No, the ICE agent isn't "kneeling on his neck". Yes, yes, ICE is the modern gestapo, a bunch of thugs operating without accountability. What I'm arguing is that you cannot "see" what's happening here. You think you clearly see the ICE agent kneeling on his neck like with George Floyd, but you don't. What you see here is the approved, trained technique of kneeling OVER the neck. The difference is that the weight is still mostly on the other leg, and on the heels. The knee is extended over the neck to prevent the target from being able to lift their head, but the weight is not on the neck, it's on the heels and other leg. You can watch the agent leading up to this point, where the ICE agent is carefully placing his foot in the right position, so that the weight stays on his back heels and other leg, then extends his knee over the target's neck. This technique is extensively trained for and approved, and is included in most training manuals. That's why you know Chauvin's claim of using his training is false. His weight was on Floyds neck, not on his other leg or heels. He was definitely NOT trained to do this. The two procedures are different for exact the reasons shown, that one can injure the person, and the other simply immobilizes them without injury. The point of this tweet isn't to defend or attack Chuavin or ICE, but to point out that you and I can see the same videos but see completely different things, and that some education is important.

This is Kerckhoffs Principle from 1880s. We can secure computers against a hyper-intelligent AI, preventing most of the doomerism scenarios. But that's only if you follow such principles. Currently, those preaching Doom want to violate all the basic principles of cybersecurity, making it more likely instead of less likely.

Now tell me what people mean when they say they've never heard BTS.

"Patriotism is supporting your country all the time, and your government when it deserves it" - Mark Twain

Still the best tweet

This "hacking attempt" is recycling the exact same website vulnerability identification ancient 25 year old scanners did. I would know, I wrote those same exact scans for Retina, the official network scanner used by the DoD. This is embarrassingly over exaggerated claims

This is the "regulatory capture" phrase that is widely passed around. If a teen commits suicide, parents will sue ChatGPT/Claude, because there is no firm definition of what it means for an AI to be "safe". A slick lawyer will usually convince a jury that corporations are evil and need to pay for tragedy, regardless of the facts. Juries are emotional. Therefore, AI leaders want government to "impose" AI safety regulations on them. While such laws punish them if they don't meet the rules, it exonerates them when they can prove they met the safety regulations, forestalling lawsuits. The leaders of every large industry do this. This is not unique to AI. They'll accept even bad regulations as long as it provides regulatory certainty and protection from ad hoc lawsuits. This also suppresses a patchwork of state laws. The leaders know that, as the biggest companies in the industry, such laws won't hurt them much, but will hurt smaller competitors. Every new regulation increases costs that large companies can easily absorb but small challengers cannot. They want this even if anti-AI activists have too much influence and create non-optimal regulations. They benefit even if they don't achieve "regulatory capture". But, since they are the loudest voices in the room, they'll have the most influence, and likely get favorable regulations. The point is that "regulatory capture" isn't the goal, but is usually the effect. The reason the leading AI CEOs want regulation is because the leaders of all big industries want regulations.

He goes on to include "advanced calculus" and "circuit design" with such "basic math". I think he's trying to have a discussion as a techie when as a CEO he's supposed to stick to media-trained messaging. I think he's just claiming that school curriculum will change with AI, with changing views about what's important. I disagree with his thinking, though. I don't think you can do high-level "systems" thinking without low-level "basics". I suspect curriculum will focus even more on the basics. That includes calculus -- summing an infinite number of infinitely small numbers is critical for any high-level understanding.

Hi. Cybersecurity expert here. I "meddle" in unusual ways with websites all the time. If you are a website administrator, you see my name in your logs. I also have meddled with your home Internet router. I'm famous for this. Some want me criminally liable for it. I worked with the DoJ so that they published recommendations to prosecutors saying such meddling is legal and not to go after "hackers" like me for it. You want such meddling by the good guys because it ends up improving the security of computers. When you say you want "criminal liability" instead of merely "civil liabilities" for damages, you are saying you want this improvement to stop. Stopping such cybersecurity work will end up making hacking doomer scenarios more likely. If you want to stop the coming Doom, you want maximize such "meddling", as much AI hacking as you can get that doesn't actually do damage. Again, I've meddled with your home router. I've meddled with EVERY device on the public Internet. This is not only legal, it should be legal. And you want AIs do be doing a lot of it.

