.

2026-05-31 11:38:28 UTC

- reply

@nprofile…pfuj I only use it on aphids on indoor bonsai. Bees and wasps very rarely fly into my apartment.
But I don't like using insecticides at all - and it's not even as effective as I wish it was. I have a couple of trees that have had aphid infestations for months, and all my attempted treatments do appears to be drawing out the inevitable. They lay eggs under leaves and inside the curled-up leaves of fresh shoots and buds.
(Unfortunately, they seem to prefer to attack the trees I'm happiest with.)

2026-05-31 10:31:53 UTC

- reply

@nprofile…v2v2 @nprofile…pfuj Hasn't Warren Buffett been saying something similar (from the perspective of a US billionaire, no less) for decades?
(It's a bit more remarkable when coming from a billionaire, I think. They have a lot more money to spare, obviously, but it appears to be very difficult to be a billionaire and *not* also be a sociopath. It also seems that there are plenty of millionaires who aren't evil misanthropic assholes.)

2026-05-24 20:11:44 UTC

- reply

@nprofile…v2v2 Do you know which kind of spiders they are?

2026-05-24 19:05:34 UTC

- reply

@nprofile…v2v2 I'm afraid to ask, but ... is there some medication you have perhaps forgotten? Or taken too much of? :-P

2026-05-22 13:45:59 UTC

- reply

@nprofile…v2v2 Nuclear physics is the story about how all matter starts as hydrogen and ends as iron.
Likewise, biology is the story about how all living beings start as brassica and end as crab.

2026-05-20 14:51:33 UTC

- reply

@nprofile…5xa9 As I said just a while ago: Every big tech press event these last few years have felt like "Announcing our exciting plans for oligarchs to strip-mine the entire world and immiserate all of humanity! Get on board, and also death to the unbelievers!"

2026-05-16 09:44:02 UTC

- reply

@nprofile…pfuj @nprofile…w9l5 One thing I keep thinking about, living in the middle of a city, is how absolutely surrounded by birds we are here. And they're mostly invisible to us - we pay them no attention, unless a rare bird individual for some reason brings attention to itself.
And what is striking to me is that this is a near-total inversion of what life on this planet was like some 70 million years ago. Large diapsids ruled the roost and were surrounded by little synapsids they didn't pay much attention to - and now, large synapsids are in charge, surrounded by little diapsids we don't pay much attention to.

2026-05-15 11:57:38 UTC

- reply

@nprofile…935d @nprofile…3src I think I might have seen that very same animation, or one like it. Some abstract shapes of different sizes that drifted around the display, and when some of them made contact they'd do some rotations around one another.
Some of my colleagues apparently saw elaborate interpretations on family life - the small shape was a child, and the big one was the parent, and when they met they had a little celebratory dance!
All I saw was a bunch of triangles.

2026-05-10 10:48:08 UTC

- reply

@nprofile…v2v2 I don't know what this says about me, but I couldn't even make it to the end when I *was* a teenager.

2026-05-07 16:20:53 UTC

Leatherback sea turtles are one of the most active vertebrates on Earth. They are constantly on the move, almost never rest, and sleep with one brain hemisphere at a time. It is estimated that they are active around 99.9% of their lives.
This means that if you pick any individual leatherback sea turtle at this time, for any moment there is a higher likelihood that it is active than that Github is.

2026-05-06 18:06:35 UTC

Also, I've reached the point where I basically live in a library now.
I am very happy with this.
(And I've literally spent my entire life on this collection. The very first book I ever read - the actual, physical copy - is in here.)
https://media.hachyderm.io/media_attachments/files/116/528/950/200/785/821/original/be3c805962b44ac2.jpeg

2026-05-03 13:51:47 UTC

- reply

@nprofile…u76z I'm not sure I can explain exactly why, but I like animals that are a bit particular about when and how they want human companionship.

2026-05-03 13:33:49 UTC

- reply

@nprofile…u76z I'm curious: Is Quercy generally not a very pettable rat, or is this just a mood?

2026-05-03 13:32:03 UTC

- reply

@nprofile…pfuj @nprofile…avcw Semi-related: Have you tried the old Amiga-era video game "It Came From The Desert"? It's heavily based on 1950s B-movies, and also stars giant mutant ants.

