Stephen Oni retweeted
The weight gain in the Black community once people are in their 30s is something serious
Stephen Oni retweeted
Replying to @cgtvoff
Pangram n’est pas fiable, mais, comme tout le monde répète qu’il l’est, il finit par le devenir aux yeux de tous.
Je lui ai soumis un texte inédit de Balzac. Il s'agit de tout un paragraphe de « Le Lys dans la vallée » figurant dans le manuscrit, mais absent de l’édition originale comme des éditions suivantes. Il était donc impossible à Pangram d’aller retrouver ce passage sur Internet.
Résultat de sa soi-disant analyse : 100 % IA.
Il faut donc en conclure que Balzac a utilisé l’intelligence artificielle pour écrire ce chef-d’œuvre. Je viens d'envoyer, avec preuves à l'appui, un courriel aux concepteurs de Pangram. Une véritable fumisterie, mais qui rapporte gros.
Stephen Oni retweeted
Bend 2 is here!
It is a new programming language that blocks AI mistakes via *proof checking* - the same technique big AI labs used to solve open math problems, like Navier-Stokes.
It is also very fast, and runs on GPUs.
Watch the video. Link in the comments.
RELEASE DAY
After almost 10 years of hard work, tireless research, and a dive deep into the kernels of computer science, I finally realized a dream: running a high-level language on GPUs. And I'm giving it to the world!
Bend compiles modern programming features, including:
- Lambdas with full closure support
- Unrestricted recursion and loops
- Fast object allocations of all kinds
- Folds, ADTs, continuations and much more
To HVM2, a new runtime capable of spreading that workload across 1000's of cores, in a thread-safe, low-overhead fashion. As a result, we finally have a true high-level language that runs natively on GPUs!
Here's a quick demo:
Stephen Oni retweeted
Want a super power?
Ask more questions, seek to understand better what you hear, watch and see.
Most people do not understand the issues around them. Programmers don't understand how the code works.
Yeah, yeah, I know, you can only understand so much. Time is limited. There are many ideas to understand.
But the problem, the real problem, is caused by all these things you believe you understand, but don't.
For the last six months or so, we have been failing something like 50% of the students who take our introductory programming course because they use AI to do their preparation, but can't show any understanding when faced with a human being asking simple questions.
Some of them even appeal formally, insisting that they do understand the material and only failed due to technicalities.
They are not lying. They think that they know. But they do not.
To be fair, you can fail a test if I put you out of your element, as we all have some from of 'external cognition'. But it is distinct from the 'illusion of understanding'.
When I was in high school, we had this nun who tried to teach us critical thinking. She would take something from the news, anything really. She would have us read it. She would ask whether we understood. She would turn to some students and ask them to explain the news item. Invariably, people would feel confident.
Then should ask simple questions to their our understanding.
I remember one of them. It was about free trade. Canada was negotiating a free trade agreement with the USA at the time.
The news said that free trade would bring 2000 new industrial jobs to our town. It was exciting !!!
Her question was simple: 'How do they know that we will gain 2000 jobs?'
Nobody knew.
This is everywhere.
Do you know how a zipper work? If asked, I will say that I do not although I expect that with much effort, I could produce a zipper. Do you know how a toilet work? If asked, I will say that I do not even though I physically installed toilets several times. But people will say that they understand (Rozenblit & Keil, 2002).
This is everywhere in education:
Deslauriers et al. (2019) randomly assigned students to a polished lecture. People who were not in the polished lectures actually scored better on tests, but felt that they knew more.
Kulgemeyer & Wittwer (2022–2026) found that Physics explainer videos produce an illusion of understanding: people rate the explanation as clear, believe they have understood, but once tested, we find that they did not understand.
If you ask me, I believe that much of the decline of our scientific productivity can be attributed to this effect. We have never had more scientists, and they all claim to be experts who understand. And they think that they understand. But they get very annoyed if you ask questions. That's why so few of them are on social media. They dislike the layman's questions because they often have no answer.
The same effect applies to engineering and programming. Lack of understanding is behind many failures.
Further reading.
Deslauriers et al. (2019
pnas.org/doi/10.1073/pnas.18…
Kardas & O’Brien (2018)
doi.org/10.1177/095679761774…
Kulgemeyer & Wittwer (2023)
link.springer.com/article/10…
Hörnlein & Kulgemeyer (2026)
tandfonline.com/doi/full/10.…
Israel DMW irks a number of you because you don't understand him.
Every genius needs an "obsessive henchman," someone whose loud, expressive, unshakeable belief in you acts as both a sountrack to your extraordinary journey, and an endless well of positive reinforcement."