No, "anthropomorphizing" is a technical problem, about how AI works, not something like woke language policing. We say it's wrong because it's "technically" wrong not "morally" wrong. If you fail to set the brakes when parking on a hill, and your car then rolls down the hill, you don't claim that the car "decided to drive away on its own". Such anthropomorphizing distorts what really happened -- that you failed to set the brakes. The same applies to AI. When you "anthropomorphize" the AI you fail at understanding what actually happened.

AGI or "general" intelligence is still far away. ASI or "super" intelligence is already here. Coding is the best example. You would never write code without an AI anymore (or else you are doing it wrong). They still have faults, but in other ways, they write superior code. I'm a long time 10x expert coder and this depresses me. We aren't going to see a leap to a "general" intelligence that matches humans in every way. We are going to see a steady erosion as AI exceeds our abilities one by one. I suspect a major advance in driving is around the corner, awaiting Moore's Law. Today's self-driving computers (like in your Tesla) has the same power as your laptop computer. Within a few years, they'll have the power that $100,000 AI servers have. When that happens, I suspect that Level 5 autonomy will be possible. At the same time, the cost of additional sensors like LIDAR will have dropped to nothing.

This political cartoon is 22 years old

I just say "you people".

Here to help!

If you mean @mtaibbi, @bariweiss, or @billmaher,, then name them and fight it out with them. I will bring cheddar popcorn. But don’t use “free speech warriors” and the “anti-cancel culture brigade” as broad terms of derision and then act surprised when actual free speech advocates push back. Some of the best people I have ever met in my life are people at FIRE, or people I’ve met through FIRE, from all across the political spectrum. They have spent years defending people they disagree with, often intensely, because they actually believe in the principle. They fought censorship and cancel culture when it came from the left, and they fight it now when it comes from the right. And they absolutely get caught in the blowback from this kind of dismissiveness . There is a very common habit of trying to score points against your political enemies by implying that “free speech people” are all hypocrites—that nobody really believes in this stuff, everybody is secretly partisan, and consistency is a fraud. I’m tired of that cynicism. I’ve spent most of my adult life around people who prove it wrong every day. So don’t casually smear an entire community of people doing difficult and genuinely admirable work. If you mean, other people name them, and they will get a chance to defend themselves. Consistent defenders of free speech exist. A lot of them are my friends and colleagues. And they deserve better than the constant eye-roll assumption that everyone is as cynical as their critics.

Why is voter ID controversial? 1. Because it's not just about showing ID. 2. It blocks more legitimate voters than illegitimate voters. 3. Because American isn't a "show your papers" country. Trump's SAVE Act is radically more than showing voter ID. It includes provisions for federal interference in elections and building a database tracking all Americans. It's clear corruption. A lot of people who want voter ID nonetheless vigorously oppose the SAVE Act. The lack of voter ID isn't a problem. This has been thoroughly studied. Trump's data clearly lies. He claims hundreds of thousands of illegitimate votes that when investigated turn out to be merely hundreds of "potential" illegitimate votes, that after further investigation end up being zero (0) illegitimate votes. The number of illegal votes that voter ID would stop amount to fewer than 10 each election even in big states like California. On the flip side, it would effectively disenfranchise thousands who cannot afford the expense. That's not a lot of people, and we can address these cases. So this isn't a big hurdle to overcome. My point is simply that the small problems voter ID causes are still bigger than the even tinier problems that it solves. Finally, American isn't a police state. When the police stop you, they cannot demand that you show them your papers. This is very different from most countries where you have to carry your papers all the time. In American fiction, police forcing people to "show us your papers" signals that the government is some sort of dictatorship.

The more we let today’s toddler AI run amok on the internet, fixing any bugs it finds, the harder it will be for tomorrow’s even more advanced AI to cause much harm. This continues ad infinitum.

As soon as Trump announced his candidacy for president in June 2015, the quotes of all politicians became more negative. He is single-handedly responsible for most of the increased negativity in our political culture. richardhanania.com/p/the-tru…

From 16,000 to 185 to Zero: Trump’s Claims of Illegal Voting in Nevada Fizzle. "The state sent a strongly worded rebuke to the federal government: Every single person on the list was a lawful citizen." nytimes.com/2026/09/24/us/po…

There is no evidence of a "hack", and a lot of good reasons to believe it's not a "hack". It is common for websites to accidentally expose non-public data to the public. Search engines often stumble across this information. An AI acting like a search engine will stumble across even more. When this happens, website owners complain that "they" (the putative "hackers") should've known the data wasn't supposed to be public and shouldn't have accessed it. They also hide details, so nobody can see that it's the website's fault and not hackers. As long as these Australian officials hide the details, we can reasonably assume they are hiding evidence of their own incompetence.