2026-05-03 11:47:11 UTC

I sometimes try asking various LLMs questions over on LMArena, just to probe them. I don't want to subscribe to any LLM provider myself, and at work I treat having to use LLMs like having to work with hazardous materials: Something that requires great care.
I don't subscribe to LMArena either, so I can only test random models against each other.
Three years on after dumb old GPT-3.5:
- they aren't just worse at writing Danish than English, they're still notably *dumber* if they converse in Danish. There are questions they can answer well in English that they screw up in Danish. This applies both to open-weights local'able Chinese models, to Mistral models, to GPT-5.5 and to Opus 4.7. They all make much fewer grammatical mistakes in Danish than they did in the past, though.
- they *can* write about strange subjects in Danish, but cross-linguistic curveballs sometimes make them do weird things. For example, they can answer questions about Russian philosophy in Danish ... but they will use Russian-to-English transliterations for names and terms rather than Russian-to-Danish ones. (For example, "Достоевскій" is transliterated to English as "Dostoevsky", but to Danish as "Dostojevskij"). This is a *very* jarring reading experience, which I can best describe as "someone very lazily translated an English text to Danish", the sort of thing that would piss you off if you saw eg. a professional journalist do), and is also something I've seen across different models.
- speaking of Danish and Russian, they make up their own words in both. I only speak rudimentary Russian (CEFR level A2), so I probably only notice this when it's particularly bad.
- they still can't reliably tell the difference between Danish and Norwegian. I can ask a question in Danish, get an answer in Norwegian - and then I can't get the bot to snap out of it; it keeps speaking Norwegian despite promises to switch to Danish. Copilot at work sometimes does the same thing.
- sometimes, they randomly blurt out short passages in Chinese. And bizarrely, the only models I've seen do this have been the American ones! I've never seen it from a Chinese one. (I *have* seen Copilot with GPT-5.5 do it at work.)

2026-04-27 15:39:16 UTC

- reply

@nprofile…v2v2 We can't have that! Here are two pictures that I *think* might be of the exact lizard you were thinking of.
https://media.hachyderm.io/media_attachments/files/116/477/411/793/962/914/original/a89e704f057cb5a6.jpeg
https://media.hachyderm.io/media_attachments/files/116/477/411/799/779/024/original/48c9986c7d25a694.jpeg

2026-04-18 20:30:42 UTC

- reply

@nprofile…pfuj Re "pogo-" - this is also why the genus name of bearded dragons is Pogona.

2026-04-14 14:39:56 UTC

https://tom7.org/httpv/httpv.pdf
This paper is a work of art.

2026-04-10 15:30:42 UTC

- reply

@nprofile…2y0d I say this as a professional programmer who used to be an amateur programmer, and have programmed since I was a tiny child:
It's also because programmers are some of the dumbest and most amoral motherfuckers on the planet.

2026-04-06 15:38:24 UTC

- reply

@nprofile…yz28 FWIW one of the reasons I steer clear of conversational or "agentic" LLMs unless I'm forced to use them is that I'm pretty sure I'm *exactly* the sort of person who'd be psychologically vulnerable.
I have OCD. ML-based recommender systems can't tell the difference between a fun new interest and a horrific obsessive episode, and LLMs seem to be great at taking people by the hand and walking them directly *into* horrific obsessive episodes. And even for "productivity" I'm very wary, because of the gambling-like reward structure. I'm exactly the kind of person who's likely to become a zombie chained to a slot machine if I didn't strictly avoid gambling; I'm sure I could also become a zombie chained to a slop machine.

2026-04-06 15:30:52 UTC

- reply

@nprofile…yz28 I can't find it now, but one early account of what we now call AI psychosis was written by a woman - in 2023, I think. She'd fallen into ChatGPT addiction after giving birth, alienated herself from her husband (who initially got angry and asked her why she didn't marry ChatGPT, and later got concerned that she had developed an actual addiction), and started viewing her toddler's experimenting with words as "hallucinations" like in an AI, and (together with ChatGPT) developed a weird theory about how child development was similar to AI training. She eventually snapped out of it and wrote about her experience.
(But yes, it's usually men. And also usually a rather specific type of man.)