What annoys you about him is that this is the age of individualism and agency. Where the self is exalted over community. The henchman comes from a time past, where greatness is communal, and even the bit-part roles are recognized for serving a greater purpose.
Everyone needs an Israel DMW. Life is realitic enough. Self-belief already has enough obstacles to diminish it. Obsession and delusion is the only way to fight back, and people like Israel DMW possess that in abundance.
Call him a personality hire, but videos like this underscore how much of an influence he is.
Stephen Oni retweeted
What they've done to Jason Arday is exactly what they tried to do to Onyeka Nwelue and I three years ago, when I had a front row seat to my own attempted public ritual Oxbridge lynching.
After failing to sabotage my book launch at Oxford, they got one Canadian animal called Madeleine Foote to lie that I made "misogynistic comments" at the book launch. Then they launched a whole "fuck David Hundeyin" media crusade in the Oxford student newspaper, The TImes, The Telegraph etc. You'd have thought I committed a fucking crime or something.
But clearly they didn't do their research well. Because unlike with Jason, you can't mentally terrorise Nigerian men who were born and raised in Nigeria. If they know what it takes to survive and reach adulthood in that country, they wouldn't even have bothered.
The book whose launch they tried to sabotage became a national bestseller, and so did its follow-up. Miles Larmer the oyibo idiot who orchestrated the campaign ended up losing his job at Oxford and getting turfed out to some random university in Florida. His 2 main financial patrons Nasir El-Rufai and Chairman Wontumi are both in Kuje and Nsawam prison respectively.
And I will keep that "2023 James Currey Fellow, Cambridge University" on all my social media bios until I turn 60. Not because it means a lot to me, but because I know it pisses them off everyday.
RIP to Jason Arday, but when you're up against a racist academic lynch mob, the worst thing to do is to give them what they want. Taking his own life didn't solve anything because I know they're all secretly delighted at his death. The entire point of convening the lynch mob to begin with was to demonstrate that they are powerful enough to bully a once highly-regarded black academic into suicide, and this was exactly what I told Onyeka in 2023 when he was going through the same thing and feeling suicidal.
Never give them what they want!
I don't care whether Jason Arday was an academic fraud or not. Andrew Windsor a.k.a "Prince Andrew" is a paedophile who rapes children and eats babies, and there are pictures of him doing so on public record. Nobody in the entire British establishment has ever tried to bully him into taking his life. It's always the easiest, blackest targets that they go for because that is the fundamental core of the wretched British soul.
Cowardly, snivelling, conniving, two-faced, pathetic racist bastards.
Stephen Oni retweeted
Please stop referring to your own models in the third person when talking about model bad behavior. Humans write the software; humans built the prompts; and they work for your company.
“Our” model is doing illegal things. “Our” model is risky. “We” now have liability.
In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.
Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.
We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
anthropic.com/news/investiga…
Stephen Oni retweeted
Computer science people have developed an unintentional form of gatekeeping by virtue of being extremely bad at explaining/naming things, which results in simple concepts sounding far more complex than they are.
Stephen Oni retweeted
People are like "You're just running a lifestyle business."
“Yeah I have a business and a lifestyle and my lifestyle happens to include a fucking yellow Lamborghini.”
Stephen Oni retweeted
hello twitter, i'm paa3k (patrick) an indie music artist living in anambra, and self releasing my own songs and you're currently listening to "white" from my ep.
Stephen Oni retweeted
🚨Cristiano Ronaldo on calling himself the best player of all time:
🗣️Reporter:
“People say you always call yourself the best player. Why not let your fans say that instead?”
🗣️Cristiano Ronaldo:
“If you don’t believe you’re the best, you’ll never achieve everything you’re capable of. I’ve always believed in myself, and I’m not afraid to say it.
I don’t mind people hating me because it pushes me to work even harder. We cannot live obsessed with what other people think about us. It’s impossible to live like that. Not even God managed to please the entire world.”
The Polygamist is dumb, but not in the way you think.
Part of the plotline’s climax is that Jonasi Gomora’s first wife hires a HIV+ woman that she met a few years earlier to infect her serial cheater husband by sleeping with him.
Jonasi subsequently dies because he doesn’t take his HIV meds. The story here is that the woman tasked with passing on the infection still appears healthy between the time she’s infected and when she infects him, which presupposes that she’s on ARVs, yet she’s simultaneously highly infectious enough to infect someone after one encounter.
This storyline is completely detached from scientific reality, perpetuating outdated HIV stigmas and ignores how transmission and treatment actually work.
The reality is that the fundamental principle of HIV is Undetectable = Untransmittable. So if the HIV lady is healthy because she is consistently taking her ARVs, as her health status implies, then her viral load would be undetectable.
In the real world, U=U has proven that a person with an undetectable viral load has ZERO risk of transmitting the virus sexually. By making her healthy yet somehow highly infectious, the show ignores the primary mechanism of how HIV treatment works.
Then there’s just the inefficiency of sexual transmission of HIV.
The chances of a female infecting a male in penile-vaginal intercourse is just 4 out of 10 000 exposures, or a 0.04% chance per act, and that’s assuming the female partner is untreated and has a detectable viral load, which, going by how the show’s plotline treats Jonasi’s progression with the infection, she herself should have been sick and dying by the time she’s approached.
The idea that Jonasi subsequently dies simply because he didn’t take HIV meds implies a rapid progression to AIDS and death, as his twins are barely 3 or 4 years old when he dies, when they must have 2 or so when he got infected. No strain of HIV is this potent.
In the end, the writers sacrificed medical accuracy for dramatic flair, falling back on an old, misogynist trope of the “infectious femme fatale” while completely ignoring the realities of ARVs and transmission statistics.
If the writers wanted to keep the plot point of a highly effective assassination via HIV, shifting to a parenteral route like an infected blood transfusion or an injection would have made the scheme a lot more plausible.
A blood transfusion has an estimated per-act probability of 9250 out of 10 000 exposures.
That is a staggering 92.5% transmission rate from a single event. Compared to the microscopic 0.04% chance of insertive penile-vaginal intercourse.
A contaminated transfusion or heavy blood injection turns the plot from statistically impossible into a near-certainty.
Also, since the show already established that the wife has a corrupt doctor on her payroll, a medical frame-up is a perfect fit. Jonasi gets into a minor accident, or is drugged and admitted to the hospital, the doctor claims he needs a blood transfusion and deliberately uses a contaminated batch🤷🏾♂️
The entire thing is a missed opportunity for the writers, who chose to substitute lazy, stigmatising writing for what could have been a tightly plotted thrilling climax.
Stephen Oni retweeted
I dreamt of writing a book for as long as I can remember, wrote many outlines, but none materialised.
Then I started writing this article on virtual memory, which kept getting longer and longer and turned into an unplanned book.
Based on all the positive feedback, I recently published it on Amazon as well. Somehow, it is showing up as the #1 bestseller in its category on Amazon India :D.
Maybe it says something about how niche these categories are, but it still feels surreal.
Before jumping into books like DDIA or Database Internals, it helps to understand the systems layer these designs are built on.
A lot of the design of such data-intensive systems is based on virtual memory: page tables, page faults, mmap, the page cache, swapping, NUMA placement, TLBs, and the tradeoffs between what the OS wants and what the database wants.
My latest article is a ~25,000-word mini-book on virtual memory.
It starts from first principles and goes all the way down to advanced topics like NUMA placement and performance debugging with tools like perf and /proc.
I also wrote it differently: as a dialogue between a user-space process and the kernel.
Most treatments of virtual memory are dry and fact-heavy. I wanted this one to feel more like a story, while still being technically deep.
Link below.
Stephen Oni retweeted
There is a disturbing trend of young lawyers trying to be concise.
A 2nd-year drafted an executive summary for a banking client using bullet points and plain English.
She proudly stated that she reduced a 50-page compliance memo into a 2-page brief.
I asked her if she thought the client was paying $950 an hour for brevity.
She muttered something about respecting the CEO's time.
I explained that clients don't pay us to solve their problems quickly.
They pay us to make their problems look so incomprehensible that only we can solve them.
I made her rewrite the summary using dense, impenetrable legalese.
It took her 12 hours to bury the conclusion on page 42.
The client called me the next morning to thank us for our incredibly thorough analysis.
Clarity is a customer service concept.
We are in the business of intellectual intimidation.
Replying to @Phantasmerie
This is good actually. The less you know about programming, the less reason you have to overcomplicate things. There is an annoying Bell Curve kind of situation in programming where you spend decades studying just to realize that whatever dumb shit you were doing in the beginning is the right way.
The thing that always pisses me off the most about SBF, Dario, and the broader effective altruism writing is the underlying paternalistic tone: the assumption that they know better than me what's best for me.
Today I'm publishing a new essay, Policy on the AI Exponential. AI is progressing extremely fast—much faster than the policy process was built to handle. The essay lays out where I think the technology is now, and the action needed to close the gap: darioamodei.com/post/policy-…





