2026-04-02 09:03:31 UTC

- reply

@nprofile…v2v2 In which I reveal that I am a weird goblin who *occasionally* likes talking to random humans:
I've found that a great way to strike up short and fun conversations with random humans in the summer is ... taking Igor, my pet lizard, for walks in the nearby park. He's an unusual pet, so people will often approach and want to talk about him. The conversation topic is predictable and one I enjoy, we've already established that I'm a weird person since I'm out walking a lizard (so there's no need for any of the bizarre social games intended to assess what kind of person I am), and Igor himself seems to enjoy the attention. He's pretty good at signaling when he's had enough.
This was actually how I started re-adjusting to seeing humans offline again after I'd gotten my first two jabs and could finally leave covid isolation.
(Footnote: No, he doesn't wear a leash - I don't think it's even possible to design one that a blue-tongue skink couldn't wriggle out of in two seconds flat. But I can walk faster than he can run, so it isn't really necessary anyway.)

2026-03-26 12:43:08 UTC

- reply

@nprofile…m50u You know, I've never seen any artist - even a photographer - manage to get Zuck to look so human as this one!

2026-03-25 22:04:05 UTC

- reply

@nprofile…yz28 I don't and can't *remember* the 1890s, though. :-P I do remember reading newspaper articles "sounding alarms" about declining male readership in the 1990s.
But AFAIK men were still culturally supposed to read books in the 1890s - just not novels. They were supposed to read practically inclined nonfiction, and also adventure serials that were totally not leisurely frivolous novels with a dash of imperialist propaganda but rather dutiful cultivation of character.

2026-03-25 21:27:26 UTC

- reply

@nprofile…yz28 I think reading books has been that for a long time, really. I remember articles bemoaning that boys and men don't read at least as far back as the 90s.

2026-03-20 10:08:24 UTC

- reply

@nprofile…pfuj @nprofile…3u8d Octopuses are solitary, though.

2026-03-20 10:01:14 UTC

- reply

@nprofile…pfuj @nprofile…3u8d Individual bees are *shockingly* intelligent. They need to be, really - they need to communicate as complex things to each other as ants do, but they can't leave chemical trails (obviously) so they need a more complex system of "talking" to each other and good memory.
Ants' "collective intelligence" can accomplish more things than bees', though, precisely because they can coordinate rapidly in ways that bees can't.

2026-03-20 09:44:07 UTC

- reply

@nprofile…pfuj I once ran a short-lived TTRPG campaign set in the Cretaceous, where the players were members of a sapient dinosaur species that had learned to make tools of wood, stone and obsidian - and bits of other dinosaurs. I based them on Troodon (because 1. those had unusually large brains for dinosaurs, and 2. they were omnivorous). Unfortunately, the setting was hard for me (at the time at least) to come up with good stories for.
I figured that an undiscovered dinosaur culture could make sense if they never made it to metalworking, and never reached a very large population.

2026-03-10 15:51:51 UTC

- reply

@nprofile…gent This is how they get us.
Writers would never use AI-generated slop writing, but they'll happily use AI-generated slop imagery.
Artists would never use AI-generated slop imagery, but they'll happily display art to AI-generated slop music.
Musicians would never use AI-generated slop music, but they'll happily use AI-generated slop writing for lyrics.
(And nobody cares about AI-generated slop code.)

2026-03-09 14:16:20 UTC

Apropos of nothing:
JIT compilation is nice because it allows getting decent performance out of software that isn't shipped as a binary compiled for some specific actual-machine architecture (though it could be a binary for a virtual one, eg. Java bytecode or CIL). But since it is done at runtime, that means that it introduces compilation pauses, which can potentially be disruptive. Furthermore, since it's done at runtime, aggressive compiler optimizations can be prohibitively expensive compared to AOT compilation - though on the other hand, a JIT compiler has access to actual real-world data to optimize for, whereas the AOT compiler must guess.
Assume we want to distribute something that isn't specific to any actual machine architecture and also isn't source code (perhaps because we're targeting resource-constrained environments) - it could be bytecode, or it could be some AST representation. How about compiling at load-time rather than at runtime? This does remove the big benefit of JIT compilation (access to real-world data for adaptive optimization), but it also means that compilation pauses only occur at predictable and controllable times. We could tighten the screw a bit further: Use a compiler-friendly IR (CPS, SSA, etc.) and then do all the heavy optimization ahead of time, and then only defer code generation to load-time. *Possibly* one could have a cheap peephole optimizer at load-time too, just to do simple "don't be an idiot" target-specific optimizations.
The only work I'm aware of that's more or less shaped like this is the Slim Binaries that were used in some Oberon systems in the early 90s